Skip to content
SourcepathBook a Demo

Reporting

The briefing leadership reads. And the evidence it was built from.

Sourcepath turns the AI answers it recorded into reporting made for the people who read it — an executive briefing, and a working view for the marketing team. Every interpretation is built from facts the platform has already established, and each one stays inspectable.

Book a Demo
An illustration of a Sourcepath executive briefing and the evidence it was built from. The briefing reads: "Your Company is named in 70% of tracked AI answers, while your own domain is cited in 40% of them — the widest gap between being described and being used as a source. AI assistants describe the company as an enterprise identity resolution platform built on semantic and RDF technologies, with no stated concerns. Third-party sources appear alongside the brand across four tracked categories, most often through reviews and customer proof." Beneath it, the complete set of 7 facts Sourcepath had already established for this project. The briefing drew on 4 of them — AI Visibility, Citation Coverage, How AI Describes the Brand, Third-Party Presence — and the remaining facts were available but not used: Sourcepath Score, Brand Strength — semantic reasoning, Brand Strength — visual modeling. Synthesis can only draw on this set; a claim citing anything outside it is discarded, and the briefing is replaced by a plain summary Sourcepath composes itself. Which facts were drawn on is recorded; the wording of individual sentences is not attributed, and Sourcepath makes no claim about how any AI model reasoned. The environment shown is representative, not customer data.

Measurement integrity

Sourcepath will tell you what changed. And when it has no basis to say.

A measurement is a set of buyer questions asked of a set of AI engines, and that set moves. Sourcepath compares only the questions both measurements actually completed — so a number never moves because the ground under it did.

An illustration of how Sourcepath decides whether it may report movement. Twelve tracked buyer questions are shown across two measurements. 8 were completed in both and are the only ones any movement is drawn across. 3 were added since the previous measurement and 1 was not completed this time, so all 4 sit outside the comparison. Movement is drawn across the 8 questions both measurements completed. The other 4 sit outside the comparison, and Reporting leaves them there. Where no overlap exists at all, Reporting draws nothing and states why: on Sourcepath Score trend it says, "Reporting baseline established at a Sourcepath Score of 58. Comparable movement will appear as additional measurements accumulate." One trustworthy measurement exists. A trend needs two. on What changed it says, "No new appearances, disappearances or meaningful citation shifts since your last comparable measurement." Nothing moved that Sourcepath could verify against the previous measurement. on Across all tracked prompts it says, "Based on 43 of 60 intended AI responses. Gemini 0 of 15. Perplexity 13 of 15." Two engines returned less than Sourcepath asked for, so the shortfall is named rather than averaged away. This is a like-for-like check, not a statistical test: an overlap either exists or it does not. An empty trend line is a result, not a gap. It means Sourcepath had nothing it could honestly compare — and it will tell you exactly what changed to make that true. The environment shown is representative, not customer data.

Movement is drawn across the 8 questions both measurements completed. The other 4 sit outside the comparison, and Reporting leaves them there.

  • Sourcepath Score trend

    Reporting baseline established at a Sourcepath Score of 58. Comparable movement will appear as additional measurements accumulate.

    One trustworthy measurement exists. A trend needs two.

  • What changed

    No new appearances, disappearances or meaningful citation shifts since your last comparable measurement.

    Nothing moved that Sourcepath could verify against the previous measurement.

  • Across all tracked prompts

    Based on 43 of 60 intended AI responses. Gemini 0 of 15. Perplexity 13 of 15.

    Two engines returned less than Sourcepath asked for, so the shortfall is named rather than averaged away.

An empty trend line is a result, not a gap. It means Sourcepath had nothing it could honestly compare — and it will tell you exactly what changed to make that true.

Buyer question performance

The number is where leadership starts. The questions underneath are where the work is.

The same report resolves into the individual buyer questions Sourcepath tracks. For each one: whether your brand was named, how many engines actually answered it, and what Sourcepath observed about the sources those answers used. Where a recommendation is already open, the row is the way into it.

A table of the individual buyer questions underneath a Sourcepath report. Each row is a question the team tracks, with whether the brand was named in the answers, how many AI engines completed it, and what Sourcepath observed about the sources those answers used. "Which platforms handle enterprise identity resolution best?" — brand named, 3 of 3 engines answered. Your domain is cited in 2 of the 3 answers. No recommendation is currently open for this question. "How much does an enterprise knowledge graph cost?" — brand named, 2 of 3 engines answered. Neither answer that named you cites your domain. A recommendation is open for this question. "What is a semantic layer?" — brand not named, 3 of 3 engines answered. Competitor One is cited in 2 answers. Your domain is cited in none. A recommendation is open for this question. "How do knowledge graphs improve RAG systems?" — brand named, 3 of 3 engines answered. Your domain is cited in 1 of the 3 answers. A recommendation is open for this question. "How do organizations ground LLMs with enterprise knowledge?" — brand not named, 2 of 3 engines answered. Neither answer names your brand, and neither cites your domain. A recommendation is open for this question. Every row is a question your team chose to track, in the wording your team chose. The way into the work only appears where Sourcepath already has a recommendation open — never as an invitation to go and find one. None of this ranks the brand, and citation position is never treated as performance. The environment shown is representative, not customer data.

Every row is a question your team chose to track, in the wording your team chose. The way into the work only appears where Sourcepath already has a recommendation open — never as an invitation to go and find one.

Executive and Marketing

The same evidence. Two documents that end differently.

Executive and Marketing are separate compositions over one body of established intelligence — not a dashboard with a toggle. Some of what they show is literally the same surface. What separates them is where each one is built to finish.

A comparison of the two finished documents Sourcepath Reporting produces from one body of established intelligence. The Executive document answers "What should leadership understand?". It contains: Executive summary, Across all tracked prompts (shared with the other document), Sourcepath Score trend, Competitive position, How AI understands your brand (shared with the other document), Leadership takeaways. It ends in a conclusion: What's working — Your brand is named in most tracked answers. Your Company appears in 70% of tracked AI responses across the engines Sourcepath measures. What to watch — Your own domain is cited far less often than your brand is named. The brand appears in 70% of tracked answers while your content is cited in 40%. That gap is the clearest opportunity currently available. Where to look next — External sources appear alongside your brand more often than your own pages do. Third-party sources span 4 tracked categories and 17 questions, most often through reviews and customer proof. Deterministic, and drawn from the same facts the executive summary was built from — never a second interpretation. The Marketing document answers "What should we work on next?". It contains: Across all tracked prompts (shared with the other document), What changed, Provider coverage, Buyer question performance, Content opportunity mix, How AI understands your brand (shared with the other document), Marketing priorities. It ends in a queue of work: New content — Build coverage for 7 buyer questions Seven tracked questions have no owned page behind them strong enough to answer them. New content is the clearest opportunity on each. Content optimization — Strengthen 3 existing pages Three tracked questions already have an owned page behind them, each carrying an open recommendation to improve it. Third-party presence — Increase presence on G2 G2 appears alongside your brand across 7 tracked questions, more than any other external source Sourcepath classifies. Six fixed levers, each gated on real evidence. A card only appears when the evidence for it exists, and none of them promises what the work will produce. Marked surfaces are the same component in both documents, reading the same evidence. Neither document derives a fact of its own. They compose what Sourcepath already established, and they stop in the two different places their readers actually need to arrive at. Neither document is generated from raw data independently, and neither claims what any recommended work will produce. The environment shown is representative, not customer data.

Marked surfaces are the same component in both documents, reading the same evidence.

Neither document derives a fact of its own. They compose what Sourcepath already established, and they stop in the two different places their readers actually need to arrive at.

Questions

What teams ask about AI search analytics. Measurement, and what makes a comparison fair.

What is AI search analytics?

It is measurement built on recorded AI answers rather than on impressions or positions. Sourcepath keeps every scan of every tracked question, so visibility, citations and the competitive picture can be read at a point in time and read again later — with the counts behind each figure kept alongside it.

How do we measure AI-search visibility over time?

Each scan produces a reading: the share of that project's tracked question and engine observations in which your brand appeared. Scans are kept rather than overwritten, so two points in time can be compared directly, and every percentage is shown with the readings that produced it rather than on its own.

Can Sourcepath report on mentions, citations and competitors?

Yes, and they are kept as separate measurements rather than blended into one number. Being named in an answer and having your page cited by it are different facts, they move independently, and a single score would delete the difference at exactly the moment it became useful. See how those observations are recorded.

How does Sourcepath compare one reading with another?

Only over the questions that completed in both scans. If a question was added, retired or did not run, it is excluded from the comparison rather than quietly counted as a change — and where there is no comparable set at all, Sourcepath will not draw a trend line. A movement you can trust is worth more than one you cannot.

Can leadership see this without losing the evidence underneath?

Yes. The executive view summarises visibility, citation coverage and competitive position, and every figure in it stays connected to the tracked questions, answers and cited pages it came from — so a number can always be opened rather than taken on trust. See the practitioner view of the same evidence.

Now see what Sourcepath finds about your brand.

Book a Demo