We paused Google Display two weeks ago and our overall CPA dropped 18%. Does that mean Display was just wasting money?
Not exactly. Display was doing two things simultaneously — generating some real conversions AND inflating your reported conversion count through view-through attribution. Pausing it removed the inflation, making your actual CPA visible for the first time.
What happened when Display was running:
• Display reported 412 conversions/month via view-through attribution (1-day window)
• But Display was serving ads to millions of impressions. Statistically, many of those people would have converted anyway through Search or organic
• When I ran a geo-holdout analysis comparing Display-on vs. Display-off markets last quarter, true incremental Display conversions were approximately 85 — not 412
What pausing Display actually did:
• You lost ~85 real conversions (the truly incremental ones)
• You lost ~327 "conversions" that were being double-counted — people who saw a Display ad but converted through Search
• Your Search and organic conversions went up by roughly 290, because those channels now get credit for what was always their traffic
• Net: total real conversions dropped by about 85, but reported CPA improved because the denominator is cleaner
Is Display worth restarting?
At the $9,200/month you were spending, Display's true cost per incremental conversion was $108 — which is actually competitive with your Search non-brand CPA of $94. The problem wasn't waste, it was measurement. If you restart Display with proper incrementality measurement instead of view-through attribution, it's a viable channel.
Recommendation: Restart at 50% budget ($4,600) with a built-in geo-holdout for ongoing measurement. I'll track true incremental CPA weekly and flag if it drifts above your $120 ceiling.
11:23 AM
Results are based on causal inference models applied to your campaign data. Always verify critical decisions with your team.