Field notes

Working notes on tests, measurement, and the results that did not hold.

We write these for ourselves first. Every number has a source, every method has the workings attached, and when we got something wrong we say so in the note rather than quietly editing it.

Aug 14, 2026

Attribution said the ads returned 4,173%. The experiment on the same spend said minus 63%.

eBay ran the experiment, published it, and the gap between the dashboard and the truth was about forty times. Then I sized a holdout for a normal advertiser and found most of them cannot run one.

Measurement
10 min
Aug 13, 2026

I sized a checkout test for a client and the answer was thirty months. Most shops cannot test their checkout at all.

A 5% lift on checkout completion needs about 15,000 sessions per variant. At a thousand checkout starts a month that is two and a half years, which is longer than the checkout will exist.

Checkout
10 min
Aug 12, 2026

I ran 200,000 tests with no winner in them. A daily peek found a winner in 27 of every 100.

The false positive rate is supposed to sit at 5%. I measured what happens to it when somebody opens the dashboard every morning, and the honest answer is 27.44%. Then I found out that a bigger site does not help.

Method
10 min