---
title: "By May 2028, Buyers Will Reject These Three AI Visibility Reporting Claims"
description: "Three checkable predictions for AI search reporting by May 2028—and how to buy repeatable measurement and implementation instead of an impressive screenshot."
url: "https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fa53d87c-8159-4e49-85f7-6963236bd6a5.png"
published: "2026-11-10T06:17:12.119Z"
modified: "2026-11-10T21:25:09.768Z"
---

November 10, 2026 · 10 min read

# By May 2028, Buyers Will Reject These Three AI Visibility Reporting Claims

[](https://groas.com/author/alexander-perelman)

[Alexander Perelman](https://groas.com/author/alexander-perelman)Personally Tested

Head Of Product @ groas

[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![A shiny chrome '94%' sculpture seen from behind, revealed as a hollow plywood facade propped up with braces and sandbags, with nothing measured behind it.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fa53d87c-8159-4e49-85f7-6963236bd6a5.png)

In this article

1. [Prediction 1: By May 2028, a visibility percentage without a disclosed denominator fails procurement](#prediction-1-by-may-2028-a-visibility-percentage-without-a-disclosed-denominator-fails-procurement)
2. [Prediction 2: By May 2028, citations and recommendations stop sharing one score](#prediction-2-by-may-2028-citations-and-recommendations-stop-sharing-one-score)
3. [Prediction 3: By May 2028, reporting-only providers must show execution or lose to firms that do both](#prediction-3-by-may-2028-reporting-only-providers-must-show-execution-or-lose-to-firms-that-do-both)
4. [Buying this week: ask for the method, the work and the coverage](#buying-this-week-ask-for-the-method-the-work-and-the-coverage)
5. [Put a May 2028 review in the agreement now](#put-a-may-2028-review-in-the-agreement-now)

**By May 2028, an AI visibility report showing a single percentage without a prompt list will fail a serious buying process.** I expect procurement teams to reject it the way they reject a rank report with no keywords: politely, and without a second meeting.

I am putting dates on this because vague forecasts are worthless. These are three reporting claims I expect buyers to reject within 18 months: a visibility percentage without a denominator, a citation counted as a recommendation, and monitoring sold as management. Each prediction has a mechanism, evidence available today, and a test that could prove me wrong.

The buying principle is simple: purchase repeatable measurement and implementation, not an impressive screenshot.

## Prediction 1: By May 2028, a visibility percentage without a disclosed denominator fails procurement

No engine publishes a visibility score. The number in a vendor report is its own construction over a chosen prompt list, and [visibility is a percentage of responses to that list](https://www.saasinsight.io/guides/what-an-ai-visibility-score-measures). Change the list and you change the score. Add ten prompts you already win and the percentage climbs without anything changing inside ChatGPT, Gemini or Perplexity.

That is the mechanism behind my prediction: buyers will discover that they can buy a better-looking number without getting better visibility.

I used to tell clients a blended score was directionally useful. I was wrong. It is useful only if you can see the denominator.

The evidence starts with the measurement itself. [Differences between prompts are 2.4x larger than day-to-day movement on any single prompt](https://www.sistrix.com/blog/measuring-ai-visibility/), and with 20 prompts accuracy sits around plus or minus 9.5 points. Twenty different prompts give you 3x more accuracy than one prompt repeated for 20 days. Self-selected sets flatter the brand in 3 of 4 cases.

Then add answer instability. Across 2,961 runs, the same question almost never returned the same brand list twice. [If your true appearance rate is 30%, a single check calls you invisible 7 times in 10](https://shruwd.io/blog/ai-visibility). Two single checks disagree 42% of the time with nothing changed.

An impressive screenshot starts looking less impressive once the buyer sees how easily the next run can contradict it.

If I were buying this week, I would ask for the method before discussing the percentage. **The denominator must travel with the score.** Every report needs four things:

- **Prompt cohort:** the exact questions tracked, with buying-intent prompts distinguished from curiosity.
- **Engines and surfaces:** ChatGPT, Gemini, Perplexity and Google AI Overviews reported separately, not blended.
- **Repetitions:** how many times each prompt ran, rather than one result presented as a stable position.
- **Measurement dates:** when the runs happened, so a peak day does not pass for sustained visibility.

That is what makes a vendor number comparable month to month. I learned the same lesson running paid search: a CPA without the match types, budget and date range attached is a story, not a number. AI visibility has the same problem, with more variance.

**Free tool:** the [AI Visibility Checker](https://groas.com/free-tools/ai-visibility-checker) asks ChatGPT, Gemini, Claude, Perplexity and Grok the question your customers ask and shows whether they recommend you.

**What would prove me wrong:** by May 2028, most renewals still go to single-percentage reports with no disclosed prompt list. That would mean buyers continue accepting the ambiguity even after repeatable measurement becomes available.

I expect the opposite. A provider that shows [how to build a repeatable prompt panel across ChatGPT, Gemini and Google AI Overviews](https://groas.com/post/how-to-measure-your-business-s-visibility-in-chatgpt-gemini) gives buyers a standard against which to judge every screenshot.

My procurement line: no denominator, no signature. If the provider calls the prompt list proprietary, I would keep looking.

## Prediction 2: By May 2028, citations and recommendations stop sharing one score

Being used as a source is not the same as making the buyer’s shortlist. A citation means your site appears as a source. A recommendation means the answer names your business as an option worth buying. Providers that blend the two sell footnotes at shortlist prices.

The split is already visible by engine. A June 2026 Semrush study found [61.7% of brand appearances in AI answers were citations where the brand was never named](https://geocitely.com/blog/ai-mentions-vs-citations). ChatGPT cited brands in 87% of appearances but named them in only 20.7%, while Gemini named them in 83.7% and cited them in 21.4%.

Count those together and the score hides the gap a buyer wants to close. Worse, a named mention is not automatically a recommendation either. The answer has to present the business as an option, not merely include its name.

The mechanism is the distinction between [retrieval and recognition](https://www.meetgeo.ai/blog/cited-not-recommended-ai-search-brand-recommendations). An engine can retrieve a useful sentence from a page without presenting the company behind it as a buyable option. Those are different outcomes, and they call for different work.

Consistent category pages, comparable specs, reviews, directory listings and press mentions are the work behind building that recognition. A reporting tool cannot create them. It can count the citations and show where the gap remains.

**For buying-intent prompts, I want recommendation rate separated from citation rate for each engine.** Neither number, by itself, proves revenue.

On a prompt such as “best vendor for a defined job,” I would ask for two separate counts:

- How often the answer names the business as an option worth buying.
- How often the business appears only as a citation beneath an answer recommending someone else.

I would weight the first at least 5x the second when the prompt carries spend behind it, say a $20k/month service category. That is my buying preference, not a revenue formula. It reflects the difference between appearing in the footnotes and appearing on the shortlist.

A provider that refuses to separate them is telling me which half of the score sells the renewal.

**What would prove me wrong:** through May 2028, buyers keep paying full retainers for blended scores and vendors face no pressure to split them. If that happens, I misread what buyers value.

My forecast is the opposite. Once a buyer sees how differently engines cite and name brands, the blended percentage becomes harder to defend. I cover that distinction further in [my 2028 forecast on citations, recommendations and engine-level proof](https://groas.com/post/within-two-years-one-ai-visibility-score-won-t-be-enough-to).

My buying rule: a report that merges footnotes with shortlists goes in the bin.

## Prediction 3: By May 2028, reporting-only providers must show execution or lose to firms that do both

A visibility gap closes when someone does the work: publishes a page, fixes access or earns a third-party mention. Reporting sophistication does not publish anything.

After testing five platforms, one practitioner found [four of five helped map gaps but left execution to the team](https://hackernoon.com/why-most-geo-tools-fail-at-ai-search-visibility-and-what-actually-works). The conclusion was that visibility compounds from output: structured content, publishing consistency, authority and citation velocity.

I have lived that split in paid search. An audit listing 40 missing negatives is worth nothing until someone adds them, checks search terms again and logs what changed. The list is a diagnosis. It is not account management.

That is why ongoing providers already sell implementation rather than monitoring alone. One [ongoing GEO service description](https://trilogi.com/en/geo-agency/) lays out the work: auditing real sector questions, rewriting or creating Q\&A and comparison content, adding schema and feeds, fixing bot access and technical SEO, building media and directory mentions, then rerunning queries each cycle.

The same page states plainly that nobody can guarantee a ChatGPT recommendation. That is the useful boundary: completed work can be documented; a particular answer cannot be promised.

**If a provider guarantees the recommendation, I would walk away. If it cannot show the work behind its service, I would not pay a management fee.**

The mechanism driving this prediction is commercial, not technical. Once a CFO asks what the visibility fee produced besides a PDF, “we told you about the gap” stops sounding like management. Reporting-only providers will have to defend their price against firms that both identify the gap and act on it.

I would keep two separate records:

- **Completed work:** pages published, schema added, bot blocks removed and mentions earned, logged with dates and URLs.
- **Observed outcomes:** changes in citations and recommendations across the disclosed prompt cohort, engines and measurement dates.

The first records what got done. The second records what happened afterward. Keeping them separate prevents a provider from claiming every favorable answer as proof of its work—or using every unfavorable answer as an excuse to do nothing.

A single-day swing proves little about either record. That is why [59% of ChatGPT citations can vanish in a day without anyone doing anything wrong](https://groas.com/post/your-brand-vanished-from-chatgpt-today-that-doesn-t-prove).

**What would prove me wrong:** standalone monitoring holds its price and renewal rate through May 2028 while done-for-you execution stalls. If buyers keep paying the same management fee for a dashboard with no pages shipped, no technical fixes and no mentions earned, I misread the incentive.

My buying rule: buy the work and its action log. Treat the score as a measurement, not the deliverable.

![Bar chart separating named recommendations from citation-only appearances by engine](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/86815c57-64a4-4dd0-a59f-273adb9ac7c9.png)

## Buying this week: ask for the method, the work and the coverage

If I am right, the buying checklist changes now, not in 18 months. **I would test the provider’s evidence before listening to its forecast.** Three requests expose most of the difference between monitoring and management.

### 1. Get a sample report with the method attached

Ask for a real client report showing the full prompt list, separate engines, repetition counts and dates. Client details can be redacted; the measurement method still needs to be visible.

If the vendor sends a screenshot showing 94% visibility and no denominator, that answers the question. I would also ask which prompts carry buying intent. Movement on “best X for Y” is not the same commercial proposition as movement on trivia.

The provider that answers in specifics is the one whose reporting will still make sense when the answers shift next month. **Buy a method that survives a rerun.**

### 2. Get the action log behind the score

Ask who publishes the pages, fixes bot blocks and schema, and earns directory and press mentions. Then ask where each completed item appears, with a date and URL.

A monitoring vendor can show a gap list and a ticket queue. A management firm should show shipped work: a comparison page that went live, an FAQ block added to a service page, a crawl fix that restored access. Those are deliverables, not recommendations waiting for someone else’s team.

At groas, that log is the product, with a named strategist accountable for the work and weekly visibility reporting on top. I would draw the same contractual line everywhere: **no execution record, no management fee.**

![Desk with a printed action log beside an AI visibility report](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/e558f628-fef5-49b3-b743-d8b222ffe439.png)

### 3. Check Google coverage alongside the AI engines

ChatGPT, Gemini and Perplexity retrieve and name businesses differently. Google AI Overviews has its own index and rules. A provider that reports one engine and calls it _AI visibility_ is selling a single checkout lane as the whole store.

I would ask for Google organic movement alongside the prompt cohort: rankings, indexed pages and technical health. An AI mention does not remove the need to check whether the site can be crawled or whether it has a supporting page to cite.

The point is not to collect another dashboard. It is to make sure the execution covers both search surfaces. **One engine is not a complete measurement plan.**

## Put a May 2028 review in the agreement now

I would write the check into the contract: same prompts, same engines, same repetition counts, rerun on fixed dates. Save the denominator and the citation-versus-recommendation split each time. Keep the dated execution log beside them.

In May 2028, pull the file and grade me. If vendors still renew single-percentage reports with no prompt list, blended citation scores still sell at shortlist prices, and dashboards with no shipped work still command management fees, I was wrong on all three.

I do not think I will be. **Buy the method and the work now:** a disclosed cohort, separate scores, and a dated log of pages and fixes. The screenshot will fade. The record of what got published will still be there.

## Frequently Asked Questions

### Why should a single AI visibility percentage in a vendor report be a red flag?

Because no engine publishes a visibility score, so the number is the vendor's own construction over a chosen prompt list. Change the list and you change the score, which means you can buy a better-looking number without getting better visibility. By May 2028, reports without a disclosed denominator are expected to fail procurement.

### What should an AI visibility report include besides the percentage?

Four things: the exact prompt cohort tracked with buying-intent prompts separated from curiosity, ChatGPT, Gemini, Perplexity and Google AI Overviews reported separately rather than blended, the number of repetitions per prompt, and the measurement dates so a single peak day is not mistaken for sustained visibility. Without these, the number is not comparable month to month.

### What is the difference between a citation and a recommendation in AI search reporting?

A citation means your site appears as a source beneath an answer; a recommendation means the answer names your business as an option worth buying. Engines differ sharply: a Semrush study found ChatGPT cited brands in 87% of appearances but named them in only 20.7%, while Gemini named them in 83.7%. Blending the two sells footnotes at shortlist prices.

### Is a monitoring tool alone enough to improve AI search visibility?

No. Reporting sophistication does not publish anything; a gap closes when someone does the work, such as publishing comparison content, adding schema, fixing bot access or earning directory and press mentions. One practitioner testing five platforms found four of five helped map gaps but left execution to the team.

### Should I trust a provider that guarantees a ChatGPT recommendation for my brand?

No. The article's position is that nobody can guarantee a particular AI answer, because completed work can be documented but a specific recommendation cannot be promised. A provider making that guarantee is a reason to walk away, while a provider that cannot show the work behind its service does not deserve a management fee.

### What should I ask an AI visibility provider before signing?

Ask for a real client report showing the full prompt list, separate engines, repetition counts and dates. Ask who publishes the pages and fixes bot blocks, and where each completed item appears with a date and URL. Also ask for Google organic coverage alongside the AI engines, since one engine is not a complete measurement plan.

### How can I hold an AI visibility provider accountable over time?

Write a review date into the contract: the same prompts, engines and repetition counts, rerun on fixed dates, saving the denominator and the citation-versus-recommendation split each time. Keep a dated execution log of pages published and fixes shipped beside the measurements, then grade the results at the review.

## Run Your Marketing On groas

Plug ChatGPT, Claude or any AI agent into groas and let it do the work of a whole marketing team: free to ask, $49 a month flat with every feature included, and autopilot whenever you’re ready.

[Start Free](https://groas.com/pricing)

[Or have groas run it for you](https://groas.com/done-for-you)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is the marketing infrastructure for your own AI: plug Claude, ChatGPT or any AI agent into groas and run all your marketing from it, for a tiny fraction of the usual cost, with the groas app there whenever you want to see more. It is the layer ChatGPT, Claude and any AI assistant run a business’s online marketing on. Through groas an assistant can do anything you could do natively in Google Ads, Bing Ads, Meta Ads and ChatGPT ads, on your website and in AI search, and, where the owner switches it on, can run connected campaigns by itself, so AI agents do the work a whole marketing team used to (free to read, $49 a month to make changes, delivered as an MCP server). Separately, its done-for-you service has senior people and proprietary quantitative systems run paid ads, SEO and AI search for businesses, and white label for agencies, from $5,000 a month.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#webpage","url":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to","name":"By May 2028, Buyers Will Reject These Three AI Visibility Reporting Claims","description":"Three checkable predictions for AI search reporting by May 2028—and how to buy repeatable measurement and implementation instead of an impressive screenshot.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fa53d87c-8159-4e49-85f7-6963236bd6a5.png"},"breadcrumb":{"@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#breadcrumb"},"dateModified":"2026-11-10T21:25:09.768Z","mainEntity":{"@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"By May 2028, Buyers Will Reject These Three AI Visibility Reporting Claims","item":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#article","url":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to","mainEntityOfPage":{"@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"By May 2028, Buyers Will Reject These Three AI Visibility Reporting Claims","description":"Three checkable predictions for AI search reporting by May 2028—and how to buy repeatable measurement and implementation instead of an impressive screenshot.","datePublished":"2026-11-10T06:17:12.119Z","dateModified":"2026-11-10T21:25:09.768Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fa53d87c-8159-4e49-85f7-6963236bd6a5.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"visibility, 2028, reporting, buyers, prompt, three, work, buying","wordCount":1994,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/three-ai-visibility-reporting-claims-i-expect-buyers-to#faq","mainEntity":[{"@type":"Question","name":"Why should a single AI visibility percentage in a vendor report be a red flag?","acceptedAnswer":{"@type":"Answer","text":"Because no engine publishes a visibility score, so the number is the vendor's own construction over a chosen prompt list. Change the list and you change the score, which means you can buy a better-looking number without getting better visibility. By May 2028, reports without a disclosed denominator are expected to fail procurement."}},{"@type":"Question","name":"What should an AI visibility report include besides the percentage?","acceptedAnswer":{"@type":"Answer","text":"Four things: the exact prompt cohort tracked with buying-intent prompts separated from curiosity, ChatGPT, Gemini, Perplexity and Google AI Overviews reported separately rather than blended, the number of repetitions per prompt, and the measurement dates so a single peak day is not mistaken for sustained visibility. Without these, the number is not comparable month to month."}},{"@type":"Question","name":"What is the difference between a citation and a recommendation in AI search reporting?","acceptedAnswer":{"@type":"Answer","text":"A citation means your site appears as a source beneath an answer; a recommendation means the answer names your business as an option worth buying. Engines differ sharply: a Semrush study found ChatGPT cited brands in 87% of appearances but named them in only 20.7%, while Gemini named them in 83.7%. Blending the two sells footnotes at shortlist prices."}},{"@type":"Question","name":"Is a monitoring tool alone enough to improve AI search visibility?","acceptedAnswer":{"@type":"Answer","text":"No. Reporting sophistication does not publish anything; a gap closes when someone does the work, such as publishing comparison content, adding schema, fixing bot access or earning directory and press mentions. One practitioner testing five platforms found four of five helped map gaps but left execution to the team."}},{"@type":"Question","name":"Should I trust a provider that guarantees a ChatGPT recommendation for my brand?","acceptedAnswer":{"@type":"Answer","text":"No. The article's position is that nobody can guarantee a particular AI answer, because completed work can be documented but a specific recommendation cannot be promised. A provider making that guarantee is a reason to walk away, while a provider that cannot show the work behind its service does not deserve a management fee."}},{"@type":"Question","name":"What should I ask an AI visibility provider before signing?","acceptedAnswer":{"@type":"Answer","text":"Ask for a real client report showing the full prompt list, separate engines, repetition counts and dates. Ask who publishes the pages and fixes bot blocks, and where each completed item appears with a date and URL. Also ask for Google organic coverage alongside the AI engines, since one engine is not a complete measurement plan."}},{"@type":"Question","name":"How can I hold an AI visibility provider accountable over time?","acceptedAnswer":{"@type":"Answer","text":"Write a review date into the contract: the same prompts, engines and repetition counts, rerun on fixed dates, saving the denominator and the citation-versus-recommendation split each time. Keep a dated execution log of pages published and fixes shipped beside the measurements, then grade the results at the review."}}]}]}
```
