Does AEO Actually Work? Results Real Clients Have Seen

We checked the case-study figures against the published case studies on 14 October 2026; how we put this question to AI engines is near the end.

Does AEO actually work?

Yes. Brands that publish clear answers to the questions their buyers put to AI engines can move from absent to cited, and the client results below show what that looked like.

OnlyAEO is an AEO and GEO agency, so both case studies in this article are our own work. Read them as vendor-reported. That is why we give the dates, the scope of each audit and what each one leaves out; both are also on our case studies page.

For an outside view, the research section further down covers an independent field test and what Google says about its own AI features.

What results did FastTrackr AI get?

FastTrackr AI, a platform that automates advisor transitions in wealth management, went from barely registering in AI answers to the top tier of its category. Its own website also became a source the engines cite, which was the harder win.

The full FastTrackr AI case study has every table; the figures that answer this question follow.

Where it started

Our baseline audit is dated February 11, 2026. It covered 11 model families and 100 buyer prompts spread over 10 distinct buyer personas.

Only 1 of the 11 models and 1 of the 10 personas showed FastTrackr at all. Buyers asking about transitions were sent to incumbents with large content libraries. The rest of the starting point is in the February column of the table below.

What the work was

The program launched in March 2026, with an automated daily pipeline live and the first article clusters shipped. Every article answered a real buyer question taken from live AI-conversation data, so the plan came from what buyers were asking engines rather than from a keyword tool.

• Uncontested topics first. The audit found topic clusters with real buyer urgency and no competitor coverage in AI answers, and those were written first.
• Citations as the target. The aim was for engines to cite fasttrackr.ai as a reference URL, in addition to naming the company.
• A short answer up top. Each article opened with a 40-to-60-word direct answer, followed by data tables and FAQ schema.

What changed by May 2026

We re-measured on May 22, 2026. The table compares the same audit on both dates.

The persona row links the work to the result most directly: according to the case study, that gain followed an article series written for exactly that audience.

We went from being essentially invisible in our category to the #3 most-cited brand inside one quarter.

What results did the New York edibles retailer get?

A state-licensed edibles retailer with roughly a dozen storefronts across New York City went from the edge of its category to near the top in AI answers, with its own website becoming the source the engines cite most.

The client is anonymous, and the New York edibles retailer case study keeps it that way.

Where it started

Our audit covered 11 model families and a structured set of buyer personas and topics. The case study dates the baseline to April 2026, when the retailer was present in its category but rarely named; the figures are in the first column of the table below.

What the work was

The program treated the retailer's local footprint as an advantage no national chain could copy.

• Every question became a target. Each question the audit captured, across every buyer type, became a piece of content.
• Neighborhood-level pages. Location-aware content covered the neighborhoods and surrounding towns each store serves.
• Its own site as the source. As with FastTrackr, the target was citations of the client's website rather than of a directory or review site.
• The same page structure. Pages used answer capsules, data tables and FAQ schema.

What changed

Second place among cited sources went to New York State's official government site, at 247 citations, behind the retailer's own site. For a regulated local retailer, being cited above the licensing authority is the result we would show a skeptic first.

What did both programs have in common?

The same habits, applied in the same order. Each is described under its client above; this section adds why it mattered.

• A baseline before any writing. Both programs began with the dated audits above, before any content shipped.
• Questions from buyers, by persona. Each buyer persona gets a different answer from the same engine, so content written for a generic reader can miss what each buyer actually asks.
• The client's own site as the cited source. Both aimed for citations of the client's own pages rather than mentions; the checklist below says why that matters.
• The answer at the top. Engines lift short, self-contained passages, which is why both used the answer-first page structure described above.

How long does AEO take to show results?

The first change you can measure usually arrives no sooner than 60 days in. Engines have to find new pages, read them and start drawing on them, and a quiet first month is normal.

FastTrackr's dates are in its section above, from a February baseline to a May re-measure. Treat that as a single client's record, not a schedule to plan around.

The retailer's case study does not give a clear start date for the work, so we do not quote a duration for it.

Do AEO results hold once you have them?

For FastTrackr they have so far. A follow-up audit dated August 26, 2026 found fasttrackr.ai had climbed to the #2 most-cited source in its category, with 110 citations.

That put FastTrackr's own site ahead of the trade publications investmentnews.com, at 81 citations, and wealthmanagement.com, at 61. In this category, those publications had long been the sources engines leaned on.

We have no published follow-up for the retailer, and a single later audit is not yet a trend. The case for keeping the work going is in why ongoing work keeps compounding.

What does independent research say about AEO?

Independent research finds a real effect, smaller and less certain than vendors' headline numbers. Google's own position is at the end of this section.

The strongest test so far is a natural experiment published on arXiv this year. Its authors applied a bundle of AEO changes in January 2026 to hundreds of thousands of YouTube question-and-answer pages on a single domain, and used the site's untreated pages as a control.

On monthly totals, ChatGPT referrals grew 5.7x, while untreated pages on the same domain grew 3.5x over the same window. After allowing for that background growth, the authors estimate a level increase of 1.82x (95% CI 1.31-2.54).

The effect held on engagement-filtered traffic, at 2.27x. Yet a placebo-in-time test gave p=0.16, and the paper says the effect is "suggestive, not conclusive, given a short and noisy pre-period".

The paper concludes that headline AEO multiples substantially overstate the causal effect. That applies to our case studies too: neither had an untreated control, so they show what happened, not how much of it the work alone caused.

Google's guide to AI features in Search is plain on the point: appearing in AI Overviews or AI Mode takes no extra requirements and no special optimization. To be a supporting link, a page must be indexed and eligible to show in Google Search with a snippet.

Google adds that no new machine-readable files, AI text files, markup or special schema.org structured data are needed. For Google, then, the FAQ schema in our case studies is optional; what counts is a useful page it can index and quote.

What have other brands published about AEO results?

A few brands have published their own numbers, and HubSpot has collected the best known of them. Like ours, these results are self-reported, with no control group.

• Apollo. In its roundup of AEO case studies, HubSpot reports a 63% brand citation rate for AI awareness prompts and 36% for category prompts. Apollo got there mainly through Reddit, without revamping its website content.
• Broworks. HubSpot reports that 10% of Broworks' organic traffic came from LLMs, and that 27% of those AI-referred sessions converted into SQLs.

Apollo's Reddit route is close to the reverse of ours. Broworks is the only example here that reports pipeline, through sales-qualified leads.

How should you judge an AEO case study?

Put the questions in this table to any case study, ours included. The last column shows how our own case studies answer each one.

To put these questions to a vendor in writing, use an RFP template for evaluating an agency. The wider buying criteria are in how to choose an AEO agency.

How does OnlyAEO measure results so you can check them?

Each monthly report tracks citation share, mention rate and how often LLMs recall your brand. Because each answer is stored with its exact prompt (see the checklist above), anyone can paste the prompt into ChatGPT or Claude and check the result.

Our AI visibility platform runs ChatGPT, Claude, Gemini, DeepSeek and Grok, and every prompt runs twice per engine, because the same prompt can draw different answers. The steps behind these numbers are in how to measure AI visibility.

For the revenue question in the checklist above, LLM Attribution shows which AI answers send you pipeline. What a report should contain for a procurement team is set out in what an AEO report should show.

Our team does the AI search work, and your team reviews and approves every piece before it goes live. To size the return before you commit, read forecasting AEO ROI before you spend.

How we tested this

Before every article, we ask ChatGPT and Claude the buyer's question and read the pages they cite. On 14 October 2026 we asked this article's question; both searched the web, and neither cited OnlyAEO.

Claude warned that most published AEO results come from vendors selling AEO, and suggested asking any vendor for its starting baseline. We agree, and the dated baselines above are our answer to it.

We also asked "Which AI company handles RIA advisor transitions in the US?" Claude searched, named FastTrackr AI as one of the dedicated transition platforms and cited fasttrackr.ai. ChatGPT answered without searching, asked a clarifying question and did not name FastTrackr.

A single day's check is a snapshot; answers change from run to run, as the measurement section above notes.

More questions about AEO results

Does OnlyAEO guarantee results?

No. There is no results guarantee, and no promise of any particular ranking, citation or placement in an AI answer.

Will AEO hurt our existing Google traffic?

In the arXiv field study, Google organic clicks to the treated pages did not fall beyond the site-wide trend, and indexation was preserved. We publish answer pages on your own domain with guardrails so they do not hurt your existing search presence.

Do AI-referred visitors convert better?

Many marketers think so: in HubSpot's 2026 State of Marketing report, 58% said visitors referred by AI tools convert at higher rates than traditional organic traffic. That is a survey of opinion rather than a measured conversion study, so check it against your own analytics.

Which AI assistants does OnlyAEO work on?

We work on ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Microsoft Copilot, Grok, DeepSeek, Meta AI (Muse) and Instinct. Our tracking platform runs a smaller set, listed in the measurement section above.