Methodology: where our numbers come from, and what they don't measure

Why this page exists

A number in an SEO tool is only as trustworthy as the measurement behind it. "You rank 3rd" or "your score is 72" means something only once you know what was measured, where, when and by which rule. This page answers those questions for every kind of data in SEOLOK, and where we don't know something or can't measure it, we say so.

1. Where ranking data comes from

Positions for your tracked keywords come from a data infrastructure that measures live Google organic results. They aren't derived from a prediction model or from Search Console averages; they're read from the results page Google serves at that moment.

• Depth: Rank tracking scans the top 30 results; the free Google rank checker scans the top 50.
• Location: Each measurement targets a whole country or one of Turkey's 81 provinces. We don't measure at district, neighbourhood or coordinate level.
• Device: Mobile or desktop, measured separately, because according to Google's documentation results are served with the user's location, language and device in mind.
• Extra sections: If the page shows a knowledge panel, images, video, a featured snippet, an AI overview or a map alongside the blue links, we record it and display a note such as "Also: Images".
• Frequency: Weekly on Free, daily on Growth and Boost.

Because a keyword is always measured with the same location and device settings, the result is a comparable time series.

2. How Search Console data is imported and stored

You grant read-only access to Search Console through your Google account. After that:

1. The last 16 months of daily data is imported: query, page, country, device, clicks, impressions, CTR and average position. Sixteen months is the maximum Google makes available in Search Console.
2. New days are added every night at around 03:00. Google's API documentation says data is typically available after 2–3 days, so the newest day in your dashboard is usually about two days old.
3. Data is stored in the SEOLOK database, so days that fall out of Google's 16-month window aren't lost on our side. The archive grows from the day you connect; we can't backfill anything older than 16 months before that day.

Google Analytics 4 connects read-only through the same sign-in. The matching property is picked automatically, and sessions, users, engagement, conversions (key events), channels and landing pages are imported daily. See Search Console analytics for details.

3. How the signal score works

The signal score in the site audit is a value from 0 to 100 for every URL and for the site as a whole. Using the score and its components, pages are grouped into Push, Protect, Zombie and Watch buckets.

The rules are deterministic. The score is calculated by rules written in code, and the same data always produces the same score. Examples of signals used: CTR gap, GA4 engagement rate, content freshness and date consistency, internal link depth and inbound internal links, thin content, title-to-query match, keyword repetition, zombie page detection, author and organisation schema, and a YMYL flag.

Inspiration: the 2024 leak. In May 2024, technical documentation that appeared to belong to Google's internal "Content Warehouse" API surfaced in a public code repository and was analysed widely across the SEO community. The documents list thousands of modules and attributes. What matters just as much is what they don't contain: they don't show which attributes are used in ranking, how they're weighted, or whether they're still in use. In a statement to the press, Google cautioned that the documents could lack context, be outdated or be incomplete.

The documents also contain names of systems associated with user interaction and click data, and the existence of systems that use click data was discussed in testimony during the US antitrust case against Google. How such systems work today, or how much weight they carry, isn't public.

That's why we treat the leak as inspiration only. For example, the presence of attribute names related to content dates prompted us to write a rule that checks whether dates on a page are consistent. We don't claim that rule has a counterpart or a weight inside Google; it exists because it also makes sense for readers. For the systems Google officially describes, the ranking systems guide is the authoritative source.

The purpose of the score isn't to predict rankings. It's to tell you which page to look at first.

4. Where AI is used, and where it never decides

AI only writes text; numbers and scores are calculated in code. The summary sentences in the monthly PDF report are also generated from data without AI. A title drafted by AI is never applied to your site until you approve it.

5. Update frequencies

6. Measurement limitations

• Personalisation: Google results can vary with location, language, device and search history. Our measurements run with fixed settings and no personal history, so a signed-in user may see something different.
• SERP volatility: Results can change within a day. Google releases core updates several times a year, and volatility rises around them. A single measurement is a snapshot.
• Location granularity: We measure at province level. Results can differ between neighbourhoods of the same province.
• CrUX eligibility: The Chrome UX Report only covers sufficiently popular pages and origins. Low-traffic sites may have no Core Web Vitals field data.
• Search Console privacy and limits: Google removes searches made by very few people from the query list as anonymised queries, and applies a daily row limit on large sites. Query totals may therefore not match site totals.
• Backlink coverage: No external link index sees the whole web.

7. Security and privacy

• Google connections are read-only; we can't change anything in your Search Console or GA4 account.
• Access tokens are stored encrypted.
• Our servers are in the European Union (Germany).
• Data is never sold. Google data is used only to show you your own data.

8. How the 30-day ranking guarantee is measured

The guarantee applies to your first paid plan and rests on measurable conditions:

1. Baseline: Positions of at least 5 tracked keywords, measured within the first 3 days after payment, form the starting point.
2. Comparison: The last measurement for those keywords on days 27–30 is compared with the baseline.
3. Condition: If there's an average improvement of at least one position or more keywords in the top 10, the guarantee condition is met.
4. Refund request: If neither happened, you can request a refund between days 30 and 45.

The guarantee doesn't promise any particular position, such as number one. The full conditions are in our terms of service.

SEOLOK is built by our team, which has worked in SEO since 2018. When any method described here changes, we update this page.