A useful report lets the client trace every conclusion to a dated buyer question, an observed answer, an exact source URL, and an approved product fact. This sample shows the structure and decisions we would deliver; it contains no invented client results.
The scope sheet comes first
Imagine a brand selling compact computers for home offices and photo editing. A real audit begins by agreeing on the products and markets in scope, the competing models, the platforms to check, and the buyer decisions that matter. It records who approved the product facts. A question such as “Which quiet mini PC is good for editing family photos?” tests a different intent from “What is the best mini PC?”; both need a stated market, budget, and date.
The sample scope might contain twenty agreed buyer questions divided into discovery, comparison, requirements, and objections. That is an example work package, not a claim that twenty questions are universally sufficient. The client should be able to approve the panel before any answer is captured.
- Scope: category, product variants, target country and language, platform, and observation dates.
- Question panel: exact wording, intent group, and why each question matters.
- Fact ledger: approved specifications, availability, limitations, and the source for each claim.
One observation row, with no invented result
A row in the delivered log would contain the exact prompt, platform and location setting, date and time, saved answer or screenshot reference, products named, whether the client brand appeared, every cited URL, and a note about uncertainty. If an answer changes on a repeat run, keep both observations. Do not replace the first one with the more flattering result.
For the photo-editing example, the reviewer checks whether the answer discusses graphics capability, memory, cooling or noise, ports, and the editing workload. They compare each statement with the approved fact ledger. An unsupported performance claim is marked for correction; absence from a broad answer is not automatically a publishing problem.
- Mention: was the brand or exact model named in the answer?
- Citation: did a linked page support the specific claim?
- Representation: were the model, features, market, and limits described accurately?
From source map to action list
The source map separates the brand website, retailer listings, independent reviews, and other cited pages. For each source, record what it says, its publication or update date when available, and whether it links to the right model. This prevents a team from treating every missing mention as the same problem.
The final action list should rank fixes by buyer relevance and the evidence gap: clarify an ambiguous product page, repair an outdated retailer description, add a genuinely useful comparison, or seek accurate third-party coverage where readers already research the category. Each recommendation names an owner, approval requirement, and check to run after publication.
What the client receives and what the report cannot prove
A complete handoff includes the scope sheet, prompt panel, dated answer log, source and fact ledger, issue register, prioritized page or editorial briefs, and a follow-up measurement plan. The report explains how many observations were made and which conditions may affect repeatability. It does not call a single mention a ranking or attribute an answer change to one page edit without evidence.
If you want to see how this format would apply to your catalog, send a product URL, one target market, and three buyer questions. We can scope a pilot and agree which files your team will receive.
Official references
Platform features and requirements can change. Check the current documentation when planning implementation.