Google subjects a website to roughly NINE distinct evaluation passes — crawl fetch, rendering, index selection, canonical selection, query-time ranking, scheduled re-crawl, freshness-driven re-crawl, broad-core-update re-evaluation, automated spam/quality system passes

Claim: A site is not "looked at" once. It is subject to nine distinct evaluation passes (per-URL = applies to each URL; site-wide = applies to the domain/host; one-time = largely once per URL; recurring = repeats for the life of the site):

1. Crawl fetch (Googlebot HTTP request). Per-URL; recurring (first fetch is one-time, but re-crawls repeat). Gated by robots.txt/reachability.
2. Rendering (WRS / headless evergreen Chromium). Per-URL; recurring (re-rendered on re-crawl). All 200-status pages are queued.
3. Index selection / quality evaluation. Per-URL decision, but heavily influenced by site-wide quality signals; recurring (re-evaluated on re-crawl, and a page can be dropped later).
4. Canonical selection (clustering + ~40 signals). Per-URL/per-cluster; recurring (can change as content/signals change).
5. Query-time ranking evaluation. Per-URL × per-query; recurring on every search — the only pass that runs continuously and combinatorially (see Query-time ranking is the only continuous, combinatorial pass — every search re-computes per-URL × per-query).
6. Scheduled re-crawl (importance-based). Per-URL; recurring. Cadence set by importance + observed change rate.
7. Freshness-driven re-crawl (lastmod / change demand). Per-URL; recurring, event-triggered by detected/declared change.
8. Broad core update re-evaluation. Site-wide and per-URL (re-weighting of which content best satisfies queries); recurring, several times a year.
9. Automated spam/quality system passes (e.g., SpamBrain and quality classifiers; the hostAge/fresh-spam mechanism sits here conceptually). Site-wide; recurring/continuous. Applies from the start, and is the pass most relevant to the "new-site caution" effect.

Bottom line: a single page is "judged" at minimum five distinct ways before it can rank (fetch, render, index-selection, canonical, ranking), and then re-judged continuously by at least four recurring systems (scheduled re-crawl, freshness re-crawl, core updates, spam/quality systems) plus a fresh ranking computation on every query. There is no single "Google looked at my site" moment.

Source: Synthesised from the compass_artifact research document and Google's public documentation.

Confidence: High on enumeration (each pass is documented). Categorisation as "9 distinct" is a useful framing rather than an official Google number.

Caveat: This is the most useful single deliverable in the brief for client conversations. Reframes "did Google review my new pages yet?" from one-shot to plural.