Moloco measures CTV incrementality two ways. Its platform has built-in ghost-bidding tests, which compare households Moloco would have shown an ad to against households that actually saw one. And through a partnership announced in October 2026, Haus runs independent geo-based tests that compare test markets with control markets. Outcomes are tracked through the advertiser's own mobile measurement partner, or through a pixel for web.
That's a reasonable setup. Here's how to judge it, and how the methods compare.
The three main CTV incrementality methods
| Method | How it works | Strengths | Weaknesses |
|---|---|---|---|
| Ghost bidding | The platform logs the auctions it would have bid on for a control group but doesn't serve the ad, then compares outcomes | Always on, no wasted media, large samples | Run by the vendor being measured; depends on how the control group is built |
| Geo testing | Ads run in some markets and not others, and outcomes are compared | Independent of any platform; captures offline and cross-channel effects | Needs enough markets and budget; regional differences add noise |
| PSA testing | The control group sees an unrelated public service ad | Easy to set up and explain | You pay for the control impressions |
How Moloco's approach fits
- Built-in ghost bidding. Moloco's CTV page says advertisers can run ghost-bidding tests to measure the lift its Performance CTV drives. The page doesn't publish test design details, so ask for them.
- Household matching. Moloco connects CTV impressions to outcomes through household and IP matching. Ask how that matching handles shared Wi-Fi and devices that leave the home.
- Your own measurement partner. App outcomes run through MMPs such as AppsFlyer, Adjust, Singular, Branch and Kochava, so results aren't only in Moloco's own reporting.
- Independent geo tests. The Haus partnership adds third-party geo-based studies for Moloco's app and CTV campaigns. That's the stronger check, because the platform isn't grading itself.
What to ask any CTV vendor about incrementality
- How is the control group built, and who chooses it?
- Does the test measure all conversions, or only the ones the platform can attribute?
- Can we run an independent geo test, and will you hold out markets for it?
- How long does the test need to run, and what lift could it reliably detect at our budget?
- Can we see results by publisher or app, not just overall?
My take
Use the platform's always-on test to optimize within CTV. Use an independent geo test to decide how much budget CTV deserves compared with your other channels. If the two disagree, trust the one the vendor doesn't run.
I design and read CTV and geo experiments for brands. See how a Measurement Audit works.
Sources: Moloco Performance CTV; Moloco and Haus partnership, Daily Research News Online, October 2026.