# Motionexa Labs > Motionexa Labs is an independent generative engine optimization (GEO) > studio that audits and improves how often early-stage SaaS and > cybersecurity companies are cited by ChatGPT, Perplexity, Gemini, > Claude and Google AI Overviews. Founded 2026. Desk in Mysuru, India; > clients worldwide. Key facts: flat-rate AI visibility audit, $1,200 USD (40+ buyer prompts run across 5 engines; deliverables: prompt log with transcripts, citation share-of-voice scorecard, source map, prioritized fix worksheet, 45-minute walkthrough; 7 business-day turnaround). Citation Sprint $1,800/month (3-month execution engagement). Standing Desk monitoring $1,200/month. White-label GEO audit fulfilment for SEO and marketing agencies from $600 per audit. Founding-client guarantee: if an audit fails to surface at least 10 buyer prompts where competitors are cited and the client is not, it is free. Contact: hello@motionexalabs.com. The published sample audit is a clearly labeled demonstration specimen, not client data. ## Core pages - [Method](https://motionexalabs.com/method/): the 7-step audit process, fully documented - [Sample audit](https://motionexalabs.com/sample-audit/): specimen of the exact deliverable - [Pricing](https://motionexalabs.com/pricing/): all rates public, including white-label terms for agencies - [Request an audit](https://motionexalabs.com/request-audit/): intake form, replied to within one business day - [About](https://motionexalabs.com/about/): who runs the lab and the standing orders it works under - [FAQ](https://motionexalabs.com/faq/): 14 plain answers, including what cannot be guaranteed ## Research (Field Notes) - [What is generative engine optimization (GEO)?](https://motionexalabs.com/field-notes/what-is-generative-engine-optimization/): definition, 2026 buyer data, how engines choose brands - [How ChatGPT chooses which brands to cite](https://motionexalabs.com/field-notes/how-chatgpt-chooses-which-brands-to-cite/): training data vs retrieval, ranked signals - [The 27-point AI visibility audit checklist](https://motionexalabs.com/field-notes/ai-visibility-audit-checklist/): the paid-audit checklist, published in full - [GEO vs SEO](https://motionexalabs.com/field-notes/geo-vs-seo/): comparison table and budget-split guidance - [llms.txt: an honest guide](https://motionexalabs.com/field-notes/llms-txt-guide/): format, worked example, honest adoption status - [How to measure AI search visibility](https://motionexalabs.com/field-notes/measure-ai-search-visibility/): the share-of-voice protocol, free to copy - [AI search statistics 2026](https://motionexalabs.com/field-notes/ai-search-statistics-2026/): 20+ sourced figures on AI-era buying; free to cite with attribution - [How to choose a GEO agency](https://motionexalabs.com/field-notes/how-to-choose-a-geo-agency/): honest provider-landscape guide with full disclosure - [AI visibility for cybersecurity companies](https://motionexalabs.com/field-notes/ai-visibility-for-cybersecurity-companies/): why security buyers moved into AI research first, and the reputation-prompt trap - [AI visibility for B2B SaaS](https://motionexalabs.com/field-notes/ai-visibility-for-b2b-saas/): the re-routed 2026 funnel, stage by stage ## Optional - [Privacy](https://motionexalabs.com/privacy/): no analytics, no cookies, form data by email only - [Terms](https://motionexalabs.com/terms/): short human-readable terms; evidence guarantee defined - [llms-full.txt](https://motionexalabs.com/llms-full.txt): full text of key pages in one file --- # Full page texts (generated 2026-06-10) ==================== ## Home ==================== first noticed this in a sales call, Feb '26 — buyer arrived with a shortlist we weren't on Entry № 014· AI visibility field notes Logged June 2026 # ChatGPT recommends your competitors. Not you. Buyers now ask AI for a vendor shortlist before they ever open Google. Three to five names make the answer — there is no page two. We run the prompts your market asks, log who gets cited, and engineer your brand into the answer set. Request an audit Read a sample audit Logged across — ChatGPT · Perplexity · Gemini · Claude · Google AI Overviews Visibility ledger · specimen page Run 03 | | Buyer prompt | Who gets cited | You | best [category] tools for startups | +2 | Absent | [competitor] alternatives | +4 | Absent | [category] pricing compared | +2 | Absent | is [your brand] legit? | thin / outdated | Wrong | top [category] platforms 2026 | +3 | Goal ✓ | Share of answer, this page: 0 / 4 prompts — the audit maps all 40+ every figure here is sourced — see the footnotes Entry № 015· The shift, on the record ## This is not a forecast. It already happened. The numbers below come from independent 2026 research, not from us. We keep them on file because most founders we audit are surprised by how far along this is. Finding 01 94% of B2B buyers used a large language model during their purchase process, per Forrester's 2026 buyer research.1 Finding 02 51% of B2B software buyers now start research in an AI chatbot rather than a search engine — and 69% changed vendor choice based on what the AI said.2 Finding 03 5.1× AI-referred visitors convert at roughly five times the rate of traditional organic traffic (14.2% vs 2.8%).3 Two details make this urgent for early-stage companies in particular. First, 37% of domains cited by AI engines never appear in classic search results4 — meaning a young brand can win citations long before it could ever win rankings. Second, roughly 70% of AI-referred traffic arrives with no referrer header,5 so most teams cannot see this channel in their analytics at all. Absence is invisible until a buyer shows up with a shortlist you are not on. - Forrester, Buyers' Journey Survey, 2026 — 18,000+ business buyers; generative AI ranked the most meaningful vendor-research source. ↩ - G2, The Answer Economy, April 2026 — survey of 1,076 B2B software decision-makers. ↩ - Exposure Ninja conversion analysis, March 2026. ↩ - Zhang et al., arXiv, December 2025 — citation overlap study. ↩ - Digital Bloom AI-traffic referrer analysis, February 2026. ↩ Entry № 016· Exhibit A — what an answer looks like ## The whole market fits in one paragraph. When a buyer asks an answer engine for "the best X," the reply is a short, confident paragraph naming a handful of vendors. That paragraph is the market now. Our job is to move a brand from outside it to inside it — and to document every step so you can verify the change yourself. The specimen on the right is illustrative (category and names generalized), but the structure is exactly what we log in every audit: which engines answer, who they name, which sources they lean on, and where the entry point is. Before After the work Specimen · run 01 best session-replay tools for an early-stage SaaS For an early-stage SaaS, the most commonly recommended options are: - — generous free tier - — strongest privacy controls - — best replay-to-analytics link Your brand: not mentioned. Not compared. Not considered. sources leaned on: two listicles (2024), one community thread, one review site Specimen · run 04, after fixes best session-replay tools for an early-stage SaaS For an early-stage SaaS, strong options include: - — generous free tier - Your brand — Cited ✓ positioned for small teams, with current pricing - — best replay-to-analytics link new sources in the mix: your comparison page, a review profile, one earned mention order matters — never write content before the prompt log exists Entry № 017· The method, in five lines ## Field work, not guesswork. - ### Build the prompt set 40+ real buyer prompts for your category — shortlists, comparisons, "alternatives to," and the question that decides deals: "is [your brand] legit?" prompt log - ### Run the engines Every prompt across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, with screenshots and source extraction for every answer. visibility ledger, run 01 - ### Trace the citations We map which pages the engines actually lean on in your category — review sites, comparison posts, docs, communities — and where your entry points are. source map - ### Fix what blocks you A prioritized worksheet: entity and schema corrections, citable comparison pages, llms.txt, review-profile gaps, and the two or three earned mentions that matter most. fix worksheet, ranked - ### Re-run and verify Same prompts, same engines, weeks later. The delta in your share of answer is the report — verifiable by anyone with a chat window. before/after delta The full method, documented → Entry № 018· From the field notes ## We publish what we learn. Working notes from live audits — the same material we use internally. No gating, no email walls. Note № 009Citation mechanicsJune 2026 ### How ChatGPT decides which brands to cite (and which to skip) Retrieval, the Bing index, authority thresholds, and why 65% of cited pages come from high-authority domains — with the early-brand workarounds. Note № 008WorksheetJune 2026 ### The 27-point AI visibility audit checklist The exact worksheet we run for clients — prompts, technical checks, entity checks, citation sources — free to copy. No. 007Logged June 20266 min read ### AI search statistics 2026 — every number sourced The reference sheet: adoption, shortlist behavior, conversion economics, citation mechanics. Free to cite with attribution. All field notes → flat rates, posted publicly — no "book a call to find out" Entry № 019· Rate card, abridged ## Posted rates. No discovery-call theatre. 01 · The audit ### Visibility Audit $1,200 one-time · 7–10 days - Prompts logged40+ - Engines5 - Fix worksheetranked Details Most filed 02 · The sprint ### Citation Sprint $1,800/mo 3 months · audit included - We implementyes - Re-runsmonthly - Delta reportverifiable Details 03 · For agencies ### White-label desk From $600 per audit · your brand on it - Turnaround7 days - Resell margin2–3× - Your logoyes Details Founding-client terms: if run 01 doesn't surface at least ten prompts where competitors are cited and you aren't, the audit is free. Full rate card and guarantee → Entry № 020· Open a file ## Find out what the engines say when you're not in the room. Four short sections, two minutes. We reply within one business day with scope and start date — findings arrive as a short recorded walkthrough. No call unless you ask for one. Open intake form 7-A Or just email us ==================== ## Method ==================== Record · MethodRevised June 2026 # The method, fully documented. Most "AI SEO" is repackaged content marketing. Ours is field work: log what the engines actually say, trace why, change the inputs, and re-run the same prompts to prove the delta. Every step below produces a document you keep. - ### Build the prompt set (days 1–2) We draft 40+ prompts the way your buyers actually phrase them — category shortlists ("best X for Y"), competitor displacement ("alternatives to Z"), use-case queries, pricing comparisons, and reputation checks ("is [brand] legit?", "[brand] reviews"). You review and add the prompts your sales team hears on calls. prompt log (yours to keep) - ### Run the engines (days 2–4) Each prompt runs across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews. We capture the full answer, the brands named, the order they appear in, the sources each engine leaned on, and a screenshot — because answers drift and you deserve evidence, not summaries. visibility ledger, run 01 — spreadsheet + screenshots - ### Score share of answer (day 4) Your share of answer is the percentage of prompts where you're cited, weighted by position. We score it per engine and per prompt cluster, alongside the three competitors who currently own the answers. This is the baseline every later claim is measured against. share-of-answer scorecard - ### Trace the citation sources (days 4–6) Engines don't invent shortlists; they assemble them from pages they trust. We extract every source URL behind your category's answers and sort them: review platforms, comparison listicles, community threads, docs, and news. The result is a map of exactly which doors you need to walk through. source map — ranked entry points - ### Diagnose the blocks (days 6–8) Why are you absent? Usually some mix of: the engines can't resolve your entity (name collisions, thin schema, no Wikipedia/Wikidata trail), your site has nothing extractable to quote (no comparison pages, no plain-language claims, no llms.txt), or the sources engines trust have never heard of you (no review profiles, no earned mentions). diagnosis memo, in plain language - ### File the fix worksheet (days 8–10) Everything becomes a ranked worksheet — effort, impact, and owner for each line. Typical items: entity & schema corrections, a citable comparison page engines can quote, llms.txt and crawler access, two review-platform profiles, and the three earned mentions most likely to enter the citation pool. Audit clients can run it themselves; Sprint clients have us run it. fix worksheet — ranked, owned, dated - ### Re-run and report the delta (Sprint, monthly) Same prompts, same engines. We report movement as a before/after delta on the original scorecard, with fresh screenshots. If a number moves, you can verify it yourself in any chat window — that's the standard we hold every claim to. delta report + updated ledger Annotations· What we will and won't do We do - Show our prompts and raw runsalways - Write citable pages worth quotingyes - Pursue earned mentionsethically - Report misses as plainly as winson record We don't - Guarantee specific rankingsnobody can - Spam communities or fake reviewsnever - Publish misleading claims about younever - Hide work behind "proprietary" fogno A note on honesty: answer engines are probabilistic and their answers move. What we guarantee is the work, the evidence, and the founding-client terms on the rate card — not a specific sentence from a specific model. Anyone promising that is selling weather control. Open intake form 7-ASee a finished audit ==================== ## Sample audit (specimen) ==================== Record · SpecimenDemonstration file # A finished audit, page by page. This is a specimen audit for a fictional early-stage SaaS — "a session-replay tool for small product teams." Category, prompts and structure are exactly what clients receive; the data is a demonstration set, marked as such on every page. Your audit looks like this, with your numbers. Page 01· Share-of-answer scorecard ## The number everything hangs on. Forty-two prompts, five engines, 210 logged answers. The scorecard reduces them to one honest figure per engine: how often you appear at all, and how high in the answer when you do. In this specimen, the subject appears in 3 of 42 prompts (7% share of answer) while the category leader appears in 31 (74%). That gap — not a keyword ranking — is what the engagement exists to close. Scorecard · specimen dataRun 01 | | Engine | Prompts cited | Share | ChatGPT | 1 / 42 | 2% | Perplexity | 2 / 42 | 5% | Gemini | 0 / 42 | 0% | Claude | 0 / 42 | 0% | AI Overviews | 0 / 42 | 0% | Blended share of answer: 7% · category leader: 74% Page 02· The prompt log ## Every prompt, on the record. The log groups prompts into four clusters. A sample of each — in the real document, every row links to the full captured answer and screenshot. Cluster A · Shortlists - "best session replay tools 2026"absent ×5 - "session replay for startups"absent ×4 - "affordable hotjar-style tools"cited ×1 Cluster B · Displacement - "[leader] alternatives"absent ×5 - "[leader] vs [rival]"absent ×5 - "cheaper than [leader]"absent ×5 Cluster C · Use case - "watch user rage clicks tool"cited ×2 - "replay + privacy for EU users"absent ×5 Cluster D · Reputation - "is [subject] legit?"thin answer - "[subject] pricing"outdated Page 03· Source map ## Where the answers come from. Across the 210 answers, the engines leaned on 61 distinct URLs. Ranked by how many answers each fed, the top of the map decides the work: - Review platform, category pagefed 38 answers · subject absent - "Best tools" listicle (updated 2026)fed 31 · subject absent - Community thread, "what do you use?"fed 19 · subject absent - Leader's own comparison pagesfed 17 · hostile framing - Subject's websitefed 2 · nothing quotable The diagnosis memo translates this: the subject isn't being skipped by accident — it's absent from every source the engines trust, and its own site gives them nothing extractable to quote. Page 04· Fix worksheet, top of the stack ## Ranked fixes, with owners and dates. Fix worksheet · specimen, first 6 of 23 linesRanked by impact ÷ effort | | Fix | Why | Effort | Create review-platform profile + first 10 honest reviews | feeds the #1 source | 2 wks | Publish "[subject] vs [leader]" comparison page with current pricing | quotable, owns displacement prompts | 1 wk | Fix Organization/Product schema + entity disambiguation | engines mis-resolve the name | 2 days | Add llms.txt + verify AI crawler access | three crawlers currently blocked | 1 day | Pitch inclusion in the 2026 listicle (it updates quarterly) | feeds the #2 source | outreach | Answer the community thread honestly, with disclosure | feeds the #3 source | 1 hr | Full worksheet: 23 lines · 9 client-side · 14 we can run in a Sprint Page 05· What the re-run shows ## The delta is the report. Sprint engagements re-run the identical prompt log monthly. The specimen below shows the format: same prompts, same engines, movement stated plainly, every figure re-checkable in a chat window. We don't promise a specific delta — engines are probabilistic and categories differ. We promise the measurement will be honest, and the founding-client terms on the rate card put teeth behind run 01. Start your run 01 Delta report · specimen formatRun 01 → 04 | | Measure | Run 01 | Run 04 | Blended share of answer | 7% | illustrative ↑ | Engines citing ≥1 prompt | 2 / 5 | illustrative ↑ | Reputation prompts answered correctly | 0 / 6 | illustrative ↑ | Specimen format — your report carries your verified numbers ==================== ## Pricing & white-label ==================== Record · Rate cardPosted June 2026 # The rate card, in full. Flat rates, posted publicly, in writing — because a 90-minute "discovery call" before anyone tells you a price is a margin-protection move, not a service. For context: published 2026 market ranges put one-time GEO audits at $1,500–$7,500 and mid-market retainers at $3,000–$10,000+/month. We price under that deliberately while we build our public track record, and these founding rates are dated so you know they'll rise. 01 · One-time ### Visibility Audit $1,200 7–10 business days - Buyer prompts logged40+ - Engines run5 - Answers captured200+ - Source mapranked - Fix worksheet~20 lines - Recorded walkthrough20 min Request this Most filed 02 · Three-month sprint ### Citation Sprint $1,800/mo 3 months · audit included - Everything in the auditincluded - We implement the worksheetyes - Citable pages written2–3/mo - Earned-mention outreachongoing - Full re-run + delta reportmonthly - Slack/email accessyes Request this 03 · Ongoing ### Standing Desk $1,200/mo after a Sprint · month-to-month - Monthly re-run, all enginesyes - Drift & misinfo alertsyes - One citable pagemonthly - Quarterly prompt refreshyes - Cancel anytime30 days Request this Founding-client terms The guarantee: if run 01 doesn't surface at least ten prompts where competitors are cited and you aren't, the audit is free. (In our runs so far, that bar has never been close.) The trade: these are deliberately-low founding rates. In exchange we ask for a short testimonial and permission to publish your before/after delta — anonymized if you prefer. If you'd rather keep everything private, add 30%. Rates are reviewed quarterly and existing clients keep their rate. Record · Trade desk· For SEO & marketing agencies ## White-label: your logo on our ledger. Your clients are asking what you're doing about ChatGPT. Answering well means tooling, prompt logs, and a fulfillment habit you may not want to build this quarter. So don't: we run the audits and sprints under your brand, you own the relationship and the markup. Industry-standard reseller markup on GEO work runs 40–60%; most partners resell our audits at 2–3×. - White-label audit (your logo, your template)$600–$900 - White-label sprint fulfillment$1,100/mo - Volume (3+ audits/mo)$500 each - Turnaround7 business days - We talk to your clientonly if you ask How partners use it The typical motion: add an "AI Visibility Audit" line to your rate card at $1,500–$2,500. We deliver the document in your template within a week. The fix worksheet becomes your upsell — either your team implements it inside your existing retainer, or we fulfill a sprint behind you at wholesale. One pilot audit at partner rate, no commitment, is the normal way to start. Email the desk or mention "agency" on the intake form. Annotations· Fair questions about money Q.01Why are you cheaper than the market ranges you quote? Because we're early and we say so. Public 2026 benchmarks put serious productized audits around $1,500–$7,500; we're underneath that while our public case file grows, and the founding terms (testimonial + published delta) are the honest price of the discount. The rates on this page are dated and will rise — clients keep the rate they signed at. Q.02Do you guarantee we'll be cited by ChatGPT? No, and you should walk away from anyone who does — answer engines are probabilistic and nobody controls them. What we guarantee is the evidence standard (every claim re-checkable), the deliverables on this page, and run 01's findings guarantee above. Q.03What do you need from us to start? The intake form (two minutes), read access to your analytics if you want referral analysis included, and one review pass on the prompt set. Audits don't require calls; sprints involve a 30-minute kickoff. Q.04Can we just buy the audit and implement it ourselves? Yes — the worksheet is written so a competent marketer or founder can run it without us. About half of audit clients do exactly that; the other half convert to a sprint because the earned-mention work is the part nobody has time for. ==================== ## FAQ ==================== Record · QuestionsRevised June 2026 # Questions, answered plainly. Fourteen questions we actually get, answered the way we'd answer them on a call — including the uncomfortable ones about guarantees and why our rates are below market. Q.01 What is an AI visibility (GEO) audit, in one paragraph? It is a documented test of how often AI assistants — ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews — name your brand when buyers ask for recommendations in your category. We run 40+ real buyer prompts across five engines, log every answer, score your share of voice against competitors, trace which sources the engines lean on, and hand you a prioritized fix plan. GEO stands for generative engine optimization: the practice of improving those results. Q.02 How is GEO different from SEO? SEO earns you a ranked position on a results page; GEO earns you a mention inside the answer itself. The inputs overlap — clean structure, authority, third-party coverage — but the scoreboard is different: an AI answer names three to five vendors and there is no page two. Forrester's 2026 buyer study found 94% of B2B buyers now use an LLM during purchase research, so the two disciplines now matter in parallel. We wrote a full comparison in the field notes. Q.03 What exactly do I get for the $1,200 audit? Five deliverables, all yours to keep: (1) the prompt log — 40+ buyer prompts run across 5 engines with full transcripts; (2) a visibility scorecard — your citation share of voice versus up to 5 competitors; (3) a source map — the specific pages and third-party sites the engines cited when they skipped you; (4) a prioritized fix worksheet — typically 15–25 items ranked by expected impact and effort; (5) a 45-minute walkthrough call. Turnaround is 7 business days. Q.04 How long until I see results? Honest answer: the audit itself takes 7 business days. Fixes that we ship (schema, llms.txt, crawler access, Bing/IndexNow submission, extractable comparison pages) can surface in retrieval-based engines like Perplexity and ChatGPT browsing within days to a few weeks, because those engines read the live web. Visibility that depends on model training data or broad third-party consensus takes months. Anyone promising guaranteed citations in week one is selling you weather control. Q.05 Can you guarantee my brand gets cited? No one can, and you should walk away from anyone who says otherwise — the engines are probabilistic and change weekly. What we do guarantee, for founding clients: if the audit doesn't surface at least 10 buyer prompts where competitors are cited and you are not, the audit is free. We can guarantee the quality of the evidence, not the behavior of someone else's model. Q.06 Which engines do you cover? ChatGPT (including its web-browsing mode), Perplexity, Google AI Overviews, Gemini, and Claude. We weight them by where your buyers actually are: for most B2B SaaS that's ChatGPT and AI Overviews first, Perplexity for technical buyers. We log model and date on every transcript because answers drift. Q.07 How do you measure 'share of voice' in AI answers? Each prompt is run on each engine; every brand mention is logged with position (named first, in-list, or footnote-only), sentiment, and whether a link was cited. Your share of voice is the percentage of prompt-runs where you appear, weighted by position. Re-running the identical prompt set later gives a clean delta. The full methodology is published in the field notes — we don't believe in black-box scores. Q.08 Do you also write the content and make the fixes? The audit is diagnosis plus prescription. If you want the surgery too, the Citation Sprint ($1,800/month, 3-month engagement) executes the worksheet: schema and technical fixes, two extractable assets per month (comparison pages, statistics pages, answers pages), third-party placement work, and a monthly re-run of the prompt set so progress is measured, not vibes. Q.09 What do you need from us to start? For the audit: nothing but the form on this site — it's all external observation. For the sprint: CMS or repository access (or a developer on your side to ship our PRs), a logo-and-facts sheet, and one 30-minute call a month. We work async-first across time zones. Q.10 Your prices look low. Established firms charge $5,000–$7,500 for this. Why? Because we're new and we're honest about it. These are founding rates: you get below-market pricing and a guarantee; we get testimonials and published before/after deltas (with your written approval, or anonymized as 'a seed-stage devtools company'). Rates rise as the case file shelf fills. The deliverable list doesn't shrink at this price — the margin does. Q.11 Do you work with agencies? Can we resell this? Yes — white-label is a standing offer, not a side door. We fulfil audits under your brand from $600, sprints from $1,100/month, with a 7-day turnaround and your logo on every document. Agencies typically resell at 2–3×. Details are on the pricing page under 'For agencies.' Q.12 ChatGPT already mentions us sometimes. Do we still need this? Maybe not — and we'll tell you if so. 'Sometimes' is the operative word: being cited on 8% of buyer prompts while a competitor sits at 60% is the common pattern we find. The ten-prompt spot check we run before any sales call settles it with evidence. If you're already winning, we close the file and say so. Q.13 Is GEO just tricking the AI? Will this get us penalized? We don't do prompt injection, hidden text, fake reviews, or fabricated statistics — those tactics are detectable, short-lived, and reputationally radioactive. Everything in our worksheet is legible to a human: clearer structure, real data points, honest comparison pages, legitimate third-party coverage, correct crawler access. We make you easier to cite, not easier to fake. Q.14 We haven't launched yet. Too early? Pre-launch is actually a good moment for the cheaper Standing Desk monitoring or a scoped-down audit: you get the baseline of who owns your category's answers today, and your launch content ships citation-ready instead of being retrofitted. Stamp 'Pre-seed' on the intake form and say you're pre-launch in the notes. Something we didn't cover? hello@motionexalabs.com — or open a case file and ask in field 10. ==================== ## About ==================== Record · The labOpened 2026 # A small lab with a big ledger. Motionexa Labs is an independent generative engine optimization (GEO) studio. We audit and improve how often early-stage SaaS and cybersecurity companies are cited by ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews — and we publish our working notes as we go. ## Why this exists Somewhere in 2024–25, the buyer's first question stopped going to a search box and started going to a chat window. Forrester's 2026 study of 18,000 buyers found 94% of B2B buyers used an LLM during purchase research; G2's 2026 survey found 51% now start vendor research in an AI chatbot. The answer those tools give contains three to five names. There is no page two. For a funded startup that's an existential detail: you can have a beautiful product, a fresh round, and a category where the machines recommend your competitors by name every single day. Most founders discover this by accident. We built a practice around discovering it on purpose — with transcripts, scorecards and fix plans instead of vibes. ## Why "field notes" GEO is young and full of confident nonsense. Our answer to that is to work like a field researcher: log what the engines actually said, date it, trace why, change one input, re-run the identical prompt, and write down the delta. Everything we publish on this site follows that discipline — and every chart in our sample audit is clearly marked as a demonstration specimen, never passed off as a client result we haven't earned yet. File card — the lab - Founded2026 - DeskMysuru, India - Working hoursasync, worldwide - FocusSaaS · cybersecurity - Engines logged5 - Tracking scripts0 Standing orders - No fabricated client data, ever. Specimens are labeled specimens. - Every statistic we publish carries a named source. - If a prospect is already well-cited, we say so and close the file. - No guarantees about someone else's model — only about our evidence. - Founding clients get below-market rates in exchange for testimonials; that trade is stated in writing, not implied. ## Who's behind the desk Motionexa Labs is founder-operated — research, audits and client calls are run by the same person who signs this page, with specialist contractors for content production during sprints. Small is the point: you brief the person doing the work, and your prompt logs aren't handed to an account manager three layers away. Published research is credited to the Motionexa Research Desk, the lab's editorial byline. We're based in Mysuru, India, and work async with clients in US, EU and Indian time zones. The fastest way to test whether any of this is real is the free ten-prompt spot check that precedes every sales call — open a case file and see what the machines say about your category this week. Request an audit  Read the method ⟶ ==================== ## Field note 001 — /field-notes/what-is-generative-engine-optimization/ ==================== Field note no. 001Logged June 2026 · 9 min read # What is generative engine optimization (GEO)? By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines Generative engine optimization (GEO) is the practice of increasing how often a brand is cited inside AI-generated answers. It matters because 94% of B2B buyers now use LLMs in purchase research (Forrester, 2026), 51% start vendor research in a chatbot (G2, 2026), and AI-referred visitors convert about 5× better than organic search (Exposure Ninja, 2026). The work: measure citation share of voice, fix crawler access and structure, publish extractable assets, build third-party consensus, re-measure. ## The definition Generative engine optimization (GEO) is the practice of increasing how often, how prominently, and how accurately a brand is cited inside the answers produced by generative AI systems — ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude and their kin. Where SEO competes for a position on a results page, GEO competes for a mention inside the answer itself. The term was formalized in a 2024 academic paper by researchers at Princeton, Georgia Tech, IIT Delhi and the Allen Institute, who tested which content changes measurably raised citation rates in generative answers [1]. ## Why this became urgent in 2026 Three numbers explain the entire discipline: - 94% of B2B buyers used a large language model somewhere in their purchase research, per Forrester's 2026 Buyers' Journey Survey of roughly 18,000 buyers — generative AI is now their top vendor-research source [2]. - 51% of software buyers now start vendor research in an AI assistant rather than a search engine, and 69% say AI answers changed which vendors made their shortlist, per G2's 2026 buyer-behavior research [3]. - 5.1× — visitors arriving from AI assistants converted at 14.2% versus 2.8% for classic organic search in Exposure Ninja's 2026 cross-client analysis, because the assistant has already pre-sold the recommendation [4]. An AI answer typically names three to five vendors. There is no page two, no position eleven, no "next 10 results." Either the model says your name or it says someone else's. ## How generative engines actually pick brands Two pipelines feed every answer. Parametric memory — what the model absorbed during training — favors brands with broad, consistent coverage across the open web over years. Retrieval — live search the engine performs before answering (Perplexity always; ChatGPT and Gemini when browsing; AI Overviews by design) — favors pages that are crawlable, extractable, recently updated, and corroborated by third parties. Research on ChatGPT's citation patterns found 65.3% of its most-cited pages sit on high-authority domains (DR 80+) [5], and an analysis published in December 2025 found roughly 37% of domains cited by AI assistants never appear in the top classic search results for the same queries [6] — the two games genuinely diverge. ## What GEO work actually consists of - Measurement first. Build a prompt set the way buyers phrase questions, run it across engines on a schedule, and log citation share of voice. (Our measurement methodology is public.) - Access & structure. Let AI crawlers in (GPTBot, ClaudeBot, PerplexityBot, Google-Extended), publish an llms.txt, ship schema markup, make pages readable without JavaScript. - Extractable assets. Comparison pages, statistics pages, plain-spoken FAQ blocks — formats engines can quote. The Princeton-led study measured a 30–40% citation lift simply from adding citable statistics and quotations to pages [1]. - Third-party consensus. Engines distrust self-praise. Review platforms, comparison sites, Reddit threads, niche directories and press are where they verify you exist — G2 found review-site presence the single strongest trust signal buyers and engines lean on [3]. - Re-measurement. Same prompts, same engines, later date. The delta is the deliverable. ## Who actually needs GEO (and who doesn't) If your buyers research before buying — B2B SaaS, security, devtools, fintech, professional services — you have an AI-visibility position whether you manage it or not. Early-stage companies have the most to gain: the incumbents own the training data, but retrieval-based answers can be moved in weeks, not years. If you sell impulse consumer goods through paid social, GEO is not your bottleneck. ## Does GEO replace SEO? No — it sits beside it, sharing perhaps 60% of the work and none of the scoreboard. Gartner projects traditional search volume falling roughly 25% by 2026 as assistants absorb informational queries [7], but search isn't disappearing; it's being re-weighted. The full comparison is in GEO vs SEO. ## Questions people ask Q.01 What does GEO stand for? Generative engine optimization — the practice of increasing how often and how favorably a brand is cited in answers produced by generative AI systems such as ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews. The term comes from a 2024 academic paper by researchers at Princeton, Georgia Tech, IIT Delhi and the Allen Institute. Q.02 Is GEO the same as AEO or AI SEO? The terms overlap heavily. AEO (answer engine optimization) historically meant optimizing for featured snippets and voice assistants; 'AI SEO' and 'LLM SEO' are looser marketing labels. GEO is the most precise term for optimizing presence inside generative answers, and the one used in the academic literature. Q.03 How long does GEO take to show results? Retrieval-driven surfaces (Perplexity, ChatGPT browsing, AI Overviews) can reflect on-site fixes and new extractable pages within days to a few weeks once crawled. Visibility rooted in model training data and broad third-party consensus moves over months. Any vendor guaranteeing specific citations on a specific date is overselling. Q.04 Can small companies realistically compete with incumbents in AI answers? Yes, selectively. Incumbents dominate parametric memory, but retrieval favors fresh, well-structured, corroborated content — and roughly 37% of AI-cited domains don't appear in top classic search results, which means the door is open to brands that never won at SEO. Niche, specific prompts ('best X for Y-type company') are where early-stage brands win first. ## Sources & further reading - [1] Aggarwal et al., "GEO: Generative Engine Optimization," SIGKDD 2024 (Princeton, Georgia Tech, IIT Delhi, Allen Institute) - [2] Forrester, Buyers' Journey Survey 2026 (~18,000 global B2B buyers) - [3] G2, "The Answer Economy" buyer research, April 2026 (1,076 B2B software buyers) - [4] Exposure Ninja, AI-referral conversion analysis, March 2026 - [5] Ahrefs, study of domains cited by ChatGPT, 2025 - [6] Zhang et al., analysis of AI-assistant citation domains vs. classic SERPs, arXiv preprint, December 2025 - [7] Gartner, prediction on traditional search volume decline, 2024–2026 coverage From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 002 — /field-notes/how-chatgpt-chooses-which-brands-to-cite/ ==================== Field note no. 002Logged June 2026 · 8 min read # How ChatGPT chooses which brands to cite By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines ChatGPT names brands from two memories: training data (built from years of consistent third-party coverage) and live retrieval (Bing-powered search over crawlable pages). The strongest levers, in order: independent corroboration on review and comparison sites (G2 found review citations the #1 trust signal), presence on high-authority pages (65.3% of ChatGPT-cited pages are DR 80+ — Ahrefs), extractable content with statistics and definitions (+30–40% citation lift — Princeton-led GEO study), crawler access and freshness, and consistent entity definition. Tricks like prompt injection and fake reviews backfire. ## One answer, two memories When ChatGPT names brands, the names come from one of two places. Training data (parametric memory): the statistical residue of everything the model read — years of articles, docs, forums, review sites. Brands mentioned consistently across many independent sources get "burned in." Live retrieval: when browsing is on (and for most shopping/vendor questions it now is), ChatGPT issues web searches — historically powered by Bing's index — reads a handful of pages, and synthesizes. Different levers move each memory, which is why this distinction is the first thing we map in every audit. ## The signals, ranked by what the evidence shows ### 1 · Third-party consensus beats self-description Models are trained to be skeptical of self-praise. A claim that appears on your homepage counts once; the same claim echoed by a review platform, two comparison posts and a Reddit thread reads as fact. G2's 2026 buyer research found review-site citations the #1 trust signal in AI-assisted buying, and one in three buyers purchased from a vendor they hadn't previously known because an AI surfaced it [1]. ### 2 · Domain authority still casts a long shadow Ahrefs' analysis of pages ChatGPT cites found 65.3% live on DR 80+ domains [2]. You can't buy that overnight — but you can borrow it by being present on the high-authority pages the engine already trusts: category roundups, review platforms, comparison sites. Getting onto the page that gets cited is often faster than making your page the cited one. ### 3 · Extractability — can a machine lift the answer cleanly? The original GEO study measured what actually raises citation rates: adding statistics, quotations and citations to a page improved its odds of being cited by roughly 30–40% [3]. Engines quote what's quotable: dated numbers, definition-shaped sentences, honest comparison tables, FAQ blocks. Walls of brand prose get skimmed and skipped. ### 4 · Crawlability and freshness Retrieval can't cite what it can't read. Blocking GPTBot or OAI-SearchBot in robots.txt, hiding content behind JavaScript, or letting pages go stale all suppress retrieval-side visibility. Recency matters: answer engines over-select recently updated pages for "best X 2026"-style prompts. ### 5 · Entity clarity The model has to know what you are before it can recommend you. Consistent naming, a clear one-line category definition repeated across your site and profiles, Organization schema, and presence in the places that define entities (Wikipedia-grade sources, Crunchbase, LinkedIn, major directories) reduce the odds of being mis-categorized or hallucinated about. ## What doesn't work (and can backfire) Keyword-stuffed "AI optimization" pages, hidden prompt-injection text, fake reviews, fabricated statistics. The first is ignored; the rest are detectable, get scrubbed at the source, and burn the third-party trust you actually need. Persuasive-but-empty language performed near the bottom of tested tactics in the GEO study [3]. ## The practical order of operations - Baseline: run 40+ buyer prompts, log who gets cited and from which sources (the source map matters more than the score). - Unblock and structure: crawler access, schema, server-rendered content, llms.txt. - Publish what engines quote: statistics pages, comparison pages, FAQ blocks — see the 27-point checklist. - Earn consensus: reviews, roundup inclusion, communities your buyers actually read. - Re-run the identical prompt set monthly; manage the delta, not the anecdote. ## Questions people ask Q.01 Does ChatGPT use Bing to find brands? When browsing/search is active, ChatGPT's retrieval has historically been powered by Bing's index, supplemented by OpenAI's own crawling (GPTBot for training, OAI-SearchBot for search). Practical consequence: being indexed and well-described in Bing — many SaaS teams never check — directly affects whether ChatGPT can retrieve you. Submitting your site to Bing Webmaster Tools and using IndexNow is cheap and routinely skipped. Q.02 Why does ChatGPT recommend my competitor and not me? Usually one or more of: the competitor appears on the high-authority roundup/review pages retrieval keeps citing; their site states what they are in clean, extractable language while yours requires interpretation; they have more independent corroboration (reviews, comparisons, community mentions); or your site blocks AI crawlers. An audit's source map shows precisely which pages the engine leaned on when it skipped you. Q.03 Can I pay to be recommended by ChatGPT? No. There is no paid placement inside organic ChatGPT answers as of mid-2026. Anyone selling 'guaranteed ChatGPT placement' is selling either ordinary GEO work with dishonest framing, or nothing. Q.04 How often do ChatGPT's brand answers change? Retrieval-backed answers can change week to week as the index refreshes and source pages update. Training-data-backed answers shift on model updates. This volatility is why we log model and date on every transcript and re-run identical prompt sets monthly rather than judging from one screenshot. ## Sources & further reading - [1] G2, "The Answer Economy," April 2026 — review-site citations as top trust signal; 1 in 3 bought from a previously unknown vendor - [2] Ahrefs, analysis of domains cited by ChatGPT (DR distribution), 2025 - [3] Aggarwal et al., "GEO: Generative Engine Optimization," SIGKDD 2024 — tactic-level citation-lift measurements - [4] OpenAI documentation on GPTBot and OAI-SearchBot crawler behavior From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 003 — /field-notes/ai-visibility-audit-checklist/ ==================== Field note no. 003Logged June 2026 · 10 min read # The 27-point AI visibility audit checklist By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines A 27-point, five-section checklist for auditing your own AI visibility: (A) build a 40-prompt baseline across five engines and map the cited sources; (B) unblock GPTBot/ClaudeBot/PerplexityBot/Google-Extended, verify Bing indexation, enable IndexNow, render without JS, publish llms.txt; (C) Organization + FAQPage + Article schema and a consistent one-line category definition; (D) statistics pages, honest comparison tables, answer-first writing; (E) get onto already-cited roundups, review platforms and communities, then re-run the identical prompt set monthly. This is the working checklist behind our paid audits, published in full. It tells you what a complete audit inspects. The operational layer — the pass/fail criteria for each item, the prompt-set design, and the competitive source mapping — ships only with the paid audit; that's the part of the craft we keep on our side of the desk. Items are ordered so the cheap, high-leverage work comes first. ## A · Baseline & evidence (items 01–05) - 01 · Build a real prompt set — 40+ prompts phrased the way buyers talk: category shortlists, “alternatives to [competitor]”, use-case and pricing questions, “is [brand] legit?” - 02 · Run all five engines — ChatGPT (with browsing), Perplexity, Google AI Overviews, Gemini, Claude — same prompts, logged with model and date. - 03 · Score share of voice — % of prompt-runs where you're named, weighted by position (first / in-list / footnote). This is your baseline number. - 04 · Map the sources — For every answer that skipped you, record which pages the engine cited instead. The source map is the to-do list. - 05 · Log competitor framing — Note the exact adjectives engines attach to rivals — that language is what you're competing against, not keywords. ## B · Crawl & access (items 06–12) - 06 · Unblock AI crawlers — robots.txt must allow GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended (and decide deliberately on CCBot/Bytespider). Many CDN bot-protection defaults silently block them. - 07 · Verify Bing indexation — ChatGPT retrieval has leaned on Bing. Register in Bing Webmaster Tools; fix anything unindexed. - 08 · Enable IndexNow — Push new/updated URLs to Bing instantly instead of waiting for a crawl — one key file plus a ping. - 09 · Render without JavaScript — curl your key pages; if the answer to “what is this product?” isn't in the raw HTML, most AI crawlers never see it. - 10 · Publish llms.txt — A plain-text site guide for language models at /llms.txt — low cost, growing convention. We wrote a full guide. - 11 · Keep clean canonical URLs — One URL per idea, no parameter soup, fast responses (under ~1s TTFB helps crawl budget). - 12 · Check your own brand SERP on Bing — What Bing shows for “[brand]” is the raw material retrieval reads first. Fix stale titles and descriptions. ## C · Entity & schema (items 13–18) - 13 · Organization schema sitewide — Name, logo, sameAs links to LinkedIn/Crunchbase/GitHub — give entity resolvers an anchor. - 14 · Schema on money pages — Service/Product with price where honest, FAQPage on real Q&A, Article with dates on posts, BreadcrumbList everywhere. - 15 · One-sentence category definition — “[Brand] is a [category] for [audience] that [differentiator].” Repeat it verbatim on homepage, about, LinkedIn, directories. - 16 · Consistent naming everywhere — One spelling, one capitalization, one tagline across web properties — inconsistency fragments the entity. - 17 · Claim the entity surfaces — Crunchbase, LinkedIn, G2/Capterra (B2B), GitHub (devtools), relevant niche directories — the places models learn entities from. - 18 · Date your pages — Visible “last updated” + dateModified in schema; answer engines over-select fresh pages for “best X 2026” prompts. ## D · Content extractability (items 19–23) - 19 · Publish a statistics page — Original or well-aggregated numbers with dates and sources — the single most-cited asset type (+30–40% lift for citable stats in the SIGKDD GEO study). - 20 · Write honest comparison pages — “[You] vs [competitor]” with a real table including where the rival wins. Engines quote balanced tables; puff pieces get skipped. - 21 · Add FAQ blocks with real questions — Phrase them exactly as buyers ask; answer in the first sentence; mark up with FAQPage. - 22 · Lead with the answer — Every key page should resolve its question in the first 60 words — models extract openings, not conclusions. - 23 · Make tables, not paragraphs, for comparisons — Structured rows survive extraction; adjectives don't. ## E · Third-party footprint (items 24–27) - 24 · Get on the pages already being cited — Your source map (item 04) lists them — roundups, “best X” posts, review categories. Pitch inclusion before building anything new. - 25 · Mind your review-platform presence — G2 research puts review-site citations as the top trust signal in AI-assisted buying; even a handful of detailed, recent reviews changes framing. - 26 · Show up where buyers actually discuss — Reddit, Hacker News, niche Slacks/forums — engines weight community consensus; participate honestly or not at all. - 27 · Re-run the identical prompt set monthly — Same prompts, same engines. The month-over-month delta — not any single screenshot — is the only honest scoreboard. ## Questions people ask Q.01 How long does this checklist take to complete? Section B (crawl & access) is a focused afternoon for a developer. Sections C and D are 2–4 weeks of part-time work for a small team. Section E never finishes — third-party consensus is a practice, not a task. The baseline in section A takes a day the first time and an hour a month thereafter. Q.02 Which single item moves the needle most? For companies starting from zero: item 06 (unblocking AI crawlers) plus item 07 (Bing indexation), because everything else is invisible until retrieval can read you. For companies already crawlable: item 19, the statistics page — citable numbers were the strongest tested lever in the SIGKDD GEO study. Q.03 Do I need special tools to run the baseline? No. A spreadsheet, the five engines, and discipline about running identical prompts on a schedule. Dedicated monitoring platforms (Profound, Peec, Otterly and similar) automate the cadence and are worth it once you're managing the number monthly — our measurement guide covers the trade-offs. ## Sources & further reading - [1] Aggarwal et al., "GEO: Generative Engine Optimization," SIGKDD 2024 — content-tactic citation lifts - [2] G2, "The Answer Economy," April 2026 — review-platform trust signal - [3] OpenAI, Anthropic, Perplexity, Google crawler documentation (GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended) - [4] Microsoft Bing — Webmaster Tools and IndexNow documentation From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 004 — /field-notes/geo-vs-seo/ ==================== Field note no. 004Logged June 2026 · 8 min read # GEO vs SEO: what actually changes By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines SEO competes for ranked links; GEO competes for named slots inside AI answers, where only 3–5 brands appear and page two doesn't exist. They share inputs (structure, authority, third-party coverage) but diverge on index (Bing matters for ChatGPT), winning formats (statistics and tables over long ranking posts), and measurement (70.6% of AI traffic shows no referrer — Digital Bloom 2026). With search volume projected down ~25% (Gartner) and 51% of buyers starting in chatbots (G2), the 2026 posture is SEO-maintained, GEO-added: reallocate 15–25%. ## Same web, different scoreboard SEO and GEO share most of their inputs — clean architecture, authority, third-party coverage — and almost none of their outputs. SEO wins you a ranked link a buyer may click; GEO wins you a named slot inside an answer a buyer reads as advice. The overlap means good SEO is a head start. The differences mean it is not a finish. ## The comparison, honestly drawn | | Dimension | Traditional SEO | GEO | Surface | Ranked results page | Synthesized answer (3–5 names total) | Unit of success | Position & click | Citation, position in answer, framing | “Page two” | Exists; weak traffic | Does not exist | Primary index | Google | Bing (ChatGPT), Google (AIO/Gemini), engine-own crawlers | Key off-site signal | Backlinks | Corroborated mentions: reviews, comparisons, communities | Content that wins | Comprehensive ranking pages | Extractable: statistics, definitions, honest tables, FAQs | Measurement | Rank trackers, Search Console | Prompt-set share of voice; mostly no referrer trail | Volatility | Algorithm updates (months) | Model/index updates (weeks) | Traffic volume | High | Lower — but converting at ~14.2% vs 2.8% (Exposure Ninja, 2026) [1] ## What the 2026 numbers say about the shift - Gartner's widely-cited projection: traditional search engine volume down roughly 25% by 2026 as chatbots absorb informational queries [2]. - Digital Bloom's traffic analysis found 70.6% of AI-assistant-driven visits arrive with no referrer — they look like “direct” in analytics, so most teams systematically under-count AI's contribution [3]. - 37% of domains cited by AI assistants never appear in top classic results for the same queries (Zhang et al., 2025) — proof the two games select different winners [4]. - Buyer behavior already moved: 51% start research in an AI chatbot; 68% of CMOs say they begin vendor discovery in AI before Google (G2 2026; Wynter 2026) [5]. ## Where one effort feeds both Server-rendered, fast, well-structured pages; Organization/FAQ/Article schema; genuinely useful comparison content; earned coverage on authoritative third-party sites. Do these once, score twice. The GEO-only additions are cheap: AI-crawler permissions, llms.txt, Bing/IndexNow hygiene, statistics-first assets, and a prompt-set measurement habit. ## How to split a small budget in 2026 Our working rule for early-stage B2B: keep the shared foundation funded (it serves both), put incremental content dollars toward extractable, citable assets rather than another 2,000-word ranking post, and put incremental measurement dollars toward a monthly prompt-set baseline — because you cannot manage a channel your analytics literally cannot see. Companies with mature SEO should treat GEO as a 15–25% reallocation, not a new line item. ## Questions people ask Q.01 Should I stop doing SEO and switch to GEO? No. Search still carries the volume; AI answers increasingly carry the decision. The rational posture for most B2B companies in 2026 is SEO-maintained, GEO-added — roughly a 15–25% reallocation of search budget toward extractable assets, crawler/index hygiene, and monthly AI-answer measurement. Q.02 Does ranking #1 on Google mean AI engines will cite me? Helpful but not sufficient. AI Overviews lean on Google's index, but ChatGPT's retrieval has leaned on Bing, and all engines re-rank for extractability and corroboration. The 2025 Zhang et al. analysis found 37% of AI-cited domains absent from top classic results — and the reverse failure (ranking well, never cited) is what most of our audits document. Q.03 Why doesn't my analytics show AI traffic? Because roughly 70% of AI-driven visits arrive with no referrer header and land in 'Direct' (Digital Bloom, 2026). Partial fixes: segment by landing pages AI engines favor, watch Bing Webmaster data, add 'how did you hear about us?' to signup, and treat prompt-set share of voice — not referral traffic — as the primary KPI. ## Sources & further reading - [1] Exposure Ninja, AI-referral vs organic conversion analysis, March 2026 - [2] Gartner, search-volume decline projection (2024 forecast for 2026) - [3] The Digital Bloom, AI traffic referrer analysis, February 2026 - [4] Zhang et al., AI citation domains vs classic SERPs, arXiv, December 2025 - [5] G2 "The Answer Economy" (April 2026); Wynter CMO survey (2026) From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 005 — /field-notes/llms-txt-guide/ ==================== Field note no. 005Logged June 2026 · 7 min read # llms.txt: an honest guide, with a working example By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines llms.txt is a Markdown file at your site root giving language models a curated map: one-paragraph definition, key facts, annotated links. Proposed by Jeremy Howard (Answer.AI, Sept 2024). Honest status in 2026: AI crawlers fetch it and adoption is growing, but no major engine guarantees it affects citations — treat it as an hour of cheap insurance plus a positioning exercise, paired with actual crawler access. Worked example and five writing rules included; this site's own file is at /llms.txt. ## What llms.txt is llms.txt is a plain-Markdown file served at your site root (/llms.txt) that gives language models a curated map of your site: what you are in one paragraph, then annotated links to the pages that matter, in an order you choose. The proposal comes from Jeremy Howard (Answer.AI) in September 2024 [1]. A companion convention, llms-full.txt, inlines the full text of key pages for models that want depth without crawling. ## The honest status report (mid-2026) Be clear-eyed about this: no major engine has committed to llms.txt as a ranking input. Adoption is a growing convention — strongest among developer-tool and AI companies (Anthropic, Mintlify-hosted docs, many devtools publish one) — and AI crawlers do fetch the file when present. Treat it as cheap insurance and a curation exercise, not a hack: it costs an hour, can't hurt, gives retrieval systems a clean self-description to lift, and the discipline of writing it usually exposes how muddled your positioning is. Anyone selling llms.txt as a guaranteed visibility lever is overselling a text file. ## The format Markdown, with a loose spec [1]: an H1 with your name, a blockquote summary, optional context paragraphs, then H2-grouped link lists with one-line annotations. An Optional section marks pages models can skip under length pressure. ## A worked example for a B2B SaaS # Acme Deploy > Acme Deploy is a deployment automation platform for > small engineering teams. Flat pricing, SOC 2 Type II, > founded 2024. Key facts: 5-minute setup; integrates GitHub/GitLab; plans from $49/month. ## Product - [How it works](https://acme.dev/how/): 90-second overview - [Pricing](https://acme.dev/pricing/): all plans, no "call us" - [Security](https://acme.dev/security/): SOC 2, data handling ## Comparisons - [Acme vs Octopus Deploy](https://acme.dev/vs/octopus/) - [Acme vs Jenkins](https://acme.dev/vs/jenkins/) ## Optional - [Changelog](https://acme.dev/changelog/) - [Blog](https://acme.dev/blog/) ## Five rules that make it useful - First paragraph = your category definition, phrased the way a buyer would ask. This is the sentence you want quoted. - Curate ruthlessly — 10–25 links beats 200. It's a reading list, not a sitemap (you already have a sitemap). - State facts machines can lift: pricing, compliance, founding year, integrations — concrete and dated. - Keep it in sync — a stale llms.txt confidently feeding old pricing to a model is worse than none. Ours regenerates with every deploy. - Pair it with crawler access — the file does nothing if robots.txt blocks GPTBot, ClaudeBot and PerplexityBot (checklist items 06–12 in the audit checklist). You can read this site's own implementation at /llms.txt — we eat the cooking. ## Questions people ask Q.01 Do ChatGPT, Perplexity or Google actually read llms.txt? AI crawlers fetch the file when present, and some retrieval pipelines use it as a site summary — but as of mid-2026 no major engine has publicly committed to it as a ranking or citation input. Honest framing: a low-cost, growing convention that gives models a clean self-description, not a proven lever. The hour it takes is justified by insurance value and the positioning clarity it forces. Q.02 What's the difference between llms.txt and robots.txt? robots.txt is access control — which crawlers may fetch which paths. llms.txt is editorial guidance — what your site is and which pages best explain it. They're complementary: robots.txt opens the door, llms.txt hands the visitor a reading list. Q.03 What's llms-full.txt? A companion file that inlines the complete text of your key pages, so a model can ingest your essential content in one fetch without crawling. Useful for documentation-heavy products; for a typical marketing site, a tight llms.txt plus crawlable pages is enough. Q.04 Where should the file live and in what format? Plain Markdown served as text at https://yourdomain.com/llms.txt — site root, no authentication, no redirect chain. Subdirectories or CMS attachments defeat the convention since tools look only at the root path. ## Sources & further reading - [1] Jeremy Howard / Answer.AI, the llms.txt proposal and specification, September 2024 (llmstxt.org) - [2] Anthropic and major devtools documentation sites publishing llms.txt implementations, 2025–2026 - [3] OpenAI / Anthropic / Perplexity crawler documentation on fetch behavior From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 006 — /field-notes/measure-ai-search-visibility/ ==================== Field note no. 006Logged June 2026 · 9 min read # How to measure AI search visibility By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines You can't manage AI visibility from analytics — ~70.6% of AI-driven visits arrive with no referrer (Digital Bloom, 2026). Measure it like field science: freeze a 40+ prompt panel across five buyer intents, run it monthly in fresh sessions on ChatGPT, Perplexity, AI Overviews, Gemini and Claude, score position-weighted share of voice, log framing adjectives and cited sources, and report the month-over-month delta on identical prompts. Platforms like Profound/Peec/Otterly automate cadence once the number goes to executives; a disciplined spreadsheet is sufficient to start. ## The measurement problem AI answers leave almost no exhaust. Roughly 70.6% of AI-assistant-driven visits arrive with no referrer and hide inside "Direct" traffic [1], and the answers themselves are personalized, probabilistic and re-generated on every ask. So you cannot manage AI visibility from analytics. You measure it the way field science measures anything unstable: fixed instrument, repeated observation, logged conditions. Here is the exact protocol behind our audits — free to copy. ## Step 1 — Build the prompt set (the instrument) 40+ prompts across five intents, phrased as buyers phrase them, frozen once written: | | Intent | Share | Examples | Category shortlist | ~35% | "best [category] for [segment]", "top [category] tools 2026" | Competitor displacement | ~20% | "alternatives to [rival]", "[rival] vs [rival2], which is better?" | Use-case | ~20% | "how do I [job-to-be-done]", "[problem] solution for [stack]" | Commercial detail | ~15% | "[category] pricing comparison", "cheapest [category] with [feature]" | Reputation | ~10% | "is [brand] legit?", "[brand] reviews", "[brand] security record" ## Step 2 — Run cadence and logging (the observation) Each prompt runs on each engine — ChatGPT (browsing on), Perplexity, AI Overviews, Gemini, Claude — in a fresh session, monthly, on a fixed week. Every run is logged with: date, engine, model version, full answer text, brands named, position of each, links cited, and one-line sentiment. One run is an anecdote; the panel over months is data. ## Step 3 — Score share of voice (the metric) Each run is scored on a position-weighted scale — a first-place recommendation counts for more than a list mention, which counts for more than a footnote citation. (The exact weighting is part of the paid methodology; any consistent scale works for your own tracking.) Share of voice = total points ÷ total runs, reported overall and per engine and per intent. Track three companions: coverage (% of prompts where you appear at all), framing (the adjectives engines attach to you, logged verbatim), and citation source mix (which of your/others' pages the engines actually leaned on). ## Step 4 — The delta is the deliverable Re-run the identical, frozen prompt set. Report month-over-month change per intent cluster — that isolates the effect of work shipped between runs. When we publish results, this is the format: same 40 prompts, same engines, dated transcripts, before-column, after-column. Anything else is screenshot theater. ## Manual ledger vs monitoring platforms Dedicated platforms — Profound, Peec, Otterly, and a fast-growing field [2] — automate daily runs, larger prompt panels, and alerting; sensible once AI share of voice is a number an executive asks about monthly. The manual ledger (a spreadsheet and discipline) costs nothing, teaches you what the engines are doing, and is fully sufficient for a quarterly baseline. We run audits manually with scripted logging because the source-reading is the analysis; we recommend platforms to clients who graduate to continuous monitoring. ## What to report upward Three lines on one page: AI share of voice (the weighted score, trended), coverage on money prompts (the 10 prompts closest to revenue), and assisted evidence — "how did you hear about us?" mentions of AI, plus conversion rate of suspected-AI landing sessions (recall AI-referred visitors converted ~5× organic in 2026 data [3]). Resist inventing precision the channel doesn't offer. ## Questions people ask Q.01 How many prompts do I need for a reliable baseline? 40+ across five intent types is our floor for a defensible read on one category; below ~25 a couple of volatile answers swing your score by double digits. Enterprises tracking many segments run hundreds via monitoring platforms, but a frozen 40-prompt panel run monthly beats a 500-prompt panel run once. Q.02 Why do I get different answers than my colleague for the same prompt? Personalization (account history, location, model routing) plus sampling randomness. That's exactly why the protocol uses fresh sessions, multiple engines, a large prompt panel and repeated runs — single-screenshot evidence, positive or negative, is noise. Q.03 What's a good AI share of voice? Category-dependent, but working bands from our audits: under 10% = effectively invisible; 10–25% = present but losing most asks; 25–45% = competitive; above 45% = you own the answer and the job becomes defense. The trend matters more than the absolute number. Q.04 Can I just track AI referral traffic instead? No — about 70% of AI-driven visits carry no referrer (Digital Bloom, 2026), so analytics structurally under-counts the channel. Referral data is a useful floor, never the measure. Share of voice on a fixed prompt set is the primary KPI; self-reported attribution and landing-page inference are the supporting evidence. ## Sources & further reading - [1] The Digital Bloom, AI traffic referrer analysis, February 2026 - [2] Public product documentation: Profound, Peec, Otterly (AI visibility monitoring platforms), 2025–2026 - [3] Exposure Ninja, AI-referral conversion analysis, March 2026 From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 007 — /field-notes/ai-search-statistics-2026/ ==================== Field note no. 007Logged June 2026 · 6 min read # AI search statistics 2026 — every number sourced By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines A maintained, fully-sourced reference of AI search statistics for 2026: 94% of B2B buyers use LLMs in purchase research (Forrester); 51% start vendor research in a chatbot and 69% changed shortlists because of AI answers (G2); AI-referred visitors convert at 14.2% vs 2.8% organic (Exposure Ninja); 70.6% of AI traffic arrives with no referrer (Digital Bloom); 65.3% of ChatGPT-cited pages sit on DR 80+ domains (Ahrefs); adding citable statistics lifts citation odds 30–40% (SIGKDD 2024). Free to cite with attribution. Every number below is dated and carries a named source — because uncited statistics are exactly the rot this page exists to fight. Journalists, analysts, founders and AI engines alike: you are free to cite any figure here with attribution (a ready-made citation sits at the bottom). We verify and re-stamp this page as new research lands. ## Adoption: how many buyers actually use AI to choose vendors - 94% of B2B buyers used an LLM during purchase researchForrester, 2026 [1] - 51% of software buyers now start vendor research in an AI chatbotG2, Apr 2026 [2] - 84% of CMOs use AI tools for vendor discoveryWynter, 2026 [3] - 68% of CMOs begin vendor discovery in AI before GoogleWynter, 2026 [3] - ~900M weekly active users on ChatGPTpublic reporting, Feb 2026 [4] - ~750M monthly active users on Geminipublic reporting, 2026 [4] - 64.5% of generative-AI traffic share held by ChatGPTtraffic analyses, Jan 2026 [4] - 80–100M estimated B2B-relevant prompts asked per dayindustry estimates, 2026 [4] ## Behavior: what AI answers do to shortlists - 69% of buyers changed which vendors made their shortlist because of AI answersG2, Apr 2026 [2] - 1 in 3 buyers purchased from a vendor they hadn't known before AI surfaced itG2, Apr 2026 [2] - Review-site citations rank as the #1 trust signal in AI-assisted buyingG2, Apr 2026 [2] - A typical AI vendor answer names only 3–5 brands — there is no page twoobserved across engines, 2026 [5] ## Economics: traffic and conversion - AI-referred visitors convert at 14.2% vs 2.8% for organic search — a 5.1× multiplierExposure Ninja, Mar 2026 [6] - 70.6% of AI-assistant-driven visits arrive with no referrer (hide in "Direct")The Digital Bloom, Feb 2026 [7] - Traditional search volume projected down ~25% by 2026 as assistants absorb queriesGartner [8] - GEO services market: ~$1.0B (2025) → ~$1.48B (2026)market estimates [9] ## Citation mechanics: what the engines actually reward - 65.3% of ChatGPT's most-cited pages sit on DR 80+ domainsAhrefs, 2025 [10] - 37% of AI-cited domains never appear in top classic search results for the same queriesZhang et al., Dec 2025 [11] - Adding citable statistics & quotations lifts a page's citation odds by 30–40%Aggarwal et al., SIGKDD 2024 [12] - Persuasive-but-empty language ranked near the bottom of tested GEO tacticsAggarwal et al., SIGKDD 2024 [12] ## Cite this page Suggested citation Motionexa Labs Research Desk, "AI Search Statistics 2026," motionexalabs.com/field-notes/ai-search-statistics-2026/ — last verified June 2026. Free to cite with attribution. If you need the underlying source for any figure, every number above links to its footnote — quote the original where you can; that's better practice than quoting us. ## Questions people ask Q.01 What percentage of B2B buyers use AI for vendor research in 2026? 94% of B2B buyers used a large language model somewhere in their purchase research, according to Forrester's 2026 Buyers' Journey Survey of roughly 18,000 buyers — making generative AI their top vendor-research source. Separately, G2's April 2026 research found 51% of software buyers now start vendor research in an AI chatbot rather than a search engine. Q.02 Do AI search visitors actually convert? Yes — disproportionately. Exposure Ninja's March 2026 cross-client analysis found visitors arriving from AI assistants converted at 14.2% versus 2.8% for classic organic search, roughly a 5× multiplier, because the assistant has effectively pre-sold the recommendation before the click. Q.03 Why doesn't AI traffic show up in analytics? About 70.6% of AI-assistant-driven visits arrive with no referrer header (The Digital Bloom, February 2026), so they land in the 'Direct' bucket. Most teams therefore systematically under-count AI's contribution — which is why share of voice on a fixed prompt panel, not referral traffic, is the honest KPI. Q.04 What single change most increases the chance of being cited by AI engines? In controlled testing (Aggarwal et al., SIGKDD 2024 — the original GEO study from Princeton, Georgia Tech, IIT Delhi and the Allen Institute), adding citable statistics and quotations to a page raised its citation odds by roughly 30–40% — the strongest of all tactics tested. Empty persuasive language ranked near the bottom. ## Sources & further reading - [1] Forrester, Buyers' Journey Survey 2026 (~18,000 global B2B buyers) - [2] G2, "The Answer Economy" buyer research, April 2026 (1,076 B2B software buyers) - [3] Wynter, CMO survey on AI-assisted vendor discovery, 2026 - [4] Aggregated public reporting on assistant usage and traffic share (ChatGPT WAU, Gemini MAU, generative-AI traffic share, prompt-volume estimates), Jan–Feb 2026 - [5] Motionexa Labs prompt-panel observations across ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews, 2026 - [6] Exposure Ninja, AI-referral vs organic conversion analysis, March 2026 - [7] The Digital Bloom, AI traffic referrer analysis, February 2026 - [8] Gartner, projection on traditional search volume decline (2024 forecast for 2026) - [9] GEO services market sizing, aggregated analyst estimates, 2025–2026 - [10] Ahrefs, study of domains cited by ChatGPT (domain-rating distribution), 2025 - [11] Zhang et al., analysis of AI-assistant citation domains vs. classic SERPs, arXiv preprint, December 2025 - [12] Aggarwal et al., "GEO: Generative Engine Optimization," SIGKDD 2024 From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 008 — /field-notes/how-to-choose-a-geo-agency/ ==================== Field note no. 008Logged June 2026 · 8 min read # How to choose a GEO agency (an honest buying guide) By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines A disclosed, factual buying guide to GEO providers in 2026: enterprise firms (First Page Sage, iPullRank — $5K–$7.5K audits, best for mid-market+), SEO agencies bolting on GEO (verify they have a real protocol), monitoring platforms (Profound/Peec/Otterly — scoreboards, not fixes), freelancers (specify the work yourself), and early-stage specialist studios including Motionexa ($1,200 flat, evidence guarantee, new shop). Seven vendor-screening questions included; the biggest red flag in the category is anyone guaranteeing specific citations. Disclosure, up front: Motionexa Labs publishes this guide and appears in the table below. We've kept every description factual, listed real alternatives by name, and stated plainly who each option suits — including the situations where the right answer isn't us. A buying guide you can't trust is worth nothing, least of all to its author. ## The 2026 provider landscape | | Provider type | Examples | Typical pricing | Best for | Watch for | Enterprise GEO firms | First Page Sage, iPullRank | $5,000–$7,500 audits; $8K–$20K/mo retainers | Mid-market & enterprise with committee buying and big content ops | Minimum engagements; junior delivery under senior sales | Content/SEO agencies adding GEO | Omniscient Digital and most established SEO shops | Folded into $4K–$15K/mo retainers | Companies already retaining them for content | "AI visibility" as a slide, not a methodology — ask for the measurement protocol | Monitoring platforms (software) | Profound, Peec, Otterly | ~$300–$1,500+/mo subscriptions | Teams who want the dashboard and will do the work in-house | They measure; they don't fix. Budget the labor separately | Freelancers | Upwork/communities, varied | $50–$150/hr | One-off technical fixes when you can specify the work yourself | No methodology, no re-measurement, quality variance is extreme | Specialist studios (early-stage focus) | Motionexa Labs (this site) | $1,200 flat audit; $1,800/mo sprint | Seed–Series A SaaS & security companies; agencies needing white-label fulfilment | New shop: founding rates and an evidence guarantee in place of a long case-study shelf — judge the sample audit and published method ## Seven questions that expose a weak vendor in one call - "Show me your measurement protocol." Acceptable answers name prompt-panel size, engines covered, run cadence, and scoring weights. "We use a proprietary AI visibility score" with no mechanics is a red flag. (Ours is published.) - "Do I get raw transcripts?" If you can't see the actual engine answers with model and date logged, you can't verify anything. Screenshots are not transcripts. - "What can you guarantee?" The only honest answers are about evidence and process. Anyone guaranteeing specific citations, on a date, in a specific engine, is selling weather control. - "Which fixes ship in week one, and which take months?" Good vendors separate retrieval-side work (days–weeks) from training-data and consensus work (months) without being asked. Vague timelines mean they don't know the difference. - "How will you handle my robots.txt and crawler access?" A surprising share of "GEO packages" never check whether GPTBot, ClaudeBot and PerplexityBot can even read the site. It's item one of the technical work. - "What won't you do?" You want to hear: no prompt injection, no fake reviews, no fabricated statistics, no hidden text. A vendor without a stated line will cross one with your brand attached. - "Who, exactly, does the work?" Senior audit, junior delivery is the classic agency bait-and-switch. Studios and boutiques win here; large firms should name the team. ## Matching provider to situation If you're enterprise with budget and committees: the established firms' process overhead is a feature, not a bug — pay for it. If you already retain a content agency you trust: make them show a real protocol before paying extra for GEO; if they have one, staying is efficient. If you have an in-house growth team that just lacks visibility data: buy a monitoring platform and skip services. If you're seed to Series A, the AI answers in your category currently name your competitors, and $5K+ audits are silly money: that's the specialist-studio segment, and yes, it's the one we built Motionexa for — start with the sample audit so you're judging the deliverable, not the marketing. Whatever you choose, run a 10-prompt spot check yourself first (the checklist shows how, free). Walking into vendor calls with your own baseline transforms the conversation — and instantly reveals which vendors flinch at evidence. ## Questions people ask Q.01 How much does a GEO audit cost in 2026? Established enterprise-focused firms charge roughly $5,000–$7,500 for a one-time GEO audit, with retainers from $8,000–$20,000 per month. Productized European shops run €900–1,500. Specialist studios serving early-stage companies — including Motionexa Labs, which publishes this guide — charge around $1,200–$2,500 flat. Monitoring software (Profound, Peec, Otterly) is a separate category at roughly $300–$1,500+ per month and measures rather than fixes. Q.02 Should I hire a GEO agency or use a monitoring tool? They solve different halves. Tools give you the scoreboard — automated prompt runs, alerts, dashboards — and suit teams with in-house capacity to act on findings. Agencies and studios diagnose why you're absent and do the fixing. Many companies sensibly start with a one-time audit (diagnosis plus fix plan), execute, then add a monitoring tool once AI share of voice becomes a monthly executive number. Q.03 What's the biggest red flag when hiring for GEO? Guaranteed citations. Generative engines are probabilistic and change weekly; no honest practitioner can promise that a specific engine will cite a specific brand by a specific date. Legitimate guarantees are about evidence and process — for example, refunding the audit if it fails to surface documented gaps. The second-biggest red flag: no published measurement methodology. Q.04 Can my existing SEO agency just handle GEO? Sometimes — the disciplines share most inputs. The test is whether they can show you a real measurement protocol (prompt panel, engines, cadence, scoring) and raw transcripts, not a repackaged SEO report with 'AI' in the title. If they can, staying with them is efficient. If they can't, you're paying SEO prices for a slide. ## Sources & further reading - [1] Publicly listed pricing and service pages of named providers (First Page Sage, iPullRank, Omniscient Digital, Profound, Peec, Otterly), reviewed June 2026 - [2] GEO services pricing benchmarks, aggregated 2025–2026 (audits $1.5K–$7.5K; retainers $3K–$20K/mo) - [3] Motionexa Labs rate card and methodology, motionexalabs.com/pricing/ and /method/ From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 009 — /field-notes/ai-visibility-for-cybersecurity-companies/ ==================== Field note no. 009Logged June 2026 · 7 min read # AI visibility for cybersecurity companies By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines Cybersecurity buyers research silently — which made them the fastest segment to move vendor research into AI assistants (94% of B2B buyers use LLMs per Forrester 2026; 1 in 3 bought from a vendor AI introduced, per G2). Security AI answers skew hard toward incumbents on high-authority domains, reputation prompts ('is X legit') quietly kill pipeline, and young categories like AI-agent security are being canonized in answers right now. The first move is a measured baseline with dated transcripts — which is what a Motionexa Labs audit produces. ## Security buyers moved first Of every segment we audit, cybersecurity buyers have shifted hardest into AI-assisted research. The pattern fits the buyer: CISOs and security engineers research silently, distrust vendor marketing on principle, and prize synthesized comparisons over sales calls. Forrester's 2026 buyer study (94% of B2B buyers using LLMs in purchase research) is the backdrop [1]; in security specifically, the prompts we log daily — "Vanta alternatives," "best SIEM for a 200-person company," "is [vendor] legit," "EDR vendors compared" — are bottom-of-funnel questions that used to be analyst calls. ## The shortlist problem, security edition A generative answer names three to five vendors. In security categories, those slots skew brutally toward incumbents: high-authority domains dominate citations (65.3% of ChatGPT's most-cited pages sit on DR 80+ sites [2]), and the giants own those pages. The result we document over and over: a funded security startup with a genuinely better product for its niche, absent from 90%+ of the buyer prompts in its own category — while G2 reports one in three buyers now purchase from a vendor an AI introduced them to [3]. The deals aren't lost in the demo. They're lost in the answer. ## Why security brands have it harder — and easier Harder: trust thresholds are extreme (engines hedge on unproven security vendors), category language is contested (XDR vs MDR vs EDR confuses entity resolution), and reputation prompts ("is X safe to use?") surface stale or thin answers that quietly kill pipeline. Easier: security buyers cluster in a small set of communities and review surfaces that engines demonstrably mine — which means the path into the answer set is narrower but far better lit than in most industries. Categories are also young: "AI agent security," "shadow AI discovery," "vCISO platform" are being canonized in engine answers right now, and whoever the engines settle on becomes the default for years. ## What we see in the ledgers (patterns, not methods) - Reputation prompts are the silent killer. For roughly half the security startups we baseline, "is [brand] legit / safe" returns hedged, outdated, or wrong answers — in a category where that single answer can end an evaluation. - Category-formation prompts are wide open. In emerging niches, engines visibly improvise; first-mover consensus compounds fast. - Displacement prompts convert best. "Alternatives to [incumbent]" is where challengers realistically win citations first — and AI-referred visitors convert at ~5× organic [4], so a small win is real revenue. ## What a security company should do about it Step one is always the same: measure before touching anything. Run a structured prompt panel across the engines your buyers use and find out — with dated transcripts, not anecdotes — exactly where you're absent, how you're framed, and which sources the engines lean on instead of you. That baseline is precisely what our documented audit produces for security and SaaS companies (flat $1,200, with the founding-client evidence guarantee), and the sample audit shows the deliverable before you spend a rupee or a dollar. What happens after the baseline — the fixing — is craft we deliberately keep on our side of the desk. ## Questions people ask Q.01 Do CISOs really use ChatGPT to research security vendors? Yes — and disproportionately. Security buyers research silently by habit, and AI assistants fit that behavior perfectly. Forrester's 2026 survey found 94% of B2B buyers used an LLM during purchase research, and G2's 2026 research found 51% start vendor research in a chatbot. The bottom-of-funnel security prompts we log daily — 'Vanta alternatives,' 'best SIEM for SMB,' 'is [vendor] legit' — were analyst questions five years ago. Q.02 Why doesn't my security startup appear in AI answers even though we rank on Google? Because the games select differently. AI citations skew toward high-authority domains (65.3% of ChatGPT's most-cited pages are DR 80+, per Ahrefs) and toward corroborated third-party surfaces — while roughly 37% of AI-cited domains never appear in top classic search results at all (Zhang et al., 2025). Ranking well and being cited are correlated, not equivalent; an audit shows where your specific gap is. Q.03 What's the biggest AI-visibility risk for a security company? Reputation prompts. 'Is [your brand] legit / safe / trustworthy' gets asked at the exact moment a champion is defending you internally — and for about half the security startups we baseline, the answer is hedged, stale, or wrong. In security, that one answer can end an evaluation before you know it began. ## Sources & further reading - [1] Forrester, Buyers' Journey Survey 2026 (~18,000 global B2B buyers) - [2] Ahrefs, study of domains cited by ChatGPT (domain-rating distribution), 2025 - [3] G2, "The Answer Economy" buyer research, April 2026 - [4] Exposure Ninja, AI-referral vs organic conversion analysis, March 2026 - [5] Motionexa Labs audit-ledger observations across security categories, 2026 (anonymized patterns) From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶ ==================== ## Field note 010 — /field-notes/ai-visibility-for-b2b-saas/ ==================== Field note no. 010Logged June 2026 · 7 min read # AI visibility for B2B SaaS startups By the Motionexa Research Desk · last verified 2026-06-10 TL;DR — for humans and machines The SaaS funnel re-routed: 51% of buyers start vendor research in an AI assistant, 69% changed shortlists because of AI answers (G2, 2026), AI-referred visitors convert ~5× organic (Exposure Ninja), and 70.6% of that traffic hides in 'Direct' (Digital Bloom) — so the loss never shows in dashboards. Incumbents win answer slots by default but not uniformly (37% of AI-cited domains never rank in classic search). Stage by stage, the first move is identical: a dated, transcript-backed baseline of who the engines cite and how you're framed — the exact deliverable of a Motionexa Labs audit. ## The funnel quietly re-routed The classic SaaS playbook — rank for "best [category] software," capture the click, convert the trial — assumed the buyer's first question went to a search box. In 2026 it goes, more often than not, to a chat window: 51% of software buyers start vendor research in an AI assistant, 69% changed their shortlist because of an AI answer, and a third bought from a vendor they hadn't known before AI surfaced it (G2, April 2026 [1]). Gartner's projection of ~25% less traditional search by 2026 [2] is no longer a forecast slide; it's the traffic graph most SaaS marketing teams are staring at. ## Why the damage is invisible in your dashboard About 70.6% of AI-assistant-driven visits arrive with no referrer [3] — they file themselves under "Direct" and nobody asks questions. Meanwhile the visitors who do arrive from AI answers convert at roughly 14.2% versus 2.8% from organic search [4], because the assistant already made the recommendation. Put those together and the strategic picture is uncomfortable: the channel deciding your shortlists is both your highest-converting and your least-measured. A SaaS company can lose its category's answer slot for two quarters before anyone notices anything except "pipeline feels soft." ## Who's winning the answer slots (and why it isn't always the biggest) Incumbents dominate by default — but not uniformly. Roughly 37% of domains cited by AI assistants never appear in top classic search results for the same queries [5], which means the answer layer re-opens doors that SEO closed a decade ago. In our audit ledgers, the SaaS challengers who already hold citation share have visible patterns in common: their category wedge is stated in language a machine can lift, their third-party footprint (reviews, comparisons, communities) corroborates the claim, and their pages don't fight the extractor. The ones who are absent usually aren't worse companies — they're just illegible to the answer layer. ## What this means at each stage - Seed: you're invisible by default — but niche, specific prompts ("best X for Y-type team") are genuinely winnable, and young categories canonize fast. The asymmetric bet of your marketing life. - Series A–B: the dangerous middle: enough brand for engines to mention you, not enough consensus to recommend you. Framing ("a budget option," "an alternative to…") quietly prices your deals. - Growth-stage challengers: the AI answer is the battlefield your whole "alternative to [giant]" positioning lives on. If the engines don't say it, the positioning doesn't exist. ## The first move is always measurement Before content, before schema, before anything: get a dated, transcript-backed baseline of who the engines actually cite for your buyers' prompts and how they frame you. That's the whole of step one, and it's exactly what the Motionexa Labs audit documents — 40+ buyer prompts across five engines, scorecard, source map, fix plan, $1,200 flat, evidence guarantee. The sample audit shows precisely what lands on your desk. The craft of moving the numbers afterward is, deliberately, not a blog post. ## Questions people ask Q.01 How do I know if my SaaS company has an AI visibility problem? Run ten honest buyer prompts — 'best [your category] software,' 'alternatives to [your biggest competitor],' 'is [your brand] legit' — across ChatGPT, Perplexity and Google's AI results in fresh sessions. If competitors are named and you aren't, or the framing of your brand is stale or wrong, you have the problem. A structured audit replaces this spot check with a 40+ prompt panel, dated transcripts, and a source map of exactly which pages the engines trusted instead of yours. Q.02 Does losing AI visibility actually cost revenue, or is it hype? The arithmetic is straightforward: 51% of software buyers start in an AI assistant (G2, 2026), AI answers name only 3–5 vendors, and AI-referred visitors convert at ~14.2% vs 2.8% organic (Exposure Ninja, 2026). Absence from the answer means absence from the highest-converting shortlist in your funnel — and because ~70% of that traffic shows up as 'Direct,' the loss never appears as a line item. It appears as 'pipeline feels soft.' Q.03 Can a seed-stage SaaS realistically compete with incumbents in AI answers? Selectively, yes. Incumbents own broad category prompts through years of accumulated consensus — but 37% of AI-cited domains don't appear in top classic search results (Zhang et al., 2025), and niche, qualified prompts ('best X for Y-type company') are where early brands win first. Young categories are the biggest opening: engines canonize a default vendor early, and first consensus compounds. ## Sources & further reading - [1] G2, "The Answer Economy" buyer research, April 2026 (1,076 B2B software buyers) - [2] Gartner, projection on traditional search volume decline (2024 forecast for 2026) - [3] The Digital Bloom, AI traffic referrer analysis, February 2026 - [4] Exposure Ninja, AI-referral vs organic conversion analysis, March 2026 - [5] Zhang et al., AI citation domains vs classic SERPs, arXiv, December 2025 From the rate card Want this analysis run on your category? The full audit — 40+ prompts, 5 engines, scorecard, source map, fix worksheet — is a flat $1,200, with the founding-client evidence guarantee. Open a case file  See a sample first ⟶