Skip to content
SourcepathBook a Demo

AI Search Intelligence

AI search doesn’t hold still.

Sourcepath runs the buyer questions you track against ChatGPT, Claude, Gemini and Perplexity, scan after scan — so where your brand stands in AI answers, and how that moves, is measured rather than guessed at.

Book a Demo
A Sourcepath Visibility Trend for a representative brand, measured across six scans between February 3 and April 14. Visibility is the share of completed AI responses in which the brand was named. ChatGPT moves from 38 percent to 61 percent, rising overall with a dip to 48 percent at the fourth scan. Claude moves from 44 percent to 57 percent, flat early and improving late. Gemini moves from 26 percent to 39 percent and is the lowest at every scan. Perplexity moves from 52 percent to 47 percent, fluctuating and ending below where it began. Two Sourcepath observations are marked: a citation gained on ChatGPT at the third scan, and a visibility decrease on Perplexity at the fifth scan. Each point is one scan; Sourcepath does not measure between scans and the values are never interpolated. The environment shown is representative, not customer data, and nothing here claims Sourcepath caused any change.

24 tracked buyer questions · 4 providers · 6 scans

24 tracked buyer questions · 4 providers · 4 most recent scans

Source evidence

Your brand is in the answer. Your domain isn’t in the sources.

Sourcepath records every citation attached to the responses for a tracked buyer question, resolves each cited domain, and classifies it as yours, a competitor’s, a recognized third-party source, or unresolved. Aggregate visibility can’t tell you which of those you are.

The citations attached to the AI responses for one representative buyer question, “Which customer data platforms handle enterprise identity resolution best?”, pooled across four providers. Your Company is named in the responses to this question. 14 citations were recorded, across 13 distinct domains — one third-party domain, cdpbuyersindex.com, was cited in two different providers’ responses, which is why there are more citations than domains. Three citations point to competitor domains: competitor-one.com, competitor-two.com and competitor-three.com. Eight point to recognized third-party sources: cdpbuyersindex.com, enterprisemartechreview.com, datastackreview.com, martechbuyersguide.com, identitydataweekly.com, cdpselectionhandbook.com and b2bsoftwareguide.com. Three are unresolved, meaning Sourcepath could not match them to a known destination or could not resolve a provider redirect to one: nordicdataadvisory.com, datapractice.blog and dataadvisorypartners.com. No citation to the brand’s own domain appeared in any of the responses to this question. The citations are an unordered set; the order they are listed in carries no meaning. The environment shown is representative, not customer data.

Which customer data platforms handle enterprise identity resolution best?

Your Company is named in the responses to this question.

  • cdpbuyersindex.com

    third-party · 2 citations

  • enterprisemartechreview.com

    third-party

  • competitor-one.com

    competitor

  • datastackreview.com

    third-party

  • martechbuyersguide.com

    third-party

  • competitor-two.com

    competitor

  • identitydataweekly.com

    third-party

  • nordicdataadvisory.com

    unresolved

  • cdpselectionhandbook.com

    third-party

  • competitor-three.com

    competitor

  • datapractice.blog

    unresolved

  • b2bsoftwareguide.com

    third-party

  • dataadvisorypartners.com

    unresolved

  • Your domain0
  • Competitor3
  • Third-party8
  • Unresolved3

Fourteen citations across four providers. Not one of them yours.

Representative buyer question · citations pooled across four providers · The environment shown is representative, not customer data.

Provider attribution

One visibility number. Four engines that look nothing alike.

Sourcepath measures every provider separately. This is where one brand’s appearances actually came from in its latest scan, and what each engine looks like when you inspect it on its own.

A ring showing where a brand’s AI appearances came from, by provider, in its latest scan. Of 60 appearances across four providers: ChatGPT, 30 percent, appeared in 18 of 24 completed responses, own domain cited in 3 of 24, up 16 percentage points since the previous scan, most-cited competitor domain competitor-one.com; Claude, 27 percent, appeared in 16 of 24 completed responses, own domain cited in 2 of 24, up 12 percentage points since the previous scan, most-cited competitor domain competitor-two.com; Gemini, 20 percent, appeared in 12 of 24 completed responses, own domain cited in 1 of 24, up 12 percentage points since the previous scan, most-cited competitor domain competitor-three.com; Perplexity, 23 percent, appeared in 14 of 24 completed responses, own domain cited in 1 of 24, up 9 percentage points since the previous scan, most-cited competitor domain none in this representative example. The ring shows each provider’s share of total appearances, which is a different measure from a provider’s own visibility rate. Sourcepath records which citations a provider returned, not how a model used them. The environment shown is representative, not customer data.

Share of appearances

60 appearances latest scan

ChatGPT

Appeared in
18 of 24 completed responses
Own domain cited in
3 of 24
Since previous scan
+16 pp
Most-cited competitor
competitor-one.com

Latest scan · 24 tracked buyer questions · 60 appearances across four providers

Representative provider evidence · competitor names are representative · The environment shown is representative, not customer data.

Prioritized action

Hundreds of observations a scan. Three things worth doing.

Sourcepath runs the same rules on every scan and attaches each piece of evidence to the finding it supports. What surfaces here is what that evidence is asking for, and the evidence stays attached to it.

The 416 observations recorded in the latest scan, drawn as one mark each. The 62 that support the three findings below are lit. Their arrangement carries no meaning — the field is not ordered, ranked or scaled.

  1. Competitive

    Address the questions where a competitor’s domain is cited and yours isn’t.

    31 observations recorded a competitor domain among the citations for a tracked question with no citation to your own domain.

    High priority 31 supporting observations 9 tracked questions

  2. Visibility

    Recover the three questions where your brand stopped appearing.

    4 observations compared this scan against the previous one and found your brand named before and absent now.

    High priority 4 supporting observations 3 tracked questions

  3. Technical

    Add FAQ schema to the 22 pages that don’t have it.

    27 observations compared pages cited for your tracked questions against your own — the cited pages carry FAQ schema where yours does not.

    Medium priority 27 supporting observations 22 pages

416 observations recorded in the latest scan · 62 of them support the three findings above · 24 tracked buyer questions

Representative findings · Priority is computed from the supporting evidence · The environment shown is representative, not customer data.

Questions

What teams ask about AI search monitoring. Direct answers, grounded in what Sourcepath records.

How does AI search monitoring work?

You choose the questions your buyers actually ask. Sourcepath runs them against the monitored AI engines on a repeating schedule and stores each answer exactly as it came back, together with the pages that answer cited. Because every scan is kept, a question is a dated reading you can return to rather than a one-off look.

Which AI engines can Sourcepath monitor?

ChatGPT, Claude, Gemini and Perplexity. Each one is asked the same tracked question, and each answer is recorded separately — which matters, because the four frequently disagree about who to name and what to cite. A blended score would hide exactly the difference worth acting on.

Can Sourcepath track brand mentions, citations and competitors together?

Yes. For every answer Sourcepath records which organisations were named in the text, which pages were cited, and how your own domain featured in either. Naming is a match on a brand's name or domain in the answer itself, so what you get is an observation rather than an inference about what the model believed. See how the competitive picture is built.

How do teams measure AI-search visibility over time?

Visibility is the share of a project's own tracked question and engine readings in which your brand appeared, in a given scan. Because scans are kept rather than overwritten, the same questions can be read again later and compared like for like — and every figure carries the readings behind it, not a percentage on its own. See how those readings are reported.

What does Sourcepath not tell us?

Why a model answered the way it did. Sourcepath observes what came back and what it cited; it has no access to a provider's ranking or retrieval logic, and it will not present a guess as a measurement. Everything on this page is something that was recorded.

Now see what Sourcepath finds about your brand.

Book a Demo