The work, opened all the way.

The situation

An ecommerce brand running a textbook funnel: awareness at the top, traffic in the middle, conversion and retargeting at the bottom. Thirteen campaigns, thirty five ad sets, two hundred and thirty five ads.

What we found

Forty five percent of the budget was going to awareness and traffic that returned 25 cents per euro. Those two layers bought 131 398 clicks and produced 23 purchases, one order for every 5 713 clicks. The two layers underneath returned one for every 107. The funnel was not broken. It was being paid for twice.

What we did

Switched the awareness and traffic layer off entirely and replaced nothing with it. Thirteen campaigns became seven, ad sets went from 35 to 14, and manual audience targeting was removed so the catalogue and the pixel do the work. Two new markets were opened as tests and both were closed when they did not clear the bar. Their cost is inside every total.

The situation

The same account ran fifteen campaign lines across two markets: three that did nothing but appear when somebody searched the brand by name, eleven built around product lines, and one new engine.

What we found

The three branded campaigns took 54% of the budget and produced 69% of the revenue, which is exactly why they had never been questioned. A brand campaign is not a growth channel, it is a toll on demand that already exists. Strip those three out of the window and what is left returned 1.26. Five of the eleven product line campaigns recorded no conversion at all.

What we did

Closed all three branded campaigns and the last product line campaigns on the same day. Fifteen campaigns became one engine per market, each fed the whole catalogue and judged on its own country. Three countries that had never been advertised in were opened one at a time, each held for a fortnight before the next was allowed to start.