Brand & Communications

AI didn't cite the publication. It cited this article.

AI engines answer your buyers' questions, and they cite their sources. When one of those sources is someone else's coverage, Sourcepath gives you the article itself: the publication, the byline, and who appears on the page.

An illustration of one third-party article that AI engines cited when answering a tracked buyer question. The article ran in Enterprise Data Review, a trade publication, under the headline "Inside the enterprise push to consolidate the customer data stack". It was published on 14 July 2026 and carries a verified byline for Elena Marsh. Sourcepath recorded it as cited when answering the tracked buyer question "Which customer data platforms are enterprises actually consolidating onto?", by ChatGPT and Perplexity. Sourcepath fetched the page and read it, and on the page itself Your Company is named and Competitor One is named. Sourcepath records which sources an AI engine cited and makes no claim about why any model chose them. The publication, the author and the environment shown are representative, not customer data.

Representation

Your competitors are named on pages you can't edit. Sourcepath checks each one for your name.

Every third-party page an AI engine cited gets fetched and read. Sourcepath looks for your brand and for each competitor you track, and reports what it found in the page's own text.

An illustration of three third-party pages that AI engines cited when answering tracked buyer questions, and who Sourcepath found named on each. A buying guide from Vantage Guides, "The 12 customer data platforms enterprise teams shortlist", handled as Lists & Comparisons. Sourcepath fetched this page and read it: Your Company is not named, Competitor One is named, Competitor Two is named, Competitor Three is named. This is a representation gap: the brand is absent and tracked competitors are present on the same page. A news article from Enterprise Data Review, "Inside the enterprise push to consolidate the customer data stack", handled as Earned Media & PR. Sourcepath fetched this page and read it: Your Company is named, Competitor One is named, Competitor Two is not named, Competitor Three is not named. A analyst document from Halberd Research, "Customer data platforms: category definition and buyer criteria", handled as Analyst Relations. Sourcepath could not read this page, so every name on it is unknown: Your Company is unknown, Competitor One is unknown, Competitor Two is unknown, Competitor Three is unknown. Sourcepath reports only what a page's own text contains and makes no claim about why any model cited it. The sources and environment shown are representative, not customer data.

None of these is a page your team can edit. What you do about each one depends on what kind of page it is.

The surfaces

A review profile is not a news article. Sourcepath tells them apart by the page, not the logo.

The sources AI cites around one buyer question are different kinds of surface, and each creates different work. Sourcepath classifies the page that was actually cited, not just the site it sits on.

An illustration of the different kinds of third-party surface that AI engines cited when answering one tracked buyer question, "Which customer data platforms are enterprises actually consolidating onto?". Each is drawn as a different shape because each is a different kind of page and creates a different kind of work. News article, handled as Earned Media & PR: What the coverage is about, and whether it opens a door for commentary or a briefing. Analyst document, handled as Analyst Relations: Which category language is used, and whether your positioning is still current in it. Review profile, handled as Reviews & Customer Proof: What themes recur in the reviews that mention you. Buying guide, handled as Lists & Comparisons: Whether the guide covers your category, and how you are described in it. Reference page, handled as Reference & Listings: Whether what it states about your company is current and accurate. Sourcepath classifies the specific page that was cited rather than the publisher it sits on, because one publisher can host more than one kind of surface. The environment shown is representative, not customer data.

Some of what these ask, Sourcepath answers from the page itself. The rest is a judgment call, and it tells you which is which.

The handoff

When the evidence stops, the work doesn't. Sourcepath hands your team the five checks.

On a page Sourcepath could read, two of these arrive already answered from the page's own text. The other three are judgment calls, and they are written for an analyst document rather than borrowed from a review site.

An illustration of how Sourcepath hands a verified third-party page over to a practitioner. The page is a analyst document from Halberd Research, "Where enterprise identity resolution is heading", cited in answers to the tracked buyer question "Which customer data platforms are enterprises actually consolidating onto?", and Sourcepath fetched and analyzed it. It then presents the analyst relations review, which has five items. Two are answered from the page's own text — Brand Presence: Your Company — named; Competitive Context: Competitor One, Competitor Two — named. Three are left open for a person, each with the question that kind of source calls for: Category Language — Which category language is used? Capability Emphasis — Which capabilities are emphasized? Positioning — Is your positioning current? Sourcepath answers only what a page's own text supports and never fills in the rest. The sources and environment shown are representative, not customer data.

The other page from that source blocks automated retrieval. Sourcepath says so, and hands over the same five checks with two of them still open.

Brand perception

Being mentioned isn't the same as being understood. Sourcepath tells you which answers actually said something.

An answer that lists you among five platforms has told your buyer nothing. Sourcepath sets those aside and works only from the answers that made a real claim about your brand, then shows you what they said.

An illustration of how AI answers characterized a brand. Of 118 tracked AI answers that named Your Company, 34 made a real claim about it; the rest named it without describing it, and Sourcepath reads only the 34. Overall, Overall: 26% positive, 62% neutral, 12% negative, from 34 characterized answers. Neutral holds the 2 answers that stated a real strength and a real concern. The overall characterization is mostly in neutral, factual terms. Known for: Enterprise identity resolution, Deterministic matching at scale, Heavy implementation lift. Strengths: Matching accuracy on messy data, Depth of the governance model. Concerns: Limited public documentation, Pricing not publicly listed. The same brand reads differently by AI platform. ChatGPT: 13% positive, 80% neutral, 7% negative, from 15 characterized answers. Claude: 55% positive, 45% neutral, 0% negative, from 11 characterized answers. Perplexity: 17% positive, 33% neutral, 50% negative, from 6 characterized answers. Gemini: not enough evidence yet, from 2 characterized answers. Sourcepath does not characterize a platform whose evidence is below its own floor. This is what the tracked AI answers themselves said, at one point in time, about this brand only. The environment shown is representative, not customer data.

That is what the answers said. Now look at the outside sources that keep appearing around the questions you track.

The landscape

The same outside sources keep turning up. Sourcepath counts how many of your buyers' questions each one appears across.

Across every question you track, Sourcepath records which sources each AI engine cited. Most appear once. A few come back again and again, and those are the ones worth a Communications team's time.

An illustration of the outside sources that recurred across a project's tracked buyer questions. Across 24 tracked buyer questions and four AI platforms. Halberd Research, handled as Analyst Relations, appeared across 14 tracked buyer questions, cited by ChatGPT and Claude and Gemini and Perplexity. Enterprise Data Review, handled as Earned Media & PR, appeared across 11 tracked buyer questions, cited by ChatGPT and Perplexity. Vantage Guides, handled as Lists & Comparisons, appeared across 9 tracked buyer questions, cited by ChatGPT and Claude and Gemini. Stackview, handled as Reviews & Customer Proof, appeared across 6 tracked buyer questions, cited by Claude and Perplexity. The Data Compendium, handled as Reference & Listings, appeared across 4 tracked buyer questions, cited by Gemini. Each count is the number of separate tracked buyer questions in which an AI answer cited that source at least once. It is not a citation volume, a score or a ranking, and Sourcepath makes no claim that any source caused or shaped what a model answered. The sources and environment shown are representative, not customer data.

You now know where you stand, and who else is in the room. What is left is the order you take it in.

The work

You don't have to work out where to start. Sourcepath hands your team the brief, in order.

For each source that matters, Sourcepath assembles what it knows, the motion the surface calls for, and the one page to open first. Then it puts the sources in the order it would take them.

An illustration of the action brief Sourcepath assembles for one third-party source. The source is Enterprise Data Review, an earned media source, and the recommended motion is Earned Media & PR. At a glance: 11 tracked buyer questions, 4 ai platforms, 7 pages identified. Why it matters: Enterprise Data Review appears across buyer questions spanning several parts of your tracked landscape, which makes how it represents you a broader question than any single article. What Sourcepath knows: Sourcepath has directly verified 4 of the 7 Enterprise Data Review pages it identified. The rest remain citation-observed. Start here: Review "Inside the enterprise push to consolidate the customer data stack" — the Enterprise Data Review page connected to the most tracked buyer questions. Sourcepath then sequences the remaining sources. Step 1, Start where the coverage is broadest, Enterprise Data Review: Open the article connected to the most tracked buyer questions and confirm how you are represented in it. Step 2, Then the guide you are absent from, Vantage Guides: Your brand is verified absent where three tracked competitors are named. Confirm the guide covers your category before anything else. Step 3, Keep the analyst source under review, Halberd Research: Its pages block automated retrieval, so representation there stays a manual check on a rhythm your team sets. The sequence is an order of work drawn from observed evidence. Sourcepath does not predict what any of it will achieve. The sources and environment shown are representative, not customer data.

Questions

What communications teams ask. About brand reputation in AI search.

How can a communications team monitor brand reputation in AI search?

Track the questions people actually ask about your brand, and Sourcepath records how each monitored AI assistant answers them — the characterization it carries, and the pages it cited to get there. Across enough answers the themes that recur become visible, with the original wording kept underneath so you can read the sentence yourself rather than a summary of it.

Which earned-media and third-party sources show up in AI answers?

Whichever ones the answers actually cite — and the useful signal is recurrence. A publication cited once across your tracked questions is noise; one cited across a dozen of them is a relationship worth having. Sourcepath groups the cited pages by destination and shows how often each comes back, so the shortlist is observed rather than assumed. See how those sources are traced.

Can Sourcepath tell us whether our brand is actually named on a cited page?

Yes, and it answers in three states rather than two: found, not found, or unknown when the page could not be read. That check is a literal read of the page's own text, not an inference — which is exactly what makes it usable in a pitch. You are not telling an editor you think you are missing from their piece; you know.

How does Sourcepath distinguish earned media from analyst, review and listing work?

Each recurring destination is routed to the marketing workstream it genuinely belongs to — Earned Media & PR, Analyst Relations, Reviews & Customer Proof, Lists & Comparisons, or Reference & Listings. A buying guide and an analyst page are both third-party sources, but they are not the same job, and they rarely belong to the same person on your team.

Can we use this to decide what to pursue first?

That is the point of it. A source carries how often it recurs, whether your brand is on it, and which competitors are, so a brief can be ordered by evidence instead of by whoever asked most recently. Sourcepath does not contact anyone or edit an outside page — it tells you which conversation is worth having, and the relationship is still yours to run. See how that picture is reported over time.

Is this media monitoring or reputation management software?

Neither. Sourcepath does not track press clippings, social mentions or review volume, and it does not suppress or manage anything. It watches one surface — what AI assistants say about you and which pages they cited — which is a narrower question than how the world is talking about you, and a more answerable one.

You already appear in answers your buyers are reading. See how you're described, and what to do about it.