Module 08 · Lesson 4 of 4 · the last lesson
Keeping Current as AI Search Changes
Everything specific in this course about surfaces and interfaces will drift. The method will not. This final lesson is about the habit that keeps the specifics honest: reading primary sources on a schedule, re-reviewing what depends on them, and refusing to act on rumour dressed as news.
The reading stack, in priority order
- Official documentation and changelogs. Google Search Central for surfaces, features and policy; the documentation of each assistant you track. When something here changes, your platform-dependent lessons change with it.
- First-party announcements. Platform blogs and verified engineering accounts for behaviour the docs have not caught up with yet. Treat as provisional until documented.
- Practitioner replication. People who show their query sets and their logs. You already know how to recognise them: they publish method, not vibes.
- Everything else. Aggregators, hot takes, "X is dead" essays. Skim monthly at most; act on none of it without replication.
Most of the industry inverts this list, acting loudest on tier four. You have spent eight modules building an instrument that outranks all of it: your own log. When a rumour claims an AI surface changed, your next run either shows it or it does not, under conditions you control. That is what the instrument is for.
The re-review schedule
Knowledge decays at different rates, so review it at different rates:
- Platform behaviour lessons (this course, Modules 2 and 7): re-check against current documentation every one to two months.
- Your measurement regime: the quarterly review already covers thresholds and drift; add "did the surfaces themselves change?" as a standing question.
- Fundamentals (Module 1): stable. Revisit when official documentation changes, which is rare and loudly announced.
Write the re-review dates into the same calendar that carries your runs. A current-awareness habit that lives in your enthusiasm instead of your calendar lasts exactly until the next busy month.
Course wrap: what you now hold
Eight modules, one system. Mechanics and intent (Module 1). How AI surfaces assemble answers (Module 2). The research instrument (Module 3). Content that gets fetched (Module 4). The entity layer (Module 5). The off-page layer (Module 6). Measurement that tells the truth (Module 7). Experiments, automation and durability (Module 8). The learners who get results from this course are the ones still running the weekly set a quarter from now, with a log that has quietly become the most valuable artefact in their search programme. Keep the cadence, keep the honesty, and the method keeps working as the surfaces move.
- Tier-four claim: "AI Overviews now ignore small sites entirely."
- You do nothing for one run cycle. Friday's run logs your six small-site citations, same as last week.
- Documentation shows no policy change; two tier-three practitioners publish logs showing the same.
- Verdict in the notes column: rumour, not replicated. Time cost: zero beyond the run you were doing anyway.
- Build the reading stack: bookmark the official documentation pages for every surface you track, and prune your feeds down to tiers one to three.
- Put the platform re-review (one to two months) and the quarterly review in the calendar, next to the runs.
- Adopt the one-run-cycle rule for acting on any claim below tier one.
- Mark this lesson complete. Then, this Friday, run the set. That is the whole course.
A blog claims a major AI surface changed behaviour yesterday. What do you do?
Why does knowledge decay at different rates?
What is the most reliable source about whether a change affects you?
Sources used in this lesson
Google Search Central: AI features and your website
Google Search Central: helpful, reliable, people-first content