AI Answer Surface Baseline (Onboarding)

AI Answer Surface Baseline (Onboarding)

1. 01

   Freeze the client's pre-engagement visibility record before any optimization work begins

   Capture dated screenshots and exports of current rankings, AI answer presence, and indexation state so the retainer starts from a documented baseline rather than a client's recollection.

2. 02

   Build a 25-query shortlist set from the client's core category terms and buying-intent phrases

   Split queries into branded, non-branded category, and comparison intents. Run each through ChatGPT, Perplexity, and Google AI Overviews and log whether the brand appears, is mentioned without a link, or is absent.

3. 03

   Crawl the client site for AI crawler access and structured data gaps

   Check robots.txt rules for GPTBot, ClaudeBot, and PerplexityBot, confirm schema markup coverage on money pages, and note pages that render content only via client-side JavaScript. Tools such as Alli AI and Screaming Frog surface these gaps in a single pass.

4. 04

   Map the client's entity footprint against the topics the shortlist queries imply

   List the named entities (people, products, locations, concepts) the brand should be associated with, then check which have supporting pages, schema, or third-party mentions. WordLift and Surfer both expose entity coverage gaps.

5. 05

   Score each gap by revenue proximity, not by keyword volume alone

   Rank fixes by whether the affected page sits within two clicks of a conversion path. A missing schema block on a pricing page outranks a missing meta description on a blog archive.

6. 06

   Package the baseline into a client-facing findings memo with three prioritized workstreams

   Present the shortlist absence rate, the crawl and schema gaps, and the entity coverage map as one narrative. Name the three workstreams the retainer will fund and the metric each one moves.

7. 07

   Set the re-measurement date and the exact query set in the statement of work

   Lock the 25 queries and the AI engines tested into the SOW so month-three reporting compares like with like. Without a fixed query set, AI visibility claims become unfalsifiable.