LEARN

What to Do in the First 90 Days

In dark space, three interlocking basal blocks support upward particles; only when all lock do particles converge into an upper dense mass, representing the foundation's critical role in scaling and review during the first 90 days
The first 90 days: 2 weeks foundation, 6 weeks scaling, 4 weeks review; scaling feedback is flawed without a complete foundation

IN ONE SENTENCE

The first 90 days: two weeks of groundwork, six of ramp, four of review — ramping early contaminates the feedback.

The first 90 days split into three phases: groundwork (about two weeks), ramp (about six), review and adjust (about four). Ramping before the groundwork is finished contaminates the feedback and invalidates every judgement after it.

OUR POSITION

Do not schedule the first two weeks as research. They have to produce verifiable results: logs proving retrieval crawlers receive complete HTML, baseline data archived with its definition, and a computed page ceiling. Missing any of the three means the next six weeks are a gamble.

01

Weeks 1–2: groundwork

Check the crawl layer and clear blockers: access status, CDN-level blocking, rendering path, and whether structured data is in server-rendered HTML.

Fix the question set and measure the baseline, archived with its definition.

Compute the page matrix, producing the ceiling and the page-type list.

Acceptance test: all three deliverables exist, and logs prove crawling works.

02

Weeks 3–8: ramp

Start with glossary and structured pages — low unit cost and enumerable — alongside a first batch of opinionated depth pages to validate the template.

Bring the technical groundwork up in parallel: column routes, internal links, structured data.

Run off-site and entity work concurrently: third-party records, description consistency, review profiles.

Acceptance test: indexing rate, and the band distribution of the first batch.

03

Weeks 9–12: review and adjust

Read band movement and cut directions sitting past thirty on high-difficulty terms.

Push page types stalled at 11–20 upward — add data, sources and quotable lines — rather than adding new pages.

Produce the first complete report: mention rate split by language, band distribution, AI-channel visits and enquiries, and movement in branded search.

⚠️ Do not use referral volume as the primary test in this phase — the absolute base is too small and too volatile.

Data behind this page

13.1 / 3.4 / 0.1 / ≈0

Monthly clicks per keyword at positions 1–3 / 4–10 / 11–20 / 21+

Source：Our keyword-level analysis of 4,074 non-branded keywords for a leading site in this category，2026-06

Sources

  1. [1]Optimizing your website for generative AI features on Google Search．Google Search Central．2026-05-15

Updated 2026-08-10