Do-it-yourself game plans · No. 04 / 22
What is a realistic 90-day plan?
Last updated
Use the first three weeks to establish evidence and priorities, then implement a small number of changes and observe them repeatedly. The schedule below is an execution allowance, not a forecast of when assistants will change their answers.
| Stage | Proposed window | Main work | Required document or output | Exit condition |
|---|---|---|---|---|
| 1. Scope the decision | Days 1–5 | Agree buyer, category, competitors, questions and limits | Pilot charter; access checklist | Budget owner approves the scope |
| 2. Establish the baseline | Days 6–12 | Collect and score repeated answers | Prompt library; observation log; baseline scorecard | Collection and scoring can be explained and checked |
| 3. Diagnose the gaps | Days 13–20 | Inspect cited sources, product fit, accuracy and technical access | Citation map; accuracy register; prioritised backlog | Each proposed intervention has evidence and an owner |
| 4. Fix access and factual foundations | Days 21–30 | Resolve approved technical problems and factual inconsistencies | Tickets; product fact register; deployment log | Changes pass acceptance checks |
| 5. Improve buying information | Days 29–45 | Publish a limited content package; request external corrections | Page briefs; approved pages; outreach register | Work is published, checked and dated |
| 6. Observe and refine | Days 45–75 | Repeat comparable tests; inspect analytics and customer evidence | Second and third measurement waves | Results remain comparable, or breaks are disclosed |
| 7. Decide | Days 76–90 | Final measurement and budget review | Decision memo; exports; next-quarter plan | Continue, change, pause or stop is explicitly agreed |
Google says recrawling can take a few days to a few weeks, and requesting a crawl does not guarantee inclusion. That is a crawling timeframe—not an AI recommendation delivery promise. (developers.google.com)
Is there a smaller first step?
Yes. Before committing to the full programme, use this proposed diagnostic:
- Six important, unbranded buying questions.
- Two relevant assistant surfaces.
- Three fresh-session observations per question and surface, spread across different days.
- 36 scheduled observations in total.
Allow 6–10 staff hours for preparation, collection, scoring and a short decision note. This is an analyst planning allowance; replace it with your actual timings after the first few answers.
The diagnostic answers: “Is there enough recurring, commercially relevant evidence to justify further investigation?” It does not establish a reliable market-wide visibility percentage.
Sources cited on this page
- developers.google.com developers.google.com/search/docs/crawling-indexing/ask-google-to-recrawl?hl=en&