How do I track my ranking position in Perplexity? What we measure
A practitioner view on how to track your ranking position in Perplexity: mention position, citation position, crawler access and how we report it to clients.
Clients ask for their "Perplexity ranking" as if it were a keyword position. It is not, and the first job is to reset that expectation. Perplexity writes an answer, names brands in it, and lists numbered sources. There are two positions to track, and they move independently.
Mention position versus citation position
Mention position is where your brand sits among the brands named in the answer text. Being named first is a stronger outcome than being named fifth. Citation position is where one of your URLs appears in the numbered sources. You can be cited without being named, and named without being cited. Each tells a different story. A citation without a mention often means your content informed the answer but a competitor took the credit. A mention without a citation often means the model knows you from third-party sources, not your own pages.
We report both per prompt group, not as one blended rank.
Build the prompt set before you track
Pick prompts by buyer stage. Early questions are broad ("how to pick a payroll provider"), middle ones compare ("X versus Y"), late ones are brand-specific. In Promptwatch we organize these with topics, tags and prompt types, and personas with persona stacking when the buyer type changes the answer. Prompt Explorer proposes candidates from Search Console, ranking keywords or People Also Ask.
Position is noisy, so we need enough prompts and enough runs to see a trend. Promptwatch runs panels daily per model, country and language, and reports position in the answer alongside Visibility Score and Share of Voice. An A/B period comparison makes before and after changes readable.
Why position drops, and how we diagnose it
We follow the same order every time.
- Competitor change. Did a rival gain a citation? Competitor heatmaps and citation analytics show it.
- Page change. Did one of your cited pages stop being cited? Page Tracker watches specific URLs, ours or a competitor's.
- Access. Did PerplexityBot or Perplexity-User hit errors on the page? Agent Analytics shows IP-verified crawler logs with failure explanations. Crawler logs start on the Professional plan at $245 a month, and Essential does not include them.
- Third-party sources. Did the forums, review sites or articles Perplexity cites change? Offsite mention tracking and the Reddit, YouTube and LinkedIn citation reports help.
Only after those four do we touch content.
Connecting position to traffic
Position matters because of visits. Visitor analytics measures Perplexity-referred sessions and conversions through a lightweight script, the GTM template or the WordPress plugin. Without it you are reporting a position nobody can value.
Plan guidance
Explore, the free plan, tracks ChatGPT only, so it will not show Perplexity. A paid plan is required: Essential at $95 for 50 prompts is a workable start for a single site, and Professional at $245 adds crawler logs. Agency plans from $199 give unlimited projects and prompts and ten seats.
Honest limits
Perplexity can change how it builds answers without notice, and results differ by phrasing and location. So we avoid promising a position. We promise a measurement method, a diagnosis order and a reporting rhythm: weekly checks for movement, monthly reviews for decisions.
Quick start
Write ten prompts. Run them in Perplexity by hand and record both positions. Then load the same ten into a trial of Promptwatch and compare your sheet with the tool's output. If they match, you can trust the tool to do the daily work and save your time for the fixes.
What a good monthly report looks like
We keep it short. One page for the headline: mention position and citation rate by prompt group, against the same competitors as last month. One page for the three biggest movers, with the diagnosis order above applied to each. One page for actions, each with an owner and a page. The Action Board in Promptwatch helps here because it ranks suggested actions and lets us assign them, and scheduled PDF reports mean the numbers reach the client without anyone rebuilding a slide deck.
The main discipline is consistency. Do not change the prompt set midstream, do not switch what counts as a mention, and write down every site change. Position in an AI answer is noisy enough without adding measurement drift on top.