The short answer
Most answer engine optimization examples published online cannot be checked: screenshots with no date, no stated prompt, no session details, numbers with no named measurement tool. This page walks through six real AEO case studies from our client work, and for each one states the before, the after, and the thing that separates a case study from a claim: how you would verify it. The methods are engine-hosted share links on our proof wall, the client's own GA4, the client's own Search Console, and in one case a robots.txt file anyone can open today. We also cover a result that did not hold up, because AI answers shift and pretending otherwise is how this industry earns its reputation.
Every number below is stated exactly as it appears on our case studies, with the same measurement attribution.
Why most AEO case studies cannot be checked
AI answers are volatile. The same prompt can return different brands run to run, engine to engine, and week to week, which our study of 90 days of AI citations measured directly. That volatility makes AEO results easy to fake: capture the one lucky run, crop the screenshot, skip the date, and you have a case study.
So we hold our own examples to four rules:
- Engine-hosted links beat screenshots. A public share link shows the original answer on infrastructure we do not control. Our proof entries state the exact prompt, capture date, session type, and region.
- Name the measurement tool. GA4 last-click, Search Console with brand terms excluded, Profound across a stated prompt set: the method is part of the claim.
- Publish the caveats with the numbers. If a positions win came with falling impressions, we say so.
- Date historical wins. When a result no longer reproduces, it is stated in the past tense with its year.
Even our SEO-side case studies follow the standard: the Ahrefs chart on the InCorp case study (non-branded organic traffic from 2K to 22K in a year) is published uncropped with the brand filter visibly unticked, and the write-up says plainly it is a Google organic win, not an AI one.
OpenSea: citation share from 5% to 27% in 7 months
The situation. OpenSea rebuilt its entire site as an application. The product experience improved and the crawlability collapsed: pages were rendered client side, so from a crawler's point of view the category leader had almost nothing readable. Across tracked NFT and Web3 prompts, OpenSea was cited in roughly 5 percent of relevant answers.
What was done. We rebuilt crawlability so pages could be read and quoted again, baselined a tracked set of 100+ prompts across NFT, Web3, and crypto with the exact source URLs behind each answer, fixed the pages that answer marketplace research questions, and ran citation work on the third-party sources engines lean on for the category, including getting OpenSea named in a Forbes article.
What resulted. Citation share grew from 5% to 27%, a 5x multiplier, in 7 months of AEO work.
How it is verified. Two ways, and the case study is explicit about the difference. The citation share was measured in Profound across the tracked set of 100+ prompts, comparing how often OpenSea appeared in AI answers at the start of the engagement against the end; that access has since lapsed, so the case study names the tool and the method rather than asking you to take the number on faith. The second half is independently checkable: a Google-hosted Gemini share link ranking OpenSea first of six NFT marketplaces for "best NFT marketplace" (captured 2026-08-09, logged-in session, United States): "Best Overall & Best for Beginners: OpenSea. OpenSea remains the most accessible, feature-packed, multi-chain platform." Open the link and run the prompt yourself.
Try Create: $2M to $25M, and the answer that outlived the engagement
The situation. Try Create had a product people wanted and a site that could not be found: broken content, and product pages not built around the terms buyers actually type.
What was done. A sprint plus retainer over 6 months: product pages restructured so each had a single clear H1 and target term, broken content repaired, technical fixes sitewide, content built for buyer and comparison intent, and Reddit seeding so the community threads the engines pull from had something to find.
What resulted. Organic revenue scaled from $2M to $25M, a 12.5x multiplier, with the headline result achieved in 2025. And the part that makes it an AEO example rather than an SEO one: Create is still ChatGPT's #1 pick for "what is the best creatine gummy" a year after the engagement ended.
How it is verified. The ChatGPT claim carries a public share link, captured 2026-08-09 in a logged-out United States session, the strongest capture condition because no chat memory or personalisation shapes the answer: "If you want a creatine gummy specifically, my top pick right now is Create Creatine Gummies." The captured answer cites a Fortune review of the product, a placement earned during the engagement: off-page placement to AI citation, end to end, both halves checkable. Gemini corroborates, naming Create in its recommendation for "best creatine gummies" (captured 2026-08-08) via its own share link. The revenue figure is stated as the client states it, which is why the case study leads with the proof a reader can open.
Enter.Health: a top content page up 28x, measured in their own Search Console
The situation. Enter.Health sells revenue cycle management into healthcare, a category where the money terms are few, expensive, and well defended. There is no long tail to hide in. Incumbents held the head terms, and the site was not built to win the featured snippets and AI Overviews that now sit above the classic results.
What was done. An 8 month build aimed squarely at the high-value money terms: revenue cycle management, RCM software, and the AI variants of both, with a content programme around those terms, placements across healthcare publications, and pages structured for snippet and AI Overview capture.
What resulted. Content pages grew 7x to 28x in impressions, and average position across the whole domain roughly halved, from 31 to 15.
How it is verified. Measured in Enter.Health's own Search Console, comparing April 2025 against January 2026. The page growth is impressions on /what-is-ipa-in-medical-billing (739 to 20,984, the 28x) and /medical-billing-automation (1,274 to 10,955, an 8.6x). The average position figure is across every query the domain ranks for, not a hand-picked set. An earlier headline that Search Console could not corroborate was swapped out for the page-level growth it can, which is the standard every number on the page now meets.
iSleep: $12.8k in revenue you can trace to AI referrals in GA4
The situation. iSleep was starting from close to nothing: barely any content, product pages that had never been optimised, no technical foundation, in a category dominated by legacy brands and venture-backed competitors.
What was done. Eleven months, built in order: technical foundation rebuilt, a content programme built from scratch around the questions buyers actually ask AI, FAQ coverage in the direct-answer format engines can lift cleanly, and product pages optimised for buyer prompts.
What resulted. $12.8k in revenue attributed to AI referrals across 71 orders, over 11 months to July 2026, with 1,983 sessions from AI referrals behind it.
How it is verified. Every tile on the case study is either in the client's GA4 or was delivered work. Nothing estimated. An attribution model that undercounts AI influence produces a smaller, harder number, and that is the trade we make on purpose.
Lux Unfiltered: Gemini's best dye-free pick, and a positions story told honestly
The situation. Lux Unfiltered was scaling fast on paid and influencer while the organic and AI layer sat broken underneath: product pages built wrong, missing alt tags, no FAQ coverage, and pages targeting terms their buyers were not searching. Beauty is an editorial category: engines answer out of publications, roundups, and review surfaces rather than brand sites.
What was done. Thirteen months and counting: a full site revamp first, then a sustained push on topical authority and the third-party surfaces beauty answers are built from, including listicle placement work and a Stacker feature syndicated onto regional news domains, at least nine of which carried it, each an indexed page naming the hero product.
What resulted. Gemini names Lux Unfiltered's Sunstruck Self Tanning Face Drops as its best dye-free pick for the prompt "best tanning face drops." In Search Console, average position across non-branded queries moved from 25.8 to 9.7, and the brand now sits at 3.2 for "non toxic self tanner."
How it is verified. The Gemini result is a public share link (captured 2026-08-08, logged-in session, United States): "The Lux Unfiltered Sunstruck Self Tanning Face Drops feature a clear, colorless formula that won't stain clothing or clog pores while providing a golden finish." The positions data is measured in Lux Unfiltered's Search Console with brand terms excluded, comparing April to July 2025 against April to August 2026, and the caveat is published with it: non-branded impressions fell over the same period, so this is a change in position rather than in volume.
Cortico Health: from blocked by every major AI crawler to open in days
The situation. Cortico Health had a clean product and good content that no AI engine could read. GPTBot, PerplexityBot, ChatGPT-User, and Google-Extended were all being turned away at the robots.txt and CDN level. Nobody knew; the block only surfaced when an AI visibility audit came back with zero citations across every tracked prompt. A zero that complete is not a content problem, it is a door problem. If your brand is absent from AI answers, rule this out first, as we cover in why isn't my brand in AI search.
What was done. A tightly scoped sprint: crawler audit across the four blocked engines, robots.txt and CDN rule changes to unblock them, schema and entity reinforcement on flagship pages, and day-by-day citation tracking once crawlers regained access.
What resulted. Cortico went from blocked by every major AI crawler to fully open to them in days, with all 4 audited engines unblocked and a 1,000+ day-one traffic lift.
How it is verified. Permanently checkable by anyone, today: open cortico.health/robots.txt and the AI crawlers are named and allowed. The case study claims the access fix, which is verifiable, rather than ongoing AI visibility, which is not. Access is necessary but not sufficient, and it says so.
What did not hold up, and why we say so
AI answers shift, and an AEO agency that only ever shows wins in the present tense is telling you something about its verification standards.
Upsun. A 6-month sprint aimed at getting a strong technical product into the seed sources language models cite for DevOps platform questions: citation gap analysis across ChatGPT, Perplexity, and Gemini, mention engineering across high-authority developer publications, and entity disambiguation. At the close of the 2025 engagement, ChatGPT ranked Upsun #1 for "best overall structured DevOps platform," with 12 cited prompts gained and Perplexity coverage moving from 0 to top 3. That #1 does not reproduce in a clean session today, so the case study states it as a 2025 result, dated and in the past tense. The result was real; the claim is scoped to when it was true.
The same discipline, applied elsewhere. Lux Unfiltered once had a ChatGPT #1 claim with a screenshot behind it; when a clean logged-out session no longer reproduced it, the screenshot came off the site and the case study now leads with the Gemini result that holds. iSleep's revenue was once estimated far higher; checking the client's GA4 produced $12.8k, and $12.8k is what the site says. Each published number went down or got narrower, and each case study got stronger, because every remaining claim is one a reader can check.
If you want to run this kind of check on your own brand before believing any agency's deck, our walkthrough on how to check if ChatGPT recommends your brand is the manual version.
Every example at a glance
| Campaign | Headline result | How it is verified |
|---|---|---|
| OpenSea | Citation share 5% → 27% in 7 months | Profound, 100+ tracked prompts; Gemini share link ranks OpenSea #1 of six marketplaces |
| Try Create | $2M → $25M organic revenue (2025); still ChatGPT's #1 for its primary prompt | Logged-out ChatGPT share link, 2026-08-09, citing the Fortune review earned during the engagement |
| Enter.Health | Top content page 28x in impressions; average position 31 → 15 | Client's own Search Console, April 2025 vs January 2026 |
| iSleep | $12.8k AI-attributed revenue, 71 orders | Client's own GA4, last-click attribution |
| Lux Unfiltered | Gemini's best dye-free pick; non-branded position 25.8 → 9.7 | Gemini share link, 2026-08-08; Search Console, brand terms excluded |
| Cortico Health | Blocked by every major AI crawler to fully open in days | Live today at cortico.health/robots.txt |
| Upsun | ChatGPT #1 for its category prompt | Dated as a 2025 result because it no longer reproduces |
Key takeaways
- Real AEO results come in different currencies: citation share (OpenSea, 5% to 27%), engine answers that name you (Try Create, Lux Unfiltered), revenue traced through GA4 (iSleep, $12.8k across 71 orders), Search Console movement (Enter.Health, 28x on its top page), and access fixes anyone can check live (Cortico Health).
- The verification method is part of the result. Engine-hosted share links beat screenshots, named tools beat bare numbers, and logged-out captures with a stated prompt, date, and region beat everything undated.
- Conservative measurement produces smaller, harder numbers. GA4 last-click undercounts AI influence; publishing it anyway is what makes the number trustworthy.
- AI answers shift. A result that was true in 2025 may not reproduce in 2026, which is why we date historical wins like Upsun's and pull proof that no longer reproduces.
- You cannot win AI search from your own site alone: our citation sources study found a brand's own website is only 2 to 6 percent of what engines cite, which is why every example above includes third-party citation work.
What these examples have in common
Every campaign above runs the same sequence: make the site readable to engines, build content around the questions buyers actually ask, earn the third-party citations that answers are assembled from, and measure it all against a tracked prompt set. That sequence is answer engine optimization, and it is what our AEO service delivers: named tools, stated windows, proof a client can open.
If you want to know what the before-and-after could look like for your brand, start with your baseline. Book a strategy call and we will show you where you stand in AI answers today, held to the same verification standard as our own case studies.