See which sources the engines cite
The Citations tab lists the domains the engines cited while answering your tracked prompts, how often each was cited, and whether it is yours. Those sites are where an answer about your category is being formed, so being present on them is how answers start including you. Visibility and share of voice work without this tab; if you have not run a check yet, start with See whether AI assistants recommend you.
Before you start
Citations need Perplexity
MarqIQ records the sources an engine returns alongside its answer. Perplexity is the engine that does so, which is why the Connect AI engines list badges it powers Citations. An engine that returns no sources cites nothing, and this page says so rather than inventing any.
Connect a Perplexity key before the run you want citations from. Sources are recorded as answers come back, so a run that did not ask Perplexity has none to show.
Fill the page
Click Connect AI engines on the AEO Overview tab.
Click Open key page beside Perplexity.
It opens www.perplexity.ai/account/api — Pro plans include monthly API credits.
Paste the key into the field beside Perplexity.
Click Verify & connect.
Click Run visibility check on the Overview tab.
Citations are recorded during a run. Connecting alone fills nothing.
Open the Citations tab once the run finishes.
Read the table
| Column | What it holds |
|---|---|
| Domain | The site, without its www. Subdomains are counted under their own name. |
| Cited | How many times that domain was cited across the run's answers, with the most-cited first. |
| Relationship | Your site where the domain is your own or a subdomain of it, Third party for everything else. |
| An icon that opens the domain in a new tab. |
Only http and https addresses are counted; anything else is dropped rather than shown as a blank row.
Below the table is the line that says what the counts rest on: the date and age of the run, how many prompts it asked, and how Perplexity fared in it — "Perplexity answered 18 of 20 calls; 2 failed". Where prompts have been removed since, it also says how many of the run's answers are still stored and therefore still counted.
Take care
Where your own site is not among the cited sources, the page says so and points at the Recommendations tab. That is not a small finding: it means the pages an engine reads about your category are all somebody else's.
Note
At most 50 sources are kept per answer, first ones first, and a duplicate address within one answer is counted once.
What to do with the list
- Treat the top third-party domains as an outreach list. A review site, directory or publication cited repeatedly is where an answer about your category is being assembled.
- Check which of your own pages are cited, and write more of that kind. A single cited page tells you what the engines found usable.
- Compare the list against the Recommendations tab's Outreach items, which are drafted from these same domains, with the most-cited URLs on each recommendation.
- Re-check after a campaign. Citations are recorded from the run that was asked, so a new list needs a new run.
If it did not work
"None of the connected engines disclose their sources."
Why: Perplexity is not connected, and no other connected engine returned sources with its answers.
Fix: Connect your own Perplexity key through Connect AI engines on the Overview tab, then run a visibility check. Visibility and share of voice keep working without it.
"The latest visibility run did not ask Perplexity — it was connected after that run."
Why: The key went in after the run, so that run's engine list never included it.
Fix: Run a new check. The next run records the sources it cites.
Perplexity did not answer in the latest run
Why: Its calls failed — no credit, a quota refusal, a throttle or a timeout. The page names the reason it recorded.
Fix: Read the amber banner on the Overview tab, which carries the action for that reason. Why one AI engine returned nothing, or half your prompts were not asked covers each one.
"Perplexity answered … without citing any web sources."
Why: The answers genuinely carried no sources. That is a fact about those answers, not a fault.
Fix: Nothing to fix. Prompts that ask for a comparison or a shortlist draw more sources than an abstract question does — Choose the questions to track covers how to phrase them.
"That run predates the Perplexity API update and could not read its sources."
Why: The run was measured before the reader that reads Perplexity's current source format landed, so its sources were never recorded.
Fix: Run a new check. Nothing can recover sources from an older run.
"Perplexity is not connected now, so the next visibility run will not add to these sources."
Why: The table is from a run that asked Perplexity, and the key has been disconnected since.
Fix: Reconnect the key if you want the list to keep updating. The figures already on screen stay, because they belong to the run that produced them.
The tab shows an upgrade panel
Why: Citation analysis is part of AEO Advanced, which is on Enterprise.
Fix: Click Try it free for 7 days on that panel, or move to Enterprise. One trial per feature per workspace, ever.
Next steps
Related articles
Was this article helpful?