How to check if Perplexity cites your brand
A name in the paragraph and a link in the numbered list are different outcomes. Read both before you call the afternoon a win.
Built by StyleLine FZE in Sharjah. Email support@mentionclock.com. We reply within one business day.
Choosing a tool is a different job from a single method. The dated vendor scorecard is best ChatGPT brand monitoring tools.
A Perplexity answer can praise a rival in the prose and still attach your documentation in the numbered list, or it can name you and cite only roundups you do not control. Checking whether Perplexity cites your brand means opening the source list, not stopping when the first sentence looks friendly. Do that on a short frozen set of buyer questions, write down the hosts, and only then decide whether the next job is a page, a profile, or a publisher.
Perplexity is built to show sources beside claims. That list is the bibliography for the answer. Mention Clock can store those URLs when Perplexity is one of the engines on the project. Paid plans query ChatGPT, the second chat engine, and Perplexity. A free project includes two of those engines, weekly. The upcoming engine is being added. Mention Clock does not scrape Google AI Overviews, so an Overview that names you is a separate manual note, not a Perplexity citation.
Name versus link
Treat two marks as separate columns.
Named. Your brand string appears in the answer text. Order matters when several vendors are listed. First and fourth are not the same conversation, but both are still a name.
Cited. At least one URL on your domain appears in the numbered source list. A citation can happen with no recommendation. A recommendation can happen with no URL of yours.
Host that mentions someone else. A review site, a rival blog, or a publisher roundup sits in the list and the page talks about competitors. That row is the work list. It is not proof that Perplexity "does not know" you. It is proof of which documents the answer leaned on.
Write Y or N for named, Y or N for cited, and the hostnames in the order Perplexity printed them. If the list is long, keep every host on the first pass. Dropping "unimportant" directories is how a pattern stays invisible.
Set up the afternoon
You need the brand string you expect to see, the domain you hope to see, and the rival names sales already loses to. You also need questions a buyer would type, not a slogan. Eight prompts are enough for a citation check. Save a weekly trend for later. Eight is small enough to finish in one sitting and large enough that one odd answer cannot dominate the notes.
Use a test login, or a window that is not your everyday Perplexity account. Personal history and custom instructions can tilt an answer toward pages you already opened. Note the model label shown in the product that day in a Notes cell. Consumer Perplexity and an API run can disagree on the same wording. If you later compare a Mention Clock row with the browser, expect direction across repeats, not a photocopy of one session.
Paste one prompt, wait until the answer and the source list both finish, then log before you paste the next prompt. Do not rewrite the question because the first answer was disappointing. A changed question is a new check.
Prompts that expose citations
Pick prompts where a buyer expects proof, because those answers grow a source list. A useful mix:
- A category question with a constraint, such as team size, industry, or region.
- A problem question that describes the job, not the product category name.
- An alternatives question that names a rival.
- A comparison that names you and one rival.
- A "best for" question a small team would ask.
- A pricing question, only if you are willing to read a wrong number calmly.
- A question about integrations or a workflow your docs already explain.
- A question a new hire would ask after reading your homepage once.
Keep the wording in a doc you can paste again. If legal or product later changes a claim, update the public page first, then keep the prompt text still so the next check is comparable.
Skip prompts that only ask for a poem, a rewrite, or a summary of text you pasted. Those answers often have no buyer bibliography. They will not tell you whether Perplexity cites your brand on a purchase question.
How to read one source list
Go line by line. For each number, record the visible title if it helps you recognize the page, the full URL when you can copy it, and the registrable host. Then mark one of these roles:
- Your domain, including docs, blog, or help center subdomains if you own them
- A rival's own site
- A review or directory profile
- A publisher roundup or news story
- A community thread
- A page you cannot classify yet
Also mark whether the answer text named you. A cited docs page plus no brand name is still useful for retrieval, and weak for a buyer who only reads the paragraph. A brand name plus only rival hosts means the prose is ahead of the documents.
If a source is a PDF or a help article, write the host anyway. The pattern is the site, not the file type. If Perplexity shows the same host twice with two paths, keep both paths. Two URLs on your docs site is a stronger citation than one homepage mention buried in a directory.
Wrong facts belong in Notes even during a citation check. A numbered link to your pricing page does not excuse an answer that quotes a retired tier. Accuracy flags in Mention Clock exist for stale price and a dead feature. On a manual pass, the Notes cell is that flag.
Sort the hosts before you assign work
After eight answers, make three piles.
Your domain appeared. Note which path. Homepage, docs, comparison, and a 2023 blog post are different assets. If only an old post is cited, that post is now part of the buying journey whether you planned it or not. Update the claim on that URL or stop linking to it from newer pages.
Rivals' domains appeared and yours did not. Read the path. Often it is a comparison, an integration story, or a glossary that defines the category in their words. Your matching page has to exist and has to be specific. A slogan on the homepage does not replace a page that answers the prompt.
Third party hosts appeared. Group them. Review sites, publishers, and community threads want different owners. A profile gap is not the same task as a contributed quote on a roundup. Do not average them into "do more PR."
Count how many of the eight answers cited you, and how many named you. Those two counts are the whole result of the afternoon. They are not a share of a market, and they are not a rank on Google. If you want the share of names across a rival set, use the definition on share of voice in AI search: your mentions divided by mentions of every tracked brand on the same prompts. That formula is a different column from "did our domain appear."
What to do before the end of the day
Assign one owner to the host that showed up most often without you. One owner, one URL, one next edit or one outreach note. Leave the other hosts on the list so they are not forgotten, but do not start seven rewrites from a single afternoon.
If your domain was cited and the facts were wrong, fix the cited page the same day if you can. Engines repeat what that URL says. If you were named and the only sources were rivals, the paragraph is borrowing their documents. Publish or update the page that matches the prompt, with dates on any price or limit you want repeated.
If you were absent from both the prose and the list, do not invent a cause from one session. Run the same eight prompts once more on a later day before you brief a team that "Perplexity does not cite us." Retrieval moves. Two afternoons with the same wording are a check. One screenshot is an anecdote.
When the check should become a habit
Four repeats on the same eight prompts, about a week apart, are enough to see whether a citation is sticky. Mention Clock runs weekly on Free and on Starter. Pro and Agency run every 3 days, which is useful when the cited pages are changing during a launch. You do not have to fill a larger prompt cap to learn from eight questions. Empty slots are allowed. The invoice does not drop because a slot is empty, and a slogan variant is a poor use of a slot.
Free is 1 project, 5 prompts, and 2 engines. If Perplexity is one of the two, pick a second engine only when you want a comparison column. Starter is $29 a month ($290 for the year) for 25 prompts and the three engines shipped today, with Slack or Teams. Pro is $79 a month ($790 for the year) and adds the cited source report so you are not exporting hosts by hand. Agency is $199 a month ($1,990 for the year) when the work is a roster of client projects and a white label PDF. Yearly prices are 10 months. The live card is pricing.
People who want a sheet instead of a project can log the same columns in the tracker:
Set the engine cell to perplexity. The Excel file has a tab that rolls hosts up. The CSV is the row log only.
The weekly method, including mention rate over several weeks, lives on how to track brand mentions in Perplexity. Use that when you already know citations matter and you want a trend. Use the afternoon above when someone asks a yes or no question: does the source list include us?
What not to claim
Do not tell a founder that a citation rate from eight prompts is the company's Perplexity performance. Say which eight prompts, which day, and whether you used a browser session or a scheduled run. Do not fold the upcoming engine or an AI Overview into the same percentage. Those surfaces are not what this check measured. The upcoming engine is being added to Mention Clock. Overviews stay something you open yourself on the queries that pay the bills, which is the point of AI Overviews brand mentions.
Do not promise that a new blog post will enter the numbered list next week. The check tells you which hosts are already trusted for these prompts. Publishing into a gap is the work. The next logged run is the evidence.
A quiet next step
When the eight prompts are worth repeating, a free project can hold five of them on two engines every week, with no card: create a free project. If you only want a first look at three questions on one engine before you open Perplexity, the AI visibility checker stays logged out.
Related guides
- How to track brand mentions in Perplexity for the weekly log and the domain rollup
- How to track brand mentions in AI chats when the same prompts should also run on ChatGPT
- How to write a weekly AI visibility report when the citation notes have to fit in a memo
- Resources for the rest of the guides
FAQ
Does a mention count as a citation?
No. A mention is your name in the answer text. A citation is your URL in the numbered source list. Log both. They lead to different edits.
Can Perplexity cite a rival and still name us?
Yes. The paragraph can list you while every number points at another company's site or at a roundup. That split is the reason to read the list instead of skimming the first lines.
Which engines can Mention Clock query for this?
Paid plans query Perplexity, ChatGPT, and the second chat engine. Free includes two of those engines. The upcoming engine is being added. The product does not scrape Google AI Overviews.
How many prompts do I need for a citation check?
Eight buyer questions are enough for an afternoon. Move to a weekly repeat on the same wording before you describe a trend. A free project holds five prompts if you want the schedule handled for you.