.png)
Both, and the honest answer is that the truth sits between the two numbers. Demand Gen is structurally a view-through channel, so a clicks-only model gives it zero credit and a view-inclusive model flatters it — and the gap between them is very large.
On this page
- What does the measured data actually show?
- Which attribution model should I report Demand Gen on?
- How do I read Demand Gen when the models disagree so widely?
- What should I fix before judging Demand Gen performance?
What does the measured data actually show?
A supplements brand's Demand Gen YouTube campaign, measured across attribution models on the same period:
| Model | New-customer CAC | New-customer efficiency |
|---|---|---|
| Clicks-only, 14 days | $161 | 0.6× |
| Clicks-only, 30 days | $171 | 0.52× |
| First-click, 7 days | $271 | — |
| First-click, 30 days | $168 | — |
| Clicks + modelled views, 7 days | $26 | 3.6× |
| Clicks + modelled views, 30 days | $23 | 4.0× |
Breakeven on that account was 1.77×.
So: underwater on every clicks-based model, and comfortably profitable on the only model that counts views. Eight to nine times more new customers attributed — 105 a week versus 12.
First-click does not rescue it, which is useful: that rules out a last-click artifact. The channel genuinely is not getting clicked much.
Which attribution model should I report Demand Gen on?
Client reports stay clicks-only. Even though the channel looks bad on it.
The reason is credibility. Switching one weak campaign to a flattering attribution model is exactly the move a sharp client asks about, and once they ask, every other number you have shown them is in question.
So it gets reported as "underwater but improving," never dressed up with view-through.
How do I read Demand Gen when the models disagree so widely?
The honest framing is that the real figure is somewhere between $161 and $26, and that the channel still needs the structural fixes rather than a better model.
If you want to know which end it is nearer, the answer is not another attribution model. It is a holdout — turn the channel off in a matched geography and see what happens to total new customers.
What should I fix before judging Demand Gen performance?
Check whether a target ROAS floor is on the campaign. On that same account, a floor on one Demand Gen campaign cut new-customer share of revenue from 39.3% to 17.7% while producing zero new-customer improvement — $161 against $162. A total-value floor is satisfied most easily by finding repeat buyers.
Related questions
How much of my non-brand Google Ads spend is going to people who already bought from me?
More than the report says, and the way to find out is not in the campaign view. Segment new versus returning at the conversion level, then check which campaigns are intercepting demand rather than creating it. Read the full answer.
My Demand Gen campaign's views exploded and sales collapsed, and I didn't change anything — what happened?
Check the channel controls before you look at creative, audience or bidding. Google has been folding standalone Display into Demand Gen, so a campaign can start buying open-web inventory without anyone having chosen that. Read the full answer.
My Demand Gen campaign is spending on junk websites — will a placement exclusion list actually fix it?
No. A blacklist is a cleanup, not a control. Read the full answer.
Written by Usama Khan, who runs Google Ads for e-commerce brands and home-services companies at AdventurAds. Figures are from accounts under active management and are anonymised. Last reviewed 2026-08-26.
