Optimising Brand Visibility in Perplexity
Perplexity is the easiest AI assistant to measure because it shows its sources for every answer. That makes it possible to see not just whether your brand is named, but which pages Perplexity read, how often your own site is among them, and which directories, reviews or competitors appear instead.
Many businesses look only at the answer text and miss most of that evidence. This guide shows how to test Perplexity under steady conditions, record mentions and sources separately, sort the sites it cites into types, and use that picture to decide where effort will raise your visibility.
- Why Perplexity is the assistant to measure first
- Holding test conditions steady
- Mentions and source share are different measures
- The fix, step by step
- What each type of cited source tells you
- Using what Perplexity shows around the answer
- Edge cases
- Keeping Perplexity visibility on track
- Questions and answers
Why Perplexity is the assistant to measure first
Perplexity is built around search. Each answer is written from pages it retrieves for the question and carries numbered references, with a sources view listing the pages used. For measurement, that brings three advantages over assistants that sometimes answer without sources:
- Every answer has evidence attached. You can see which pages informed the answer, not just the text that came out.
- You can separate being named from being used. Your brand can be named in the answer while your site is absent from the sources, or your page can be a source while the answer names someone else.
- Changes are traceable. Because answers depend on retrieved pages, a change to the pages that are cited can show up in later answers, and you can see when a new source starts to appear.
Perplexity does not publish how it ranks or chooses sources, and what appears varies between runs of the same question. So the goal is not to reverse-engineer a formula. It is to collect enough runs that the pattern of who gets cited, and why, becomes clear.
If your own site never appears in the sources, even for questions about your business by name, start with access instead; see fixing invisibility in Perplexity. For making individual pages more likely to be chosen, see optimising for Perplexity. This guide sits between the two: measuring brand visibility and using the measurement to direct the work.
Holding test conditions steady
Perplexity offers more than one way to answer a question, and the choice changes how many sources are read and which are cited. Before you start tracking, decide these conditions and write them down:
| Condition | Why it matters | What to do |
|---|---|---|
| Search mode | Standard answers and the more extensive research-style modes read different numbers of pages. | Use the default mode most customers would use, and never mix modes within one data set. |
| Source filters | Perplexity lets users narrow sources, for example towards academic or social content. | Leave filters on the default web setting unless you are tracking a specific filter deliberately. |
| Account | Signed-in accounts can hold preferences and history that shape answers. | Test signed out or with a clean account used only for tracking. |
| Model | Paid plans can choose between models, which write answers differently. | Stay on one model and note it with each run. |
| Location | Local questions depend on where the request seems to come from. | Name the town or region in the prompt itself. |
Mentions and source share are different measures
Record two separate things for every run, because they answer different questions and are improved by different work.
Brand mention rate
The share of runs in which the answer text names your business. This is what the person asking actually reads. You can be named even when your site is not a source, because a review site, directory or article that mentions you was cited instead.
Source share
The share of all cited sources, across all runs, that are pages on your domain. A related measure is source presence: the share of runs in which at least one of your pages is cited. Source share shows whether Perplexity is using your own site as evidence.
How the two combine
| Named? | Your site a source? | What it usually means | Where to focus |
|---|---|---|---|
| Yes | Yes | You are both known and used. | Accuracy and position within the answer. |
| Yes | No | Third-party pages carry your reputation. | Keep those pages accurate; make your own page the clearest source on your facts. |
| No | Yes | Your page was read, but you were not named. | The page may not make clear who you are or that you offer what was asked. |
| No | No | Invisible for this question. | Study the cited sources and whether you appear on any of them. |
The AI Visibility Tracker records brand presence as cited, mentioned, recommended or invisible and identifies citation gaps against competitors, which covers the same ground if you would rather not keep the records by hand.
The fix, step by step
Step 1
Fix the mode and session you test in
Choose the conditions you will use for every Perplexity check and write them down: signed out or a clean tracking account, the default search mode, the default web sources filter, and a named model if your plan lets you choose. Put the town or region in the text of local prompts. Changing any of these between runs changes which pages are read, so a rise or fall could reflect the set-up rather than your brand.
Step 2
Save the full sources list for every run
For each run, save the answer text and every source listed, with its full URL, its position in the list and whether the page mentions your brand. Do this for every prompt, not only the ones where you are missing. Over a few weeks the records show which domains Perplexity relies on for each question, which is the most useful evidence you can collect about why you are, or are not, named.
Step 3
Count mentions and source share separately
Calculate two figures per prompt group: the share of runs where the answer names your business, and the share of cited sources that are on your domain. Keep them apart rather than merging them into one score. A high mention rate with low source share means third parties carry your reputation; a decent source share with few mentions means your pages are read but do not make clear who you are or what you offer.
Step 4
Sort the cited domains into types
Label every cited domain by type: your own site, a direct competitor, a review or directory site, a comparison or news article, a forum or community thread, a video page, or an official or reference source. Count how often each type appears for each prompt group. This shows where the answers come from for your market, and therefore which kind of work, on your site or elsewhere, is likely to change them.
Step 5
Work on the sources that keep appearing
Take the third-party domains that appear most often for prompts where you are not named, and check whether you are listed or mentioned on them. Claim and complete directory and review profiles, correct wrong details, and ask real customers to leave reviews where they already buy. Where a comparison article recurs, ask the publisher whether it takes updates. Take part in community discussions openly as the business, never through disguised accounts or paid posts.
Step 6
Make your page the clearest source on your own facts
For prompts about your own services, prices or location, make sure one page
on your site answers the question directly, in text that the server sends without relying on scripts,
naming your business in the relevant sentences. Check the robots.txt and firewall rules for PerplexityBot
as well. When your page is reachable and states the facts plainly, Perplexity has less reason to rely
on a directory that may be out of date.
Step 7
Add prompts from the related questions
Perplexity suggests follow-up questions under its answers. Record the related questions that appear for your tracked prompts and look for ones that match real customer needs. Add the useful ones to your prompt set as a new, dated batch rather than replacing existing prompts, so the history stays comparable. These suggestions show where conversations tend to go next, which often exposes topics you have never written about.
Step 8
Compare months against a fixed domain list
Keep a fixed list of the domains you care about: your own, your main competitors and the third-party sites that matter most. Each month, report how often each one is cited and how often each brand is named, per prompt group. Mark on the same timeline the changes you made and any change you notice in how Perplexity works. Movement that repeats across several prompts is worth acting on; a single run is not.
What each type of cited source tells you
Once cited domains are sorted into types, the mix for each prompt group suggests where to put effort. These are patterns to investigate, not rules.
| Source type dominating | What it suggests | Sensible response |
|---|---|---|
| Competitors' own sites | Their pages answer the question more directly than yours, or yours cannot be read. | Compare the cited competitor page with yours, question by question; check your access. |
| Review and directory sites | Perplexity relies on aggregated listings for this kind of question. | Make sure your profiles exist, are complete and are accurate on each recurring site. |
| Comparison and news articles | Editorial round-ups decide who is named. | Offer publishers accurate information and something newsworthy; never pay for disguised coverage. |
| Forums and community threads | People's own experiences are treated as useful evidence. | Take part openly and helpfully where your customers already talk. |
| Video pages | Demonstrations or explanations answer the question well. | Consider a clear video with a full written description and transcript. |
| Official and reference sources | The question is factual or regulated. | Aim to be named as a next step, and link to the official source from your own page. |
An example for a fictional business, Wren & Oak Dog Grooming in Chesterfield, after three weeks of runs:
Prompt group: category ("dog groomer in Chesterfield for a nervous dog")
Runs: 15
Named: 4 of 15
Own site cited: 1 of 15
Most cited: example-petdirectory.co.uk (11), example-forum.co.uk thread (6),
competitor example-groomers.co.uk (5)
Finding: Directory profile lists wrong services and no mention of
nervous dogs; forum thread predates the business
Action: update directory profile; add a "nervous dogs" section to
services page with the approach in plain textEdge cases
Your content cited from someone else's domain
If a copy of your article on another site is cited instead of your original, the copy counts as someone else's source. Ask syndication partners to link to your original with a canonical link, and deal with unauthorised copies separately.
You are named, but described wrongly
Open the source attached to the wrong statement. It is usually an outdated directory entry, an old article or a page on your own site that was never updated. Correct the source; the answer can only be as accurate as the pages behind it.
Answers that change sharply from week to week
Some questions have many possible sources of similar quality, so the cited set rotates. Pool results over several weeks for these prompts before drawing conclusions.
Very small markets
For niche local questions there may be only a handful of relevant pages on the web. One new directory listing or review page can change the picture quickly, in either direction.
Keeping Perplexity visibility on track
- Keep test conditions fixed and record the mode and model with every run.
- Keep the domain list stable so month-on-month figures stay comparable, and add new domains as a separate note.
- Review recurring third-party profiles quarterly for outdated details.
- Watch PerplexityBot in your logs after any robots.txt, firewall or CDN change, since a new block cuts your site out of the sources quickly.
- Check AI referrals. Visits from Perplexity usually arrive with perplexity.ai as the referrer, so an analytics segment for that source shows whether being cited turns into visits.
For making the same measurement work across every assistant, see fixing AI visibility tracking, and for the ChatGPT equivalent of this guide, see brand visibility in ChatGPT.
1. How does Perplexity choose which sources to cite?
Perplexity has not published a full description of how it ranks or selects sources. What is visible is that it searches the web for each question, reads a set of pages and writes an answer with numbered references to them. Pages it cannot fetch cannot be cited, and pages that answer the question directly appear to be used more often. Beyond that, treat confident claims about exact ranking factors with caution and rely on your own records of who gets cited.
2. Why is my brand mentioned in Perplexity but my website not cited?
Because the pages Perplexity read to answer the question were about you rather than by you. Review sites, directories, forum threads and articles often mention businesses, and Perplexity can name you based on them. That is still useful visibility, but it means those third parties control how you are described. Keep your profiles on them accurate, and make sure your own site states the same facts clearly so it can be used as a source too.
3. Why do Perplexity sources change every time I ask?
Several pages can answer the same question about equally well, and the set Perplexity retrieves can differ between runs. The answer is also written fresh each time. Mode, model and source filters add further differences if they are not kept the same. This is why a single check is unreliable. Run each tracked prompt several times under fixed conditions and look at how often each domain appears across all runs rather than in one answer.
4. Should I test Perplexity signed in or signed out?
Signed out, or with a clean account used only for tracking, is the safer choice, because a personal account can carry preferences and history that shape the answers you see. If most of your customers use a paid plan with a particular model, you may prefer to track under that plan instead. Whichever you choose, use it for every run, record it in your method notes, and do not mix conditions within the same set of figures.
5. Can I pay to appear in Perplexity answers?
Perplexity has experimented with advertising, but advertising is separate from which sources an answer cites, and the details of any programme can change. For the answers themselves, Perplexity publishes no way for businesses to buy or request inclusion. The routes that fit how it works are being reachable to its crawler, publishing clear pages that answer real questions, and being described accurately on the other sites it tends to cite for your market.
6. What is a good source share in Perplexity?
There is no published benchmark, and it depends heavily on the market and question. A local business answering questions about its own services should expect to be a source often; a new brand in a crowded national category may rarely be one. The useful comparison is with your own history and with the competitors on your fixed domain list. Track the figure per prompt group and judge progress by the trend over several months.
7. Do Perplexity related questions help my visibility?
They do not change your visibility directly, but they are useful research. The related questions under an answer show where people tend to go next with a topic. If they point to subjects your site does not cover, such as costs, alternatives or aftercare, that is evidence of a content gap. Add the useful ones to your tracked prompts as a new batch, and answer them on your site where they fit.
The AI Visibility Tracker queries Perplexity alongside ChatGPT, Claude, Gemini, Copilot and Brave Leo with your target prompts, records whether your brand is cited, mentioned, recommended or invisible, and identifies citation gaps against competitors.
Open AI Visibility Fixes →