Mention Clock

What to do when ChatGPT recommends a rival

Treat the answer as a work order for a page and a listing. The next run is a measurement. It is not a promise that you will be named.

Built by StyleLine FZE in Sharjah. Email support@mentionclock.com. We reply within one business day.

ChatGPT recommended a competitor and left you out, or it mentioned you under a rival with a sentence that is no longer true. The useful response is a log plus two kinds of edits: a page you control, and a public listing buyers already read. There is no switch that inserts a brand into an answer. Publishing can change what a later answer is able to cite. It does not guarantee that it will.

Do not start by rewriting the prompt until your brand appears. A friendlier prompt is a different question. You will feel better and you will know less. How to turn a ChatGPT screenshot into a weekly check is how to recover the wording from an image. The steps below start once you have that wording, or once a scheduled run already stored it.

Log the answer before anyone edits

Write four facts while the answer is still on screen.

The prompt, character for character. If a colleague paraphrased it, mark the note as reconstructed.

The brand names, in the order they appeared. Include yourself if you appeared, and the place (first, second, later, or absent).

Every cited URL, including rival help docs, review roundups, and your own pages if any were cited.

Any sentence about your product that is false: old pricing, a missing integration, the wrong buyer, a feature you retired.

That is the whole incident. Do not add a theory about why the model "prefers" the rival. You do not have that evidence. You have a prompt, an order, a URL list, and a fact check.

If the session was the second chat engine or Perplexity, label the engine. A miss on ChatGPT is not a miss on the others, and a win on one is not a win on the others. Mention Clock can run the same prompt on ChatGPT, the second chat engine, and Perplexity on paid plans. Free runs two of those. The upcoming engine is being added, so the upcoming engine screenshot is a hand note until a run includes it. AI Overviews are not a scheduled engine. Keep them in a separate manual note if you check them at all.

Content actions you can finish

Pick one cited URL that is not yours. Read the claim that earned the citation. Your job is to make the same fact easy to quote on a URL you own, and to correct your own URL if the answer repeated a stale fact.

Your comparison or job page. If the prompt was "what should a 40 person 3PL use for dock appointments," the page that should exist is about that job, that team size, and the constraint, in the first paragraph. A homepage slogan does not answer that prompt. Say what the product does, who it is for, and what it does not do. Models and buyers both get lost in a page that only says you are modern.

Your docs. If the cited rival page is a help article on a specific behavior (webhooks, SSO, a file import), your matching doc should state that behavior in plain sentences, with the limit included. A feature grid that hides the limit is how wrong answers survive.

A correction on your own site. If ChatGPT said your price, your headquarters, or your integrations incorrectly, change the HTML on the pricing page, the security page, and the one sentence description so they match. Do this before you ask anyone else to update a profile. Third parties copy you. If you are inconsistent, they will copy the inconsistency.

A note to the author of a page you do not own. If the citation is a roundup or a docs site at another company, you can send a factual correction: a wrong plan name, a retired feature, a missing constraint. You cannot require them to add you, and you should not describe a sent email as a result. The result is the next logged run, later.

Ship one of these, not all four, in the week of the miss. A pile of unpublished drafts does not change what can be cited. How to get mentioned by ChatGPT is the longer publishing week if you do not yet have the page. Use it as the production list. Use the log above as the reason you picked that URL.

Listing actions you can finish

Listings are the public profiles a buyer already trusts enough to ask an assistant about the category: your app marketplace entry, your review profile, your docs subdomain, and the partner directory of a platform you integrate with. The names will differ by product. Update the fields you control so they match the site you just fixed.

Fields worth aligning, when the profile has them:

The one sentence description. It should match the sentence on your homepage and on your pricing page, not a campaign tagline from last quarter.

The category. If sales is losing deals for "dock scheduling" and the profile says "logistics AI," the assistant is being handed a vaguer shelf than your buyers use.

Integrations and constraints. SSO, the systems you sync with, and the company size you actually serve. Omit a logo you do not integrate with. A false logo becomes a false sentence in a later answer.

Pricing pointers. Link to the live pricing URL. Do not paste a discount that has ended.

Screenshots and dates. Replace a UI you retired. An old screenshot is a citation for a product you no longer ship.

You do not control the star rating, the order of a reviewer roundup, or whether a directory accepts the edit. Record what you submitted and when. Do not record it as "we fixed ChatGPT."

Avoid buying placement or asking for a swapped mention. If a page exists only because you paid for a ranking, say that to your own team. Do not feed that page into the story as if it were an independent source. The scheduled run will cite whatever it cites. Your report should not pretend a paid logo is an organic recommendation.

What not to do in the same week

Do not change the prompt wording to include your brand "so we show up." That measures a different question.

Do not average this ChatGPT session with a Perplexity session and call it one rank. Keep the rows separate. How to track brand mentions in Perplexity is the citation habit for that engine. How to track brand mentions in AI chats is the habit for ChatGPT.

Do not open a second project for the same brand because the first answer was embarrassing. You will split the history. One brand, one prompt list.

Do not promise the client, the founder, or the sales channel that the next answer will name you. You can promise the log, the page, and the listing edit. The following run confirms whether the answer moved. Sometimes the cited host does not change. Report that. An honest unchanged row is more useful than a victory sentence the next miss will contradict.

Do not add five synonym prompts to "cover" the loss. You will spend the weekly review on duplicates. If the original prompt is a real buyer question, keep it and do the page work.

Put the same prompt on a clock

One session is an anecdote. A repeated run of the same wording is the evidence that a page edit did or did not show up in later answers. Wait until you have the prompt frozen, then track it.

Free is $0 for 1 project, 5 prompts, and 2 engines, weekly, with no card. Put this prompt and the closest sibling questions in that five. Paid plans add the third engine. Starter lists at $29 each month, and at $290 billed yearly, for 25 prompts. Pro is $79 a month or $790 a year for 100 prompts, every 3 days, and a cited source report that gathers the URLs so you are not copying them by hand. Agency is $199 a month or $1,990 a year when the work is one of several client brands.

On the next run, compare only these fields to the log you wrote: were you named, what place, which rivals, which URLs. If you edited your own page, check whether that URL appears. If it does not, the edit can still have been the right page. Say "our URL was not cited." Do not say "the edit failed" unless you also know the answer got worse on a fact you changed.

Personalized chats and a scheduled run can disagree. A buyer on a phone with memory turned on is not the same setting as a clean run. Keep the buyer's screenshot in the notes. Do not overwrite the scheduled row with it, and do not throw the scheduled row away because a founder saw something else at lunch.

When you are ready for that repeat, create a free project and paste the recovered prompt unchanged. The rest of the writeups, including the weekly note a boss can forward, sit on resources.

A practice log, not a case study

Use this empty shape in the doc where you store runs. It is a form, not a story about a customer, and not a result you should expect.

Prompt: (paste)

Engine: ChatGPT, the second chat engine, or Perplexity. The upcoming engine only if you checked it by hand.

Date and whether memory or custom instructions were on.

Named, in order: (list)

Your place: absent, or the position you saw.

Cited URLs: (list)

False sentence, if any: (quote)

Content action this week: one URL you will edit, or one author you will write to.

Listing action this week: one profile field you will align with the site.

Next run: the date the same prompt will be asked again.

If a cell is unknown, write unknown. Guessing a citation is how the following week argues about the wrong page.

Related guides

FAQ

Will these edits make ChatGPT recommend us?

There is no basis to promise that. The edits give a later answer a page and a listing that state the facts. The next run of the same prompt is how you see whether the names and the URLs changed. They might not.

Should I contact the competitor who was cited?

No. Read their page, learn which claim was citable, and put that claim truthfully on a URL you own if it is also true of you. If it is not true of you, do not copy it. Write the constraint you actually meet.

Do I fix the second chat engine and Perplexity the same week?

Log them separately if you have the rows. A content fix is still one URL. A listing fix is still one profile. Do not assume a ChatGPT miss is also the second chat engine miss. Paid plans can run all three live engines: ChatGPT, the second chat engine, and Perplexity. Free runs two.

What do I tell sales today?

Give them the four facts from the log and the one action you will ship. Tell them a single session is not a rank. Offer the date of the next run. Do not offer a date by which the brand will be recommended.

Create a free project