How to Measure Casino Keyword Rankings

What a ranking actually measures

Two numbers are both called "position", and mixing them is the most common reporting error in casino SEO.

• Search Console position. Google defines it as "a relative ranking of the position of your link on Google, where 1 is the topmost position". The figure in the Performance report is the topmost position your link held, "averaged across all queries in which your property appeared". It is weighted by real searches, and it includes every location and device unless you filter.
• Observed rank. A rank tracker, or a person checking by hand, records where a URL appeared in one results page, for one query, at one moment, from one location and device. It says nothing about how many people saw that page.

Neither is wrong. They answer different questions: "how did real searchers see us?" and "where did we appear in this controlled sample?". A report should say which one it shows. Our guide to Search Console average position covers the first in depth.

Define the keyword set before you check anything

A keyword set (we call it a cohort) is the fixed list of queries you will measure. Decide it first, because a list chosen after seeing the rankings will flatter you.

1. One cohort per market and language. "online casino" in the United States and "online casino deutschland" in Germany are different measurements. Never pool them into one average.
2. Tag every keyword with its intent (informational, commercial, transactional, navigational), whether it contains your brand, and the product it relates to (casino, sportsbook, poker, slots).
3. Exclude what you cannot serve lawfully. If a product is not licensed for your brand in that market, its keywords do not belong in the cohort. Tracking them invites content you cannot publish.
4. Version the list. When you add or remove keywords, start a new version with a date. Compare like with like, or show the change as a break in the series.

Fix location, device and language

Google is explicit that results differ between people. Its help page lists location ("The location used for your search is displayed in the footer"), language and device ("shows the best results based on your screen type, like mobile or desktop"). It also says results can differ "depending on which data center is responding to your search", and that some information, including location, is used even with personalized results turned off.

For a casino sample this means:

• Location at the right level. Country is enough for Great Britain. It is not enough for the United States or Canada, where gambling law is set by state or province. Our market protocols give the level to use for each market.
• Device as a separate series. Report desktop and mobile separately. A single blended number hides the device where most of your traffic arrives.
• Interface language recorded. In Canada, run French and English queries for Québec as separate cohorts.
• Logged out, clean session. For manual checks, use a browser profile with no search history and say so in the method notes.

Where ranking data can come from

Be careful with home-made scrapers. Google's Terms of Service prohibit "using automated means to access content from any of our services" in violation of machine-readable instructions, and its spam policies define machine-generated traffic to include "scraping results for rank-checking purposes". In December 2025 Google sued a search-results data company, SerpApi, in federal court in Northern California over scraping. In July 2026 the court dismissed the complaint, partly with leave to amend; Google filed an amended complaint in August 2026, and the case was still open in October 2026. We do not take a legal position on any provider. We do recommend asking how a provider collects results and keeping the answer with your method notes, as our data sources page does for SpinSERP.

How often to check, and how deep

Daily checks show noise; weekly checks show direction. For most casino cohorts, weekly is enough for reporting, with daily data kept for investigating sudden changes. Always record the time and time zone of each check.

Depth matters too. In September 2025, many rank trackers had to change how they collected positions beyond the first page after Google stopped honoring a URL parameter that returned 100 results at once. Google told Search Engine Land that the parameter was "not something that we formally support". If your provider changed its method at that time, mark the date as a break in any long-term chart.

Handle missing ranks honestly

If you check the top 20 and a keyword is not there, its position is unknown, not 21 and not 101. Substituting a number distorts every average that includes it.

• Record it as "not found in top 20".
• Report coverage: the share of the cohort found within the checked depth.
• Calculate averages and medians only over keywords that were found, and print the count next to the figure.

Worked example (synthetic)

A demo cohort of five UK desktop keywords, checked to depth 20 on two dates. The numbers are invented to show the arithmetic.

Coverage was 4 of 5 at both checks. The median position of found keywords moved from 6.5 (positions 2, 4, 9, 15) to 7.5 (2, 3, 12, 17). A report should say exactly that, rather than "average rank improved" or "worsened", because the two lists contain different keywords.

Common mistakes

• Comparing a rank tracker's position with Search Console's average position as if they were the same metric.
• Pooling markets or devices into one average.
• Adding new keywords mid-series without marking a break.
• Treating "not found" as position 101.
• Reporting a position without the date, location and device it came from.

Questions

Spotted an error? Tell us with a link to the source. Corrections are logged on the page they affect (corrections policy).