1. Define a manageable scope
We confirm the business, its verified website, the audience and the service area. We agree the questions and the AI interfaces to examine, then set the repeat count and capture conditions.
We size the work using a timed pilot. Adding questions, repeats or another interface changes the scope; it is agreed before the baseline is collected.
2. Preserve the observation
Each planned answer has a recorded question, interface, repeat number and capture time. We preserve the original response and a faithful transcription.
Sources shown in footnotes, links or source panels are recorded separately. A human checks that each recorded link is actually presented as a citation.
3. Count the right things
We report mentions and captures citing your verified website separately. A mention is not necessarily a citation, and a citation is not a guarantee of a recommendation.
Questions that name your business are separated from questions that ask generally for a provider. Failed or missing captures are disclosed and excluded from the valid-answer denominator.
We report a position only when the answer contains an explicitly ordered list. We do not turn ordinary prose into a ranking.
4. Turn findings into decisions
The weekly brief connects observations to useful actions: correcting an inaccurate fact, strengthening an explanation, developing a supported viewpoint or investigating a relevant editorial opportunity.
Each recommendation points to its evidence. An interesting idea is labeled an idea; it is not presented as a verified result.
5. Review before use
You approve your business facts and decide which recommendations to implement. Regulated claims require the appropriate client review.
The introductory service supplies briefs and recommendations. Sending, publishing, press-release production and implementation are separate work unless expressly included in your order.
6. Repeat the agreed measurement
The full measurement runs monthly. We compare periods only when the business details, protocol and capture conditions meet the comparison rules and both sets of captures are complete.
Changes describe the captured answers. They do not establish that a particular action caused the change.