RankGear.wikiThe independent, sourced reference on RankGear SEO software and the methods behind it

SEO testing

From RankGear.wiki

SEO testing is the practice of checking whether a specific change to a web page affects its position in search results, rather than relying on general advice. Search engine optimization itself is described by Wikipedia as "the practice of improving the visibility and overall performance of websites and web pages in search engine results pages"[5] . Because the leading search engines do not disclose their ranking algorithms, practitioners use controlled tests to gather their own evidence[5][1] .

Key facts. A fair SEO test changes exactly one variable while holding everything else constant, waits for indexing, and accounts for search result noise and core updates[1][3][4] .

Overview

Wikipedia notes that Google, Bing and Yahoo do not disclose the algorithms they use to rank pages[5] . Correlation studies can show which page features tend to appear on ranking pages, but RankGear's blog points out that a correlation "cannot tell you that changing the factor will move your page — for that you need a controlled test"[1] . The difference is explained in Correlation and causation in SEO. Testing has also been a theme for practitioners such as Clint Butler, who has stressed trying advice yourself to see what works[6] .

Controlled tests and A/B tests

Wikipedia describes A/B testing as a randomized experiment that usually involves two variants, and calls "A/B testing" a shorthand for a simple randomized controlled experiment[2] . It notes that A/B tests are sensitive to variance and "require a large sample size in order to reduce standard error and produce a statistically significant result"[2] . Its example is a company that sends two versions of an email, identical except for the call to action, to two groups of 1,000 customers and compares the response[2] .

RankGear's blog describes a design in which a control page and variants are built from one template and deployed one page per independent domain, so that "no single site's authority skews the comparison"[1] . RankGear's own implementation is covered in Experiments.

Isolating one variable

RankGear's blog defines a controlled SEO experiment as one in which "exactly one variable changes and everything else is held constant across independent test units"[1] . It says the control and variants should share the same template, content length, publish timing and internal linking[1] . If a title, schema and introduction all change at once, "a ranking move teaches you nothing about which one did it"[1] .

The same post names site authority as the biggest confounder: two variants on the same strong domain make it impossible to tell whether the domain or the variable moved them[1] . RankGear's experiments page adds that tests need real pages, real domains and time for indexing[7] .

Noise, volatility and core updates

Search results move even when no page has changed. RankGear's Volatility tool repeats the same search several times to measure how much the result set rearranges on its own; a calm field makes a rank move more meaningful, while in a restless field "small moves are probably noise"[3] . The maker notes that a field can be calm one week and restless the next, so the baseline should be refreshed when it matters[3] . These tools are covered in Diff, Trends and Volatility.

Larger shifts come from Google itself. Google says that "Several times a year" it makes significant, broad changes to its search algorithms and systems, which it calls core updates[4] . For anyone diagnosing a drop, Google recommends confirming the update has finished rolling out on its Search Status Dashboard and "waiting at least a full week after a core update completes" before analyzing data in Search Console[4] .

Recording and repeating results

RankGear's blog calls a single leaderboard "a data point, not a verdict" and says confidence comes from a large enough sample, replication and publishing tests that showed nothing[1] . It treats a null result as a result: "this variable didn't move rank in this design" is worth knowing[1] . It says each published result should include the method, sample size, dates, variable tested and dataset, so others can "weigh a result, replicate it, or disagree with it on the merits"[1] .

Wikipedia's account of A/B testing traces the method back to advertising, noting that Claude Hopkins used promotional coupons to test campaigns and described this in his 1923 book Scientific Advertising, though without modern concepts of statistical significance[2] . The statistics of significance and sample size are covered in Spearman and Pearson correlation.

Common questions

How do you run a fair SEO test?

Change exactly one variable and hold everything else constant, including template, content length, publish timing and internal linking[1] . Wait until the pages are indexed before reading results[1] .

What is an SEO A/B test?

It applies the A/B testing idea, a randomized experiment that usually compares two variants, to search rankings[2] . One design places each variant on its own independent domain so domain authority does not skew the comparison[1] .

How do Google core updates affect SEO tests?

Google makes broad core updates several times a year, which can shift rankings independently of any test[4] . Google suggests waiting at least a full week after an update completes before analyzing performance data[4] .

How long should an SEO test run?

No fixed duration is published in the sources reviewed, but RankGear's blog says a leaderboard built before pages are indexed "measures indexing speed, not the change"[1] . It also says repeated results are stronger than a single run[1] .

See also

References

  1. ^ a b c d e f g h i j k l m n o p q r "How we run a controlled SEO experiment". RankGear (rankgear.com). Retrieved November 2026.
  2. ^ a b c d e f "A/B testing". Wikipedia. Retrieved November 2026.
  3. ^ a b c d "Volatility". RankGear (rankgear.com). Retrieved November 2026.
  4. ^ a b c d e f "Google Search's core updates". Google Search Central. Retrieved November 2026.
  5. ^ a b c "Search engine optimization". Wikipedia. Retrieved November 2026.
  6. ^ "Clint Butler: Starting Your Own SEO Agency". Knowledge Bomb podcast (Buzzsprout), September 22, 2023. Retrieved November 2026.
  7. ^ "Experiments". RankGear (rankgear.com). Retrieved November 2026.