Your AI Visibility Score Fell 4 Points. Your Reporting Tool Changed Nothing.

Subject: Your September AI Visibility Report is ready
From: insights@definitely-doing-something.ai
To: you, paying $399/month

Executive Summary (Generated)

Your AI Visibility Index fell from 41 to 37. Nothing on your website changed. Our system monitored 50 prompts across ChatGPT, Gemini, Perplexity and Google AI Overviews, then confirmed that assistants answered most of them without mentioning you. We have classified the decline as directionally insightful.

ChatGPT mentioned a competitor 14 times. It mentioned you zero times. We have classified that as an opportunity, which is less expensive for us than classifying it as work. The engine-by-engine breakdown follows, along with the same recommendations we sent last month. Please note that your logo remains prominently displayed in the dashboard header. It has not yet appeared in an assistant’s answer, but we consider this a promising placement within our own product.

Your report is available now. Your next report will be available next month, whether or not anyone acts on this one.

Engine-by-Engine Breakdown: Four Views of the Same Absence

ChatGPT: competitor mentioned 14 times; you, zero. Across 20 tracked prompts, the assistant answered confidently and cited someone else. We can show you each prompt, the competitor’s name and the date we noticed. The dashboard cannot edit the page that failed to answer the question. It can, however, export the observation as a PDF for the person who might edit it.

Gemini: sentiment neutral; presence absent. Your brand surfaced in one run, where the tone was neutral. We have rendered that finding as a donut chart for executive consumption. A small slice indicates the mention. The rest of the donut indicates the amount of meeting time available to discuss it. We recommend keeping the chart in the deck, where it will continue to look like progress from a distance.

Perplexity: tracked. Your prompt-level results are available in a separate tab. Keeping this engine separate is useful because its answers are not the same as ChatGPT’s or Gemini’s. Combining them into one index is also useful, principally for putting one large number at the top of this email. We have chosen the large number for the subject line and the separate results for anyone prepared to open a tab.

Google AI Overviews: we noticed. You were not cited. Over 90% of AI Overview citations come from pages already ranking in the top 10, which suggests that a missing or poorly ranked page will not become quotable merely because we tracked its absence. We have logged this insight for next month. No page was created, improved or moved closer to being cited by the act of logging it.

Opportunities Identified (Carried Over From Last Month)

1. “Best CRM for contractors”: A competitor was cited; you were not. Root cause: content gap. Suggested owner: your team.
2. “How much does X cost”: A competitor was cited; you were not. Root cause: content gap. Suggested owner: your team.
3. Fourteen answers with material an assistant could quote: The cited pages contained a table, a short definition or an FAQ block you do not have. Suggested owner: also your team.

We have retained the word opportunity from the previous two reports. It continues to fit in the column without requiring any change to your site. Each row has a status field, currently set to identified. We can also set it to reviewed once someone opens this email. Neither status tells us whether a useful answer has been published.

This is a familiar service model. Ranking roundups describe mentions, citations, sentiment, share of voice and prompt-level counts across the major engines. They also describe the practitioner complaint: monitoring without action. We have reproduced that complaint faithfully for three consecutive months. Your competitors, regrettably, appear to be reproducing pages an assistant can quote.

Actions Taken by Your Platform This Month

None. Your plan includes tracking, scoring and alerting. Content rewrites, technical fixes and off-site citations are available as insights on your current tier. On Enterprise, they are available as longer insights. We can identify the missing paragraph, display a preview of the missing paragraph’s likely location and send an alert when the paragraph is still missing. Writing it is outside the scope of your subscription.

I used to tell clients a score was the first step toward getting cited. I was wrong. A score is what you look at while someone else does the work. If no page changes, no technical fix ships and no third-party source corroborates what you say, our score has not supplied an assistant with a better source. That distinction is less attractive on a sales slide than a rising line, but it is straightforward: measurement records the gap; execution closes it.

Our activity log for September therefore contains 50 prompts monitored, several findings grouped, one report generated and zero changes shipped. The log is complete. It is also a useful demonstration of why a complete activity log is not necessarily a useful work log.

Recommended Next Step: Review the Review

Your account manager, represented here by a calendar link, recommends upgrading to Enterprise for deeper insights. Deeper insights means the same donut, tracked more often, with opportunity in a larger font. A strategy call is available to review this report. A second call can review the conclusions from the first call. No pages will be rewritten on either call. Calls are where execution goes to be discussed.

For context, the upsell is a sturdy business model. Otterly Lite starts at $29/month for 15 prompts, with Standard at $189 and Premium at $489; Profound Starter is $99/month for ChatGPT-only tracking and Growth is $399 for multi-engine tracking; Evertune Pro starts at $800/month. Enterprise Profound contracts typically run $30k to $100k+ per year. These are different products and tiers, but the price list makes one point worth carrying into your next buying decision: more coverage is not, by itself, more work done. What the deck calls expanded coverage, I call paying more to watch yourself lose on additional engines.

Renewal Reminder: Your Score Is Not an Edit Log

Your annual renewal is due in 23 days. Your AI Visibility Index fell 4 points this month, which our retention model flags as a strong reason to keep paying. After all, if you cancel now, who will tell you it fell 5 points next month?

We understand the appeal of continuity. You can compare September with August, examine the falling line and attach both reports to a budget request. You can also compare the opportunities sections: same missing answers, new date. The comparison will tell you that the problem persisted. It will not tell you who owned the edit, when a fix shipped or whether the next set of answers changed after it did. For those details, please consult a system that performs work.

Methodology footnote, in 6-point type: Scores are calculated from a small set of single runs. A SparkToro study of 2,961 runs found the same brand list in the same order about 1 in 1,000 times, which is why we encourage you to treat one Tuesday afternoon sample as your definitive market position. We also combine ChatGPT, Gemini, Perplexity and AI Overviews into one number, although AI Overviews appeared on 43% of 160 test queries, almost always informational, and AI Overviews showed much higher week-to-week volatility with 20-plus citations, versus 5–7 source cards in AI Mode. One number to rule them all. Very scientific. Very stable.

Please do not ask whether ChatGPT-referred visitors converting at 15.9% versus 1.76% for organic makes your 14 uncited competitor mentions a revenue question. That falls outside the scope of monitoring. Our Q4 roadmap includes additional sentiment colours and an AI-written summary of the summary. Execution remains on your roadmap, not ours.

Pre-Renewal Checklist: Evidence We Do Not Currently Provide

Before approving another year, open last month’s report beside this one. For each repeated opportunity, ask what changed between the two dates. The buying distinction is dashboard or workflow: a report of what happened, or a system that decides what to monitor, what to fix, what to publish and how to prove it. Please request the following in order:

1. Pages rewritten or created for lost prompts. Ask for the page, the descriptive question heading and a standalone answer near the top; check whether a useful table, definition or FAQ block was added where the cited competitor had one.
2. Technical fixes shipped and dated. Check crawl access, indexation, clean HTML, schema and internal links; observed is not the same status as fixed.
3. Before-and-after mention rates by engine. Use repeated runs after the edits ship; one run is noise, and one blended index conceals which engine changed.
4. Off-site citations and mentions earned. Name the third-party page that corroborates your claim; this is the item most often skipped, and skipping it leaves the report identifying a credibility gap nobody has closed.

We anticipate a question about item three. Repeated buyer prompts across ChatGPT, Gemini and Google AI Overviews can make the reporting cleaner. Separate results can show where you appeared, where you did not and which competitor appeared instead. They cannot explain the loss on their own. A trend line does not write the missing answer. We learned that distinction after sending clients attractive lines that moved down and to the right while their sites stayed the same.

A usable content-gap view names the prompt, shows the competitor passage an engine used and points to the page on your site that should answer the question. It then connects that finding to an edit and a later set of runs. Opportunity is a category. A dated change to a specific page is a piece of work. If your platform displays the former but cannot show the latter, keep the screenshots for your records; they may be the fullest deliverable you receive.

For a longer look at the execution test, see tools that help you get featured in Google AI Overviews, not just monitored. For the work that sits beyond a tracking subscription, see the bill beyond tracking. If you would rather not staff the edits, fixes and off-site work yourself, groas Earned Search is the done-for-you alternative: it performs the work and logs what moved. This comparison is inconvenient for a monitoring product, so we have placed it after the methodology footnote.

P.S. From Your Platform

We have attached last month’s report for comparison. The numbers are slightly worse and the recommendations are identical. Our system detected that you opened this email; we will log that as engagement. If you forward it to your boss, please include the donut. Bosses love the donut. The donut is the only part of this product anyone has ever complimented.

If your next PDF cannot point to a page it rewrote, a fix it shipped or a third-party mention it earned, it can still offer an excellent account of the problem. You can compare prompt counts, inspect sentiment colours and schedule a call about the missing answer. You can even circle one lost prompt and ask for the edit log. We will send you the opportunity row again, now with a later date.

I would cancel the expensive thermometer and pay for the work.