The speed and mobile fixes show up in the numbers within days, because they cost you conversions on every visit. Anything that depends on a test needs enough traffic to be sure it is a result and not a good week -- for most sites that is three to six weeks per test. We will tell you which of the two you are looking at before we start.

Lower than most people assume. Split testing needs volume, but the leaks above do not -- a form asking for eight fields or a page taking six seconds is costing you whether you get a thousand visitors a month or a hundred thousand. Below roughly a thousand visits a month we fix the obvious and skip the testing, and we say so rather than billing for tests that cannot conclude.

The audit with every leak priced against your own numbers, the rebuilt pages or components, and the test record -- what we changed, what it did, and what we left alone. Plus the tracking set up properly, so the next decision is read off data rather than argued in a meeting.