Every Mavrtr brief ships a CBO campaign structure recommendation. When we started seeing how teams actually applied those structures inside Meta Ads Manager, we noticed the same four mistakes recurring — sometimes from teams who'd been running ads for years.
Here they are, in the order they cost teams the most money.
Mistake 1 — Too many ad sets
The pattern: a team takes the three ad sets we recommend in the brief and adds two or three more "just to test." The campaign ends up with 5+ ad sets sharing the budget.
Why it costs money: CBO's algorithm needs enough data per ad set to know which one is winning. The minimum is around 50 conversion events per ad set per week. With a $50/day budget and 5 ad sets, you're splitting roughly $70 a week across five buckets. None of them clear the threshold. The algorithm shrugs and spreads the budget randomly because it has no signal.
The fix: cap CBO campaigns at 3 ad sets unless you're spending $200+/day. If you want to test more concepts, run a separate ABO campaign in parallel.
Mistake 2 — Overlapping audiences
The pattern: an interest-based ad set and a lookalike ad set in the same CBO, where the lookalike is built from a small purchaser audience and ends up significantly overlapping with the interest set.
Why it costs money: Meta will let you bid against yourself in the auction without telling you. Your CPMs creep up, your CTR doesn't, and the algorithm reads the inefficient delivery as a signal that the campaign is fatiguing.
The fix: run the Audience Overlap tool in Ads Manager before launching. If two ad sets overlap by more than 25%, collapse them or move one to a different campaign. We bake this check into the Mavrtr strategy section but stores still skip it.
Mistake 3 — Killing ad sets before they finish learning
The pattern: a team launches a CBO on Monday, sees one ad set spending less than the others by Wednesday, and pauses it.
Why it costs money: Meta's learning phase is 50 conversions per ad set. Pausing before that means you've spent money on data collection and thrown the data away before it became usable. Worse, pausing an ad set inside a CBO often re-triggers learning on the remaining ad sets — you reset all the progress.
The fix: don't pause anything for at least 7 days unless something is catastrophically broken (zero impressions, policy reject, broken landing page). The discomfort of watching an ad set underperform for a week is the price of getting clean data on whether it actually does underperform.
Mistake 4 — Mixing campaign objectives
The pattern: a single CBO contains prospecting ad sets and retargeting ad sets. The team's logic is that they're all "purchase" objective, so they should compete for the same budget.
Why it costs money: retargeting will look like it's winning every time. The audiences are smaller, the conversion rates are higher, and the algorithm will pour budget into them. You'll spend your budget on people who were going to buy anyway and starve the prospecting that's supposed to fill the funnel.
The fix: prospecting and retargeting belong in separate campaigns, with separate budgets. We default to a 70/30 prospecting/retargeting split for most ecommerce stores, but the right ratio depends on your repeat-purchase rate.
What this tells us about how teams think about CBO
The common thread across all four mistakes is treating CBO as if it's an ABO with one big budget. It isn't. CBO is a feedback loop — you're trusting the algorithm to allocate spend based on early signal, which means you need to (a) give it clean signal and (b) wait long enough for the signal to be real.
The teams that get the most out of CBO are the ones who set the structure once and don't touch it for a week. The teams that get the least out of CBO are the ones who optimize daily.
Mavrtr's brief won't stop you from making any of these mistakes after the campaign goes live. What it will do is hand back a structure that doesn't start with them — three ad sets, no audience overlap, separate budgets for prospecting and retargeting.
The rest is discipline.