Display Campaign Underspending After a Target CPA Cut
An ecommerce retailer selling running shoes runs a Display campaign with Target CPA bidding at $22 and a daily budget of $400. Conversion tracking counts every purchase as one conversion with the same value; average order value is $85, but some orders exceed $300. Over the last 14 days the campaign spent about $105/day, recorded 38 conversions, and CPA is $19.40. The diagnostics page does not show 'Limited by budget'. The tCPA was lowered from $35 to $22 eight days ago. The account also runs Performance Max and Search campaigns with tCPA $18. A separate Display remarketing campaign with Maximize Conversions spends its full $150/day, has CPA $12, and 70% of its conversions come from users who converted in the previous 7 days. The team proposes doubling the daily budget to $800 and lowering tCPA to $18 to 'force more volume'.
1. Why is the campaign underspending while the budget is far from exhausted? Explain the mechanisms that produce this pattern.
Read the worked answer
Underspending without 'Limited by budget' means the budget is not the binding constraint; the bid strategy is. Target CPA bidding sets bids to hit $22, and eight days ago that target was cut from $35. With 38 conversions in 14 days (about 2.7/day), the system has limited conversion data and is still adapting to the tighter target. A $22 target may be below the clearing price for the inventory that actually converts, so the campaign wins only cheap impressions that it predicts can convert at or under $22 - and once it runs out of those, it stops spending rather than overpaying. That produces low spend and a CPA below target.
Also check audience and placement restrictions, device bid adjustments, ad strength, and conversion lag: purchases may be recorded days after the click, so recent spend looks unproductive. The $19.40 CPA beating the $22 target is a sign the constraint is the target, not the budget.
2. Evaluate the proposal to double the budget and lower tCPA simultaneously. What would you do instead?
Read the worked answer
Both changes push in the wrong direction and confound each other. Raising the budget to $800 cannot increase spend when the campaign already stops at $105/day for bid-strategy reasons. Lowering tCPA to $18 tightens the constraint further, likely reducing volume and pushing the campaign toward cheaper, lower-quality inventory, while also resetting learning. Making two changes at once means you cannot attribute the result.
Instead, change one lever: raise tCPA moderately, for example to $28–$32, and hold budget at $400 for two weeks. Monitor spend, CPA, and conversion volume against the 30-conversion learning threshold. If volume rises but CPA drifts above $30, tighten gradually. Separately, address the value problem: all purchases count as one conversion at the same value, so the system cannot distinguish a $40 order from a $300 order. Import order revenue and test Target ROAS or value-based bidding, which is the more durable fix for an ecommerce campaign with variable order values.
3. How would you reallocate bidding and budget across the prospecting Display, Performance Max, Search, and remarketing campaigns, given the overlap?
Read the worked answer
Start from incrementality, not platform CPA. The remarketing campaign's $12 CPA is flattered by the fact that 70% of its conversions come from users who converted in the last 7 days; exclude recent purchasers (for example, a 30-day purchaser list) and re-measure. If CPA rises sharply, much of that campaign was harvesting demand Search or PMax would have captured anyway.
For prospecting Display, keep the $400 budget and use a tCPA of roughly $28–$32 as a test, since the $22 target is limiting delivery. Do not cut Search or PMax to fund a Display budget increase that cannot be spent. Set a portfolio-level target or at least align targets so the same user is not chased at different CPAs by three campaigns. Use placement and audience exclusions to reduce overlap - exclude existing purchasers from prospecting, and exclude Display remarketing audiences from PMax if PMax is already retargeting. Judge each campaign on incremental conversions and revenue, not last-click CPA.