Your Agency's Best Month Was Your Brand's Worst
The ROAS hit 12x. The agency sent champagne emojis. But new customer acquisition dropped 30%. The metric moved; the business didn't.
The Pattern
Smart Bidding optimises for whatever you tell it to. If you tell it to maximise conversions at target ROAS, it will find the cheapest conversions available. Those are almost always people who were going to buy anyway.
The Celebration
Every agency has a version of this story. A month where ROAS doubled, CPA halved, and the deck practically wrote itself. The client was thrilled. The account manager got a bonus. Nobody asked the question that mattered.
The question that mattered was: did more people buy, or did we just get better at claiming credit for people who were already buying?
What Actually Happened
When you set a target ROAS and tell the algorithm to maximise conversion value, it learns fast. It learns that branded searches convert at 15x ROAS. It learns that retargeting past purchasers converts at 20x. It learns that showing ads to people who just received your email converts beautifully.
So it does more of that. Much more. Your budget shifts silently from prospecting to harvesting. The dashboard glows green. The P&L tells a different story.
"The algorithm did exactly what you asked. The problem is you asked for the wrong thing."
Branded Cannibalisation
The clearest symptom is branded cannibalisation. As paid branded spend increases, organic branded clicks decrease by roughly the same amount. Total revenue stays flat. But now you are paying for traffic you used to get for free.
This is not a Google Ads problem. It is a measurement problem. If your agency reports paid performance in isolation, they will never see it. The only way to catch it is to look at blended metrics: total revenue, total traffic, total new customers, regardless of channel.
Agency Report Says
ROAS up 40%. CPA down 25%. Conversion volume up 15%. "Best month ever."
P&L Says
Total revenue flat. New customer orders down 30%. Organic traffic dropped 18%. Contribution margin unchanged.
The Acquisition Flatline
The second symptom is subtler. New customer acquisition stalls while returning customer revenue holds steady. The agency report does not distinguish between the two, so the numbers look fine. But the business is slowly eating its own seed corn.
Every brand has a natural repurchase cycle. If you stop acquiring new customers, revenue holds for 3-6 months before the decline becomes visible. By then, the agency has moved on to the next client.
The algorithm is not doing anything wrong. It is optimising for the signal you gave it. If conversion value is the goal, harvesting existing demand is the rational strategy. Prospecting is expensive, uncertain, and lowers ROAS.
What to Measure Instead
The fix is structural, not tactical. You need metrics that cannot be gamed by shifting spend to branded terms.
Metrics That Tell the Truth
- •Blended new customer acquisition cost: total marketing spend divided by total new customers, regardless of channel
- •Incremental revenue: measured through holdout tests, not attribution
- •Non-brand paid traffic share: what percentage of paid clicks come from genuinely new queries
- •Contribution margin per order: revenue minus COGS minus ad cost, at order level
- •Organic cannibalisation rate: change in organic branded clicks as paid branded spend changes
None of these metrics appear in a standard Google Ads report. That is the point. If your measurement framework only uses data from the ad platform, you are asking Google whether Google works. The answer is always yes.
Next Steps
Ask your agency one question: what happened to new customer acquisition last month? If they cannot answer without checking, they are optimising for a metric, not for your business.
The Bottom Line
A good month for the agency and a good month for the brand are not always the same thing. The difference is whether new customers are growing or whether the algorithm is simply getting better at taking credit.
Get an Honest Assessment