«Burned the budget on tests» is the most common way failure gets described. There are four mechanisms behind it, and all four are fixed by rules rather than experience.
Testing with no stop rule
You launch, watch, wait for «just a bit more», and continue to the end of the budget. The rule has to be set before launch and in numbers.
Too many variants at once
Five texts on a hundred contacts is twenty each — noise instead of data.
Testing the thing that does not decide
People spend weeks on wording when the problem is the audience. The order is the reverse: who first, what second. The first changes results by multiples, the second by percentages.
No stage-by-stage measurement
One number saying «no conversions» does not say where it broke. You need four: delivered, replied, clicked, converted.
What to do
Budget tests separately from working spend and treat that money as the price of information. A test that made no profit but produced numbers is a successful test. See how to test a setup.