The Lab.

My own website is the test site. Every experiment gets a written hypothesis before it starts and a screenshot after it ends — including the ones that fail. Especially the ones that fail: those are the only entries here you can be certain aren't marketing.

Running now · 1

Experiment 003 29 September 2026 RUNNING

Does rewriting a meta description change the snippet Google shows?

Google Search · Meta description · Screenshots HypothesisIf I rewrite the meta description of one article to state the exact answer, Google's search snippet will use that wording within two weeks. Google often ignores meta descriptions and writes its own snippet,… ResultNo result yet. I changed the description on 29 September 2026 and check Google every day for two weeks, until 13 October 2026. Read the experiment

Queued · 2

Experiment 002 QUEUED

Does posting weekly on Google Business Profile increase profile views?

Google Business Profile · Google Business Profile Performance · Google Analytics 4 HypothesisWeeks with one Google Business Profile post will show more profile views than the weeks without one. I expect any effect to be small, and a null result is a real possibility: Sterling Sky,… ResultNo result yet. My Google Business Profile was verified on 29 September 2026. The test starts on 27 October 2026, after a four-week settling period that isn't measured. Read the experiment Experiment 004 QUEUED

Does a caching and image fix lift a WordPress site’s PageSpeed score by at least 20 points?

Google PageSpeed Insights · WordPress HypothesisA site-wide caching and image fix on a WordPress test site will raise its mobile PageSpeed score by at least 20 points, measured as the median of five runs before and five runs after. ResultNo result yet. The test is planned to start on 8 October 2026. Read the experiment

Failed · 1

Experiment 001 1 August 2026 FAILED

Do more Google reviews improve local ranking?

Local Falcon · Google Business Profile · Google Maps HypothesisMore reviews should mean a higher map-pack position. I expected review count to be the strongest single factor. ResultWrong. Review count did not predict visibility. The clinic with 1,681 reviews appeared in the Google Maps top 3 at only 3 of my 25 test points. A… Read the experiment

Get in touch

Want this run on your site?

Same method, same honesty about what worked and what didn’t.

Start a conversation → See services

Let's connect

Chat with me Usually replies within a half hour