Fewer campaigns, more profit

Challenge

Profitable on paper, wafer-thin in practice

Alongside Shopping, this webshop ran a set of search campaigns that largely overlapped: the same product categories, the same search terms, separate budgets and separate learning processes. Together they accounted for around EUR 1,700 in monthly spend.

At first glance it worked: the set was profitable. But POAS stood at 1.3, and break-even is 1.0. Of every euro of ad budget, only 30 cents of gross profit remained. Net, the webshop kept around EUR 550 per month. One disappointing month and it tips into a loss.

Approach

One structure, profit as the steering signal

In mid-December the campaign set was merged into a single structure, with profit-driven bidding as the engine. Conversion value in this account is gross profit: product costs are already deducted before the algorithm ever sees it. So target ROAS here doesn't steer on revenue with thin margins, but on what's left at the bottom line.

From that moment, all conversion signal landed in a single learning process, every keyword got one place and one bid, and budget could move freely to the categories that deliver. No extra budget, no new ads. Just structure.

Result

2.7 times as much net profit, on the same spend

In the quarter after the consolidation, POAS rose from 1.3 to 2.1. Net profit after ad spend went from around EUR 550 to almost EUR 1,500 per month. The same traffic, the same products, and a structure that finally let the algorithm do what it's good at.

Best practice

Why consolidation works

Smart Bidding learns per campaign. Every campaign you add cuts the available conversion signal into smaller pieces. Two campaigns on the same traffic means two half learning processes that both respond slowly to changes in the auction, the season and search behavior.

Google preaches this philosophy itself under the name Hagakure: as few campaigns as possible, as much data per decision as possible. In practice, almost every account moves the other way. A campaign gets added for the promo week, one for a new product line, one because a colleague found it easier to oversee. Five years later the account is bidding against itself and budget is stuck in silos, while the algorithm should be steering it toward the winners.

Consolidation turns that around: structuring the account for what the algorithm needs, not for whatever grew over the years.

Self-check

Signals that an account is due for consolidation

Consolidation is not a goal in itself, by the way. Splitting is fine, as long as every campaign keeps enough signal to steer on. Structure follows the data, not the other way around.