A six figure budget, and the expensive part turned out to be a dropdown.

The situation

An organisation that runs personal growth programmes internationally had already committed a six figure budget behind an online summit running across three days. Sign ups were coming in. Each one just cost far more than the event could carry, and nobody could tell me which half of the spend was responsible.

The constraint

I could not trust the numbers, which meant I could not cut anything safely. Attribution was inconsistent enough that any decision to move budget would have been a guess dressed up as analysis. Underneath that, the account had been built as though everyone in it was the same person, so someone who had never heard of the organisation was being bid on with the same logic as someone already deep in its world. That arrangement quietly overpays for strangers. And this is not an audience you can rush. It is spread across countries, the decision is a personal one, and people arrive wary, so the usual urgency levers do very little.

What I did

• Started with the measurement, because everything else depended on it. Until I could see which audiences were actually producing sign ups, moving budget would have been superstition.
• Split the audience in two, strangers and people already engaged, and gave each half its own bidding logic so the cold side stopped being priced off the back of people who were already convinced.
• Rebuilt the account around that split rather than patching the one I inherited, which had been assembled for a different problem.
• Took budget off the segments that the corrected reporting showed were producing nothing.
• Then stopped treating it as a media problem at all. The ads were doing their job and landing people on a page that asked an international audience to locate their own time zone in a list of twenty to thirty entries before it would let them register. I had the page cut back until it asked for one thing.

The outcome

The budget never went up. Cost per acquisition came down by six to seven times, more people registered than before, and the ones who did were closer to the people the programme was actually built for.

What I took from it

I come back to this one whenever an account looks like it needs more money. The media was never the expensive part. What was expensive was asking a wary stranger, in a time zone nobody had bothered to detect, to solve a small puzzle before they were allowed to say yes.