Perplexity Search API Citations and Brand Mentions Monitoring for AI Search

Perplexity Search API citations and brand mentions monitoring are different products. The official Search API returns real-time ranked web results as structured data, with title, url, snippet, and optional dates, plus domain filters, languages, and regions. Perplexity's own docs say to use Search when you need raw results for your own processing. For an LLM-generated answer with citations, use the Agent API. The split is in the docs on purpose. Search is a retrieval endpoint. Agent is a generation endpoint. They answer different developer questions, and neither one answers the marketing question.

Neither API is a brand-mention monitor against what a buyer sees in Perplexity. Building a nightly Search-API scrape of your category is engineering, not a GEO dashboard, and the person who maintains the scrape is now on your payroll. The API gives you rows. It does not give you a share of voice, a citation trend, or a slide a client can read. A scrape also drifts. Perplexity changes its ranking, the API changes its fields, and your parser breaks on a Tuesday. The maintenance cost is the hidden cost, and it does not produce a report anyone outside engineering wants to look at.

If you are a developer shipping ranked links or RAG, you are in the right docs. If you are SEO asking "did Perplexity name us," you are in the wrong checkout. The two questions sound similar and they buy different things. The developer question is "give me ranked URLs I can process." The SEO question is "when a buyer typed this prompt in the consumer box, did we appear and where." The API answers the first. It does not answer the second, because the consumer UI runs its own presentation layer that the API does not reproduce.

Promptwatch paid plans replay your prompts against the Perplexity UI, daily, with mention versus citation and citation analytics across page, domain, Reddit, and YouTube. Essential is $95/mo. Explore is ChatGPT-only. Site: promptwatch.com. Perplexity has 22M+ monthly active users. The daily replay is the part that makes it a monitor rather than a scrape. The mention versus citation split is the part that makes it a report rather than a count.

When the API is the right buy

You are shipping a product that needs ranked links or you are piping results into your own RAG. Domain allowlists and denylists are documented. Multi-query search exists. That is developer infrastructure, and it is a legitimate purchase when the use case is a product rather than a report. The API is built for this. You pass a query, you get ranked results with fields you can filter and sort, and you build on top of them. That is a real product surface and a real reason to pay for the API.

Title, url, snippet, and optional dates. Languages and regions. Domain filters. Those fields are a search index return. You can count whether your domain appeared in raw results. That is not the consumer answer. Do not report it as brand visibility, because a raw result is not a footnote a buyer saw, and a ranked URL is not a citation in a generated answer. The fields tell you what the index returned. They do not tell you what the consumer UI rendered, because the consumer UI is a different surface with its own ranking and formatting.

The Agent API is the other official door: cited answers in your own product. Still not a frozen prompt list against the consumer Perplexity UI. Still not "share of voice on these 15 sales prompts." If your question is a product feature, the Agent API is the right door. If your question is a report, neither API is. The Agent API generates an answer with citations inside your app. That is a feature you ship to your users. It is not a monitoring dashboard you read for your brand.

When the question is "did Perplexity name us on this sales prompt, and whose URL did it footnote," buy monitoring. Otterly.AI includes Perplexity in the base four with lag. Peec AI can include it in three models. Profound Starter cannot, because Starter is ChatGPT-only. The lag on Otterly is the trade for the cheap price. The three-model cap on Peec is the trade for the tracker UI. The Starter gap on Profound is the trade for the ChatGPT-only entry tier.

Promptwatch's claim is real UI monitoring, not this API. Explore will not include Perplexity. Essential at $95/mo will, with visitor analytics but no listed crawler-log allowance. Professional at $245/mo adds Agent Analytics for PerplexityBot, which is the tier where a miss can be checked against a fetch log. Agency Kick-off is $199/mo with its own crawler-log allowance. Method: how we rank. Tools list. The tier progression matters. Essential gives you the UI replay and the citation split. Professional adds the crawler log, which is what turns a miss into a diagnosis. Agency Kick-off carries the same diagnosis for multi-client work.

A blocked PerplexityBot is a fetch problem. An API hit that lists your URL is a raw-result problem. Only one of those is close to "the buyer saw a footnote." The close one is still the UI replay, not the Search API. The three problems need three tickets, and conflating them is how a content team ends up rewriting a page that was never fetched. The fetch problem goes to engineering or to whoever controls the edge. The raw-result problem is not really a problem, it is just the API doing what it does. The citation problem is the one a content team can fix, and only after you confirm the fetch happened and the page was read.

Who should pick which

Pick the Search API if you are a developer building ranked-link features or RAG. Pick the Agent API if you want cited answers inside your own product. Pick Promptwatch Essential if you need daily Perplexity citation share on a fixed prompt list. Pick Promptwatch Professional if you also want to know whether PerplexityBot fetched the page. Pick Otterly if a laggy mention check is enough. Pick Peec if you want a three-model tracker UI that can include Perplexity. Skip Profound Starter for Perplexity because it does not cover the engine. The picks split by role. Developers get the APIs. SEOs get the monitor. The monitor picks split by cadence and surface, with Promptwatch on daily real UI, Otterly on laggy mentions, and Peec on a tracker UI.

FAQ

Can I use Search API hits as a citation share?

You can count whether your domain appeared in raw results. That is not the consumer answer. Do not report it as brand visibility. Citation share on a buyer prompt is mention versus citation in the UI, plus the cited domain. The API returns a row. The UI returns a footnote. They are not the same number.

Does Promptwatch wrap the Search API?

Promptwatch's claim is real UI monitoring, not this API. The UI replay is a different method from an API scrape, and it produces a different number.

What to do this week

1. If you are a developer, read the Search vs Agent API split on Perplexity's docs. Search for raw results. Agent for generated answers with citations in your app.
2. If you are SEO, skip the API key, because the API does not answer the question you typed.
3. Load 15 Perplexity prompts into Promptwatch Essential.
4. Save cited URLs, not just names. Page, domain, Reddit, YouTube when they appear.
5. Allow PerplexityBot, so a miss can be checked against a fetch log rather than guessed at.

22M+ monthly users still see the consumer UI. The Search API is raw results. The Agent API is for your product. Brand-mention monitoring is a third thing. Buy the third thing if that is the question you typed into this title.