---
title: "Postmortem: Six Months of AI Visibility Tracking, Zero New Citations"
description: "Six months. About 200 prompts tracked across ChatGPT, Perplexity and Google AI Overviews."
url: "https://groas.com/post/postmortem-six-months-of-ai-visibility-d"
image: "https://groas.com/media/blog/78ea398f8cc316f2a9d03fd8e565a4c3b314cdf4b7a11fac7c6d8a1ea6f09835.png"
published: "2026-10-05T05:28:42.782Z"
modified: "2026-10-05T05:28:43.226Z"
---

[AI For Google Ads](https://groas.com/category/ai-for-google-ads) · October 5, 2026 · 9 min read

# Postmortem: Six Months of AI Visibility Tracking, Zero New Citations

[DavidFounder & CEO @ groas](https://groas.com/author/david)

![Cover image for: Postmortem: Six Months of AI Visibility Tracking, Zero New Citations](https://groas.com/media/blog/78ea398f8cc316f2a9d03fd8e565a4c3b314cdf4b7a11fac7c6d8a1ea6f09835.png)

In this article

1. [Month 0: The team buys a tracker, not a way to ship](#month-0-the-team-buys-a-tracker-not-a-way-to-ship)
2. [Month 1: The baseline feels like progress](#month-1-the-baseline-feels-like-progress)
3. [Month 2: The report lands in a shared drive](#month-2-the-report-lands-in-a-shared-drive)
4. [Months 3–5: The chart updates; the site does not](#months-35-the-chart-updates-the-site-does-not)
5. [Month 6: The renewal exposes the missing owner](#month-6-the-renewal-exposes-the-missing-owner)
6. [The fix: put a publishable change beside every finding](#the-fix-put-a-publishable-change-beside-every-finding)
7. [Pre-purchase checklist: will anything change after the report?](#pre-purchase-checklist-will-anything-change-after-the-report)
8. [The receipt at the end of six months](#the-receipt-at-the-end-of-six-months)

Six months. About 200 prompts tracked across ChatGPT, Perplexity and Google AI Overviews. **Zero new citations** and a very accurate chart showing competitors winning.

This is a composite, not a client account or a measured case study. The figures illustrate the version of this failure I keep seeing in AI search, after watching the same pattern with PPC graders and audits for years: a business buys measurement without assigning execution. Then it watches the dashboard record the loss in high resolution. The mistake happens in month zero, long before the renewal email arrives.

## Month 0: The team buys a tracker, not a way to ship

Say you run a service business spending $20k a month. Someone forwards a screenshot in which ChatGPT recommends three competitors and not you. The team agrees to “see where we stand first.” Sensible enough. It buys a tracker. Entry prices make that an easy decision: [Otterly at $29 a month for 15 prompts, Peec for 50 prompts, and Scrunch at $250 a month for 125 prompts](https://www.sona.com/blog/how-to-choose-an-aeo-checker-that-measures-more-than-mentions/). Nobody asks who will rewrite the pages or get technical changes published. That feels like a question for later.

I used to bless this sequence in PPC. Run the free grader, get the 40-point audit, then figure out resourcing. I was wrong about the order. What the deck calls a phased approach, I call buying a speedometer when the car has no engine. [HubSpot describes the category job as measuring and then recommending](https://blog.hubspot.com/): check citations in ChatGPT, Perplexity and Gemini, check presence in AI Overviews, and turn the findings into recommendations. That is useful work. It is also where this story stalls.

The purchase order names the tool. It does not name the person who can change the site. **That omission is the day-one decision.**

## Month 1: The baseline feels like progress

Someone loads 200 buyer-intent prompts, tags them by service line and sets a regular check. The first report puts share of voice at 11%. Mention rate is lower; citation quality is thinner. Sentiment looks fine, which tells the team little about whether a buyer will find it. The chart looks great in the Monday meeting. At last, everyone can see where the business stands.

The [usual monitoring loop covers prompt coverage, mention rate, citation quality, sentiment and share of voice](https://pinggy.io/blog/best_ai_search_analytics_monitoring_tools/). In this composite, the team completes the measurement half. It does not connect each gap to a page, a technical change and someone with the authority to publish it. Because the tool was bought to observe answers, observing answers feels like using it correctly.

That is the trap. A baseline can tell the team which questions matter and which competitors appear. It cannot decide whether a service page needs a direct answer, get copy approved or push a fix through the CMS. By the next meeting, the chart has refreshed. No URL has changed because of it.

## Month 2: The report lands in a shared drive

By week six, the tool has done its job. The composite report shows 47 prompts where a competitor gets cited and the business does not. It flags three service pages that do not answer the buyer’s question directly, missing FAQ schema and thin location pages. The export runs to 30 lines.

Then it sits in a shared drive.

Marketing commissioned the report and assumes IT or the agency will ticket the work. IT assumes marketing owns the copy. Leadership said to “get it into the queue” but never defined whose queue, what goes first or who can approve publication. [That is how audit findings can end with low-risk tweaks shipped while structural and crawl issues sit untouched](https://bestwebsite.com/blog/how-to-decide-who-owns-website-audit-findings-after-the-report/). The list is specific enough to make people feel busy and vague enough to let everyone wait.

I watched this stall with PPC audits. A list could call for ad-group restructuring, broad-keyword cuts and landing-page rewrites. The team would pick the bid adjustment it could make that afternoon and leave the harder changes alone. [Search Engine Land makes the same point about long, unprioritized audits: teams cherry-pick easy fixes](https://searchengineland.com/seo-audits-waste-fix-439063). The problem is not that the easy fix is worthless. It is that completing it lets the rest of the list look less urgent.

**The report feels like progress. It is inventory.** Until someone turns a finding into a published change, the team has merely counted the work.

## Months 3–5: The chart updates; the site does not

By month three, the gap between knowing and shipping appears in the composite numbers. Share of voice slips from 11% to 9%. The competitor previously cited on 47 prompts now appears on 61. In this version of events, that competitor has published the FAQ pages and comparisons the team’s own report pointed toward. Meanwhile, the task list gains schema work, entity pages, review markup and internal linking. Only two title-tag tweaks leave it.

The dashboard is doing its job perfectly. It is accurately recording the loss.

![Wall of monitors displaying search visibility reports](https://groas.com/media/blog/89efe31d498f9ed754ae850a1a5525c30a5df5625e607385b35f52275f29eb08.png)

The tool did not promise to do all the work. [Peec AI, for example, tracks citations across ChatGPT, Perplexity and Google AI Overviews, but does not provide entity extraction, content briefs or CMS publishing](https://pixis.ai/blog/peec-ai-vs-pixis-visibility-2026-platform-comparison/). When the report exposes a gap, the team still needs someone to interpret it, someone to make the content or technical change and someone to publish. Each handoff is another place for the work to stop. You bought a smoke detector and assumed you had bought a sprinkler system.

I understand the small team that decides to monitor manually instead: one person, three hours a week, a spreadsheet of 25 buyer prompts checked across ChatGPT, Gemini and AI Overviews. That approach is at least honest about what it buys. It can keep the questions visible without adding another subscription. But it does not remove the handoffs. When the content queue fills, the spreadsheet waits beside the paid dashboard.

**Monitoring was never the bottleneck. Shipping was.** Changing the price or frequency of the check does not change who can act on it.

![Dusty car speedometer](https://groas.com/media/blog/fb43ac1126846730a909bf290aa8091b80cb6d38bed0385a093e7b76a0f3397e.png)

## Month 6: The renewal exposes the missing owner

The renewal email arrives, and the direct cost looks almost harmless. Six months at, say, $95 a month comes to $570, plus maybe 15 hours spent pulling reports and building slides. Under $2,000 in this composite. Cheap, until the team asks what happened during those six months: buyer questions kept returning competitors, while the business kept documenting the result.

The expensive part is not the subscription. It is the work that never entered a publishable queue. Teams can [measure visibility across AI search surfaces](https://www.sona.com/blog/how-to-choose-an-aeo-checker-that-measures-more-than-mentions/) and bring week-over-week screenshots to every meeting, yet struggle to name a page they published because of what they learned. More precise reporting does not close that gap. It can make the meeting about the curve instead of the site.

On day 180, someone finally asks the question that belonged on day one: **Who was supposed to change the site?** Not “What is our share of voice?” Not “How many prompts do we track?” Who ships? A tracking vendor can offer refreshes, exports and an API without promising new citations. Holding it responsible for changes nobody assigned is an excellent way to learn nothing from the renewal.

There is an exception. If a team buys monitoring as research before a site rebuild, then uses the findings to do the rebuild, it has connected measurement to execution. The tracker served its purpose. This postmortem is about the team that buys the first part and assumes the second will somehow follow.

## The fix: put a publishable change beside every finding

The fix was not a prettier dashboard. It was assigning execution before the first baseline arrived: one owner with time to prioritize findings, access to a writer or developer who can make the change, and a route to publication that does not depend on three undefined approvals. Then each check of the buyer-intent prompts has a destination. A finding becomes a proposed URL change, a person takes it on, and the team records what shipped before it reads the next chart.

I look for that action log first now. It should let me distinguish positions observed from [work done to address them](https://www.hubspot.com/). If a team can show me months of prompt history but cannot show me the pages it changed, I do not need another view of the trend. I need to know why the work stopped. A tool may reveal a gap; it cannot settle an ownership dispute nobody has named.

The distinction also matters when comparing products. If you want a business [featured in Google AI Overviews instead of only monitored](https://groas.com/post/tools-to-get-featured-in-google-ai-overviews), ask what happens after a gap appears. Does someone produce and publish the relevant content? Can someone close the technical gap? Does the team re-check the same questions afterward? That is the work. The chart tells you where to start and whether the answer later changes; it does not make either thing happen.

## Pre-purchase checklist: will anything change after the report?

Run this before paying for a tracker or an execution service. The point is not to award the best-looking chart. It is to find out who can change a cited URL, how the work reaches publication and whether anyone can verify what happened afterward. If the subscription looks cheap, read [what it actually costs to get cited](https://groas.com/post/what-it-actually-costs-to-get-cited-in-g) alongside it. The dashboard is only one line of the bill.

1. **Does the product change the site or only report on it?** Exports and recommendations still require someone to write, develop and publish.
2. What will monitoring cost in-house versus alongside execution? Count staff time as well as the tracker fee.
3. Does the comparison use cited URLs rather than only proprietary scores? [Check the same 25 buyer-intent prompts across shortlisted tools in the same week, compare cited URLs and test the export](https://www.sona.com/blog/how-to-choose-an-aeo-checker-that-measures-more-than-mentions/).
4. Does coverage include ChatGPT, Gemini, Perplexity and Google AI Overviews for the prompts you care about? Do not mistake one surface for the whole picture.
5. **Is there a log of actions taken, not just positions observed?** Look for the date, the URL and what changed. A movement on a chart is not an account of the work.
6. **Who ships the content and technical fixes, and when?** “The team” is not an owner. Leave this item blank and you pay for another month of findings with no publish date.

## The receipt at the end of six months

In this composite, the tracker records a fall from 11% to 9% share of voice and a competitor’s cited prompts rising from 47 to 61. The site does not change in the ways the report calls for. The dashboard did not lie. It did what a receipt does: showed what the business paid to observe.

I would not cancel a tracker in month four just to prove a point. I would stop adding prompts, take the 25 buyer-intent questions closest to a purchase, and assign one owner to get the relevant changes published. Then I would re-check those same questions. Citations may or may not move immediately, but at least the team would know what it changed and why. That is a better basis for a renewal decision than six more screenshots.

The PPC version taught me the same thing without the newer vocabulary. A grader could identify waste. An audit could explain it. Neither could cut a keyword, rewrite a landing page or resolve an approval bottleneck on its own. The AI visibility dashboard is that audit with a monthly charge. My rule now is one line: **never buy measurement without assigning execution on the same day.**

## Frequently Asked Questions

### Why did tracking AI visibility for six months produce zero new citations?

Because the team bought a tracker without assigning anyone to change the site. The dashboard accurately recorded competitor wins while no page, schema or technical fix was ever published, so citations never moved.

### When should a team decide who is responsible for acting on AI visibility reports?

On day one, before the first baseline is measured. The purchase order names the tool but rarely names the person who can change the site, and the article argues that omission is the decision that causes the whole failure.

### Can an AI search tracking tool fix the problems it finds?

No. A tracker can show which prompts return competitors and which pages miss the buyer's question, but it cannot write copy, get it approved or push a fix through the CMS. Someone with authority to publish still has to do that work.

### What does it mean that a report is inventory and not progress?

A list of findings like missing FAQ schema or pages that do not answer the buyer's question only counts the work. Until someone turns a finding into a published change, nothing has moved, and completing easy tweaks can make the harder ones look less urgent.

### What is the real cost of an AI visibility tracker that goes unused?

Not the subscription itself. In the composite example, six months at $95 a month plus reporting time came to under $2,000, but the expensive part was the work that never entered a publishable queue while competitors kept winning the buyer prompts.

### How do you avoid buying measurement without execution?

Assign execution the same day you buy measurement. Name one owner with time to prioritize findings, give them access to a writer or developer, and define a route to publication so each finding becomes a proposed URL change that someone ships and records.

### What should I check before buying an AI visibility tracking tool?

Ask whether the product changes the site or only reports on it, count in-house staff time alongside the fee, and compare tools on cited URLs rather than proprietary scores. Also check for a log of actions taken, with dates and URLs, and confirm who will ship the fixes and when.

### Is there any case where buying an AI visibility tracker on its own makes sense?

Yes, when a team buys monitoring as research before a site rebuild and then uses the findings to do that rebuild. In that case measurement is connected to execution. The failure mode is assuming the execution part will somehow follow on its own.

## Related Posts

- [![Clay figure pours a wheelbarrow of articles into a wall funnel whose pipe is disconnected, so the pages pile on the floor beside an unused wrench.](https://groas.com/media/blog/e1a99963f159a4756e614472ee73400643c5900fa7bde718abe8fa563e8cdc77.png)](https://groas.com/post/can-chatgpt-even-read-your-site-a-45-min)
  ### [Can ChatGPT Read Your Site? A 45-Minute Check Before You Write More AEO Content](https://groas.com/post/can-chatgpt-even-read-your-site-a-45-min)
  [October 6, 2026 • 10 min read](https://groas.com/post/can-chatgpt-even-read-your-site-a-45-min)

  [Written by Alexander Perelman](https://groas.com/post/can-chatgpt-even-read-your-site-a-45-min)
- [![Cartoon of a marketer noting 'position two' after one pull of a slot machine whose reels are already spinning to a new result.](https://groas.com/media/blog/038bfefb4237637a1718ff66b379d5bdc42d57615f25797529aa5ef5682a4f7e.png)](https://groas.com/post/i-measured-ai-visibility-like-a-ppc-acco)
  ### [Six PPC Habits That Misread AI Visibility](https://groas.com/post/i-measured-ai-visibility-like-a-ppc-acco)
  [October 6, 2026 • 10 min read](https://groas.com/post/i-measured-ai-visibility-like-a-ppc-acco)

  [Written by David](https://groas.com/post/i-measured-ai-visibility-like-a-ppc-acco)
- [![A tiny worker on a ladder fixes up five model shopfronts while a giant dusty 'take a number' dispenser looms behind, its ticket queue ignored.](https://groas.com/media/blog/35c8ce6c06242ab08178711c5abe08dda0df51ec927a2dd6bc9770aa5f83bdbf.png)](https://groas.com/post/one-strategist-zero-developers-five-clie)
  ### [One Strategist, Zero Developers, Five Domains: A 30-Day AEO Launch Playbook](https://groas.com/post/one-strategist-zero-developers-five-clie)
  [October 6, 2026 • 10 min read](https://groas.com/post/one-strategist-zero-developers-five-clie)

  [Written by David](https://groas.com/post/one-strategist-zero-developers-five-clie)

## The Machines Already Run Search, You Should Probably Own One

[get started free](https://groas.typeform.com/to/xC1bQNUT)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","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/postmortem-six-months-of-ai-visibility-d#webpage","url":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d","name":"Postmortem: Six Months of AI Visibility Tracking, Zero New Citations","description":"Six months. About 200 prompts tracked across ChatGPT, Perplexity and Google AI Overviews.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://groas.com/media/blog/78ea398f8cc316f2a9d03fd8e565a4c3b314cdf4b7a11fac7c6d8a1ea6f09835.png"},"breadcrumb":{"@id":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d#breadcrumb"},"dateModified":"2026-10-05T05:28:43.226Z","mainEntity":{"@id":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d#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":"AI For Google Ads","item":"https://groas.com/category/ai-for-google-ads"},{"@type":"ListItem","position":4,"name":"Postmortem: Six Months of AI Visibility Tracking, Zero New Citations","item":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d#article","url":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d","mainEntityOfPage":{"@id":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"Postmortem: Six Months of AI Visibility Tracking, Zero New Citations","description":"Six months. About 200 prompts tracked across ChatGPT, Perplexity and Google AI Overviews. Zero new citations and a very accurate chart showing competitors winning. This is a composite, not a client account or a measured case study. The f…","datePublished":"2026-10-05T05:28:42.782Z","dateModified":"2026-10-05T05:28:43.226Z","image":{"@type":"ImageObject","url":"https://groas.com/media/blog/78ea398f8cc316f2a9d03fd8e565a4c3b314cdf4b7a11fac7c6d8a1ea6f09835.png"},"author":{"@id":"https://groas.com/author/david#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"articleSection":"AI For Google Ads","keywords":"Google Ads, AI For Google Ads","wordCount":1928,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/david#person","name":"David","jobTitle":"Founder & CEO @ groas","image":"https://groas.com/media/blog/363be9ad5255816654dac065d01bf1c6063110b400bb38cc199570267232027c.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"FAQPage","@id":"https://groas.com/post/postmortem-six-months-of-ai-visibility-d#faq","mainEntity":[{"@type":"Question","name":"Why did tracking AI visibility for six months produce zero new citations?","acceptedAnswer":{"@type":"Answer","text":"Because the team bought a tracker without assigning anyone to change the site. The dashboard accurately recorded competitor wins while no page, schema or technical fix was ever published, so citations never moved."}},{"@type":"Question","name":"When should a team decide who is responsible for acting on AI visibility reports?","acceptedAnswer":{"@type":"Answer","text":"On day one, before the first baseline is measured. The purchase order names the tool but rarely names the person who can change the site, and the article argues that omission is the decision that causes the whole failure."}},{"@type":"Question","name":"Can an AI search tracking tool fix the problems it finds?","acceptedAnswer":{"@type":"Answer","text":"No. A tracker can show which prompts return competitors and which pages miss the buyer's question, but it cannot write copy, get it approved or push a fix through the CMS. Someone with authority to publish still has to do that work."}},{"@type":"Question","name":"What does it mean that a report is inventory and not progress?","acceptedAnswer":{"@type":"Answer","text":"A list of findings like missing FAQ schema or pages that do not answer the buyer's question only counts the work. Until someone turns a finding into a published change, nothing has moved, and completing easy tweaks can make the harder ones look less urgent."}},{"@type":"Question","name":"What is the real cost of an AI visibility tracker that goes unused?","acceptedAnswer":{"@type":"Answer","text":"Not the subscription itself. In the composite example, six months at $95 a month plus reporting time came to under $2,000, but the expensive part was the work that never entered a publishable queue while competitors kept winning the buyer prompts."}},{"@type":"Question","name":"How do you avoid buying measurement without execution?","acceptedAnswer":{"@type":"Answer","text":"Assign execution the same day you buy measurement. Name one owner with time to prioritize findings, give them access to a writer or developer, and define a route to publication so each finding becomes a proposed URL change that someone ships and records."}},{"@type":"Question","name":"What should I check before buying an AI visibility tracking tool?","acceptedAnswer":{"@type":"Answer","text":"Ask whether the product changes the site or only reports on it, count in-house staff time alongside the fee, and compare tools on cited URLs rather than proprietary scores. Also check for a log of actions taken, with dates and URLs, and confirm who will ship the fixes and when."}},{"@type":"Question","name":"Is there any case where buying an AI visibility tracker on its own makes sense?","acceptedAnswer":{"@type":"Answer","text":"Yes, when a team buys monitoring as research before a site rebuild and then uses the findings to do that rebuild. In that case measurement is connected to execution. The failure mode is assuming the execution part will somehow follow on its own."}}]}]}
```