Data Methodology: Where Every Number on This Site Comes From

The algorithm update tracker

Every confirmed update listed on this site comes from Google itself. Nothing is added on the basis of forum chatter or third-party speculation.

Aspect	Method	Measurement (verification)
Primary source	Google Search Status Dashboard incident feed	Only incidents affecting the Ranking product are recorded
Sync frequency	Twice daily, automated	New confirmed updates appear within hours of announcement
Coverage from Feb 2023	Directly from the dashboard, including the official rollout log	Start date, end date and each status message are reproduced verbatim
Coverage 2020 to Feb 2023	Manually recorded from Google Search Central announcements	The dashboard does not cover this period, so these entries are labelled as such
Update type	Derived from the official name Google gave the update	Core, spam, helpful content, reviews, link, Discover, page experience, local
Impact level	Assigned by update type, not measured	An editorial judgement, and clearly a generalisation

What the tracker cannot tell you. Google confirms only the largest changes and ships thousands of unannounced ones each year. An empty period on the tracker means Google announced nothing, not that nothing happened. That gap is exactly why the sensor exists.

The SERP volatility sensor

The sensor is original measurement rather than a reproduction of anyone else’s index. It exists because unconfirmed changes are now more common than confirmed ones.

• A fixed basket of 150 keywords across 15 categories. Why: Ten keywords per category, spanning health, finance, news, travel, shopping, local services and others. The basket is deliberately stable, because changing keywords would break comparability with earlier readings.
• One snapshot per keyword per day, taken at the same hour. Why: Results vary through the day, so a fixed collection time removes that as a variable.
• The top 10 organic domains are recorded for each keyword. Why: Positions beyond 10 move constantly and add noise without adding signal.
• Each day is compared against the previous day. Why: The score measures displacement: how far each domain moved, plus a penalty weight for domains that entered or left the top 10 entirely.
• The result is normalised to a 0 to 10 scale. Why: Raw displacement figures mean nothing to a reader. The scale is chosen so that a quiet day reads under 2 and a major shake-up reads above 8.
• Data source: the Serper API, which returns live Google results. Why: Named because you should know whose pipes the data comes through.

Current scope. United States, desktop, English. Other countries and mobile are planned but not live, and the page will say so when that changes.

What the sensor cannot tell you. It measures the web, not your site. A reading of 8 does not mean you were hit, and a reading of 1 does not mean you were not. It also cannot distinguish a Google update from a large seasonal shift in one category. Treat a high reading as a prompt to check your own Search Console, never as evidence on its own.

Recovery timelines and gap figures

The 82 day average that appears throughout the site is not a rule of thumb. It is calculated from the last eight intervals between confirmed core updates, measured from the completion date of one to the start date of the next, using the dates on the tracker above. The range across those intervals is 7 to 147 days, and both numbers are quoted alongside the average wherever it appears, because an average without its spread is misleading.

The figures update automatically as new core updates are confirmed, so a page written six months ago will show today’s numbers rather than the ones that were current when it was written.

Outcome statistics from client work

Several pages quote a distribution of recovery outcomes: roughly three in ten full recoveries, five in ten strong partial recoveries of 60 to 90 percent, one in ten stabilised, one in ten not recovered. These deserve a clear caveat.

• They come from client engagements between 2017 and 2026. Why: This is my own practice, not an industry study, and it is not a random sample.
• They are rounded proportions, not precise measurements. Why: Stated as three in ten rather than 31.4 percent, because a decimal point would imply a precision that does not exist.
• Client identities are withheld where the engagement was under NDA. Why: Screenshots on the portfolio are redacted for the same reason. Named case studies are named with permission.
• Selection bias applies. Why: Sites that reach a recovery specialist are not typical. They tend to have real revenue at stake and a drop large enough to justify paying for help.

The free tools

All the tools that accept your data process it entirely in your browser. Nothing you paste or upload is transmitted to this server or to anyone else, and nothing is stored.

Tool	Where your data goes	Measurement (what it computes)
Search Console drop analyzer	Browser only, never uploaded	Centred 7 day average, then 28 day windows either side of every candidate date
Page scope analyzer	Browser only, never uploaded	Per-page and per-section click change between two periods
Index bloat ratio	Browser only, never uploaded	Share of listed URLs under 10 impressions, plus unlisted indexed pages if supplied
Disavow builder	Browser only, never uploaded	Syntax normalisation and duplicate removal
E-E-A-T rubric and audit checklist	Browser only, nothing saved	Weighted scoring from your own answers
Date checker, countdown, timeline planner	No personal data involved	Your input compared against the tracker dates

The one exception is the alert signup, which necessarily stores an email address. That is covered in the privacy policy.

Corrections

If a date, figure or claim on this site is wrong, tell me and it gets fixed. Substantive corrections are noted on the page rather than edited away silently, because a quiet edit destroys the reader’s ability to know what changed. Use the contact page.

The editorial policy covers how content here is researched, sourced and updated.