SEO Testing Overview

SEO Testing lets you run controlled experiments on a client's pages: change one thing (a title tag, a meta description, an H1), wait a set period, and compare Google Search Console performance before and after the change. Every test records what was on the page (the control), what you changed it to (the variant), and the click, impression, CTR, and ranking-position deltas between the two windows.

Use it when you want proof that an on-page change helped, or an early warning that it hurt, instead of guessing from a traffic chart.

How an SEO A/B test works

An SEO test is a time-based before/after comparison rather than a split-traffic test. Search engines show one version of a page to everyone, so you can't send half your visitors to a variant the way a conversion-rate tool does. Instead:

1. RankNest captures a baseline snapshot of the page's Search Console performance (clicks, impressions, CTR, average position) while the original version is live.
2. You publish the change on the actual site and start the test.
3. When the test ends, RankNest pulls the same metrics for the test window and compares them against the baseline.

Because the comparison spans time rather than traffic, keep one rule in mind: change only the variable you're testing. If you rewrite the title and also overhaul the page content mid-test, you won't know which change moved the numbers.

Where to find it

Open a client and click SEO Testing in the sidebar. The view has two tabs in the top bar:

• Tests: the working view, with a left rail of tested URLs and a detail pane showing the selected page's experiments and click history. See Managing Tests.
• Results: the roll-up dashboard, covering cumulative impact, win rate over time, and which variables win most often. See Reading Results.

The dashboard header also shows a four-stat strip for the client: Running, Wins, Win Rate, and Avg CTR Lift.

Prerequisite: connect Google Search Console and assign a property to the client. See Google Search Console. Without it, tests still record your control and variant values, but RankNest can't pull clicks, impressions, CTR, or position data, so there is nothing to measure results against.

Test lifecycle

Every experiment moves through a fixed set of statuses:

1. Draft: the test exists but nothing has changed on the site. You can edit the control and variant, sync the baseline, and re-crawl the page freely.
2. Running: you clicked Start Experiment after publishing the change. The baseline locks at this point so the "before" snapshot can't be overwritten.
3. Completed: the test ended, either because you clicked End Experiment or because RankNest auto-ended it after its end date (checked nightly). Ending a test pulls the variant-period metrics from Search Console. A completed test can't be set back to running.

Two more statuses exist alongside these, Paused and Cancelled, and a test in either state shows that label on its card. The editor's Status dropdown, which is where you move a test between them, appears only for tests that are already paused or cancelled. On a draft, a running test, or a completed test, the status is shown as read-only text instead.

What gets measured

For each completed test, RankNest compares the baseline and test windows on:

• Clicks: total clicks to the page from Google Search.
• Impressions: how often the page appeared in results.
• CTR: click-through rate, with the change shown in percentage points.
• Position: average ranking position for the page's queries.

Beyond the aggregate numbers, the experiment editor includes a Query Performance panel showing which individual search queries were retained, gained, or lost between the two windows. See Reading Results. If the client also has Google Analytics 4 connected, test cards show a Key Ev. Δ figure (GA4 key events on the page over the same window) as context. It is not attributed to the test and plays no part in deciding whether a test won.

Tips

• Test pages that already get impressions. A page with no Search Console history produces nothing to compare against.
• Give tests a full window. The default duration is 30 days; low-traffic pages produce noisy results on anything shorter.
• Search Console data lags by two to three days, so freshly synced numbers may trail what happened on the site this week.
• Write a hypothesis when you create the test. When you review results a month later, "shorter title with the city name should raise CTR" is far more useful than a blank field.

Related

• Creating Tests
• Managing Tests
• Reading Results
• Google Search Console