Harvard Business School tested comment management on live ads.
Here is what the paper found, and how to read it.
48% more return on ad spend
In a one-month A/B test on Facebook and Instagram, the campaigns with automated comment management returned 48% more on ad spend. It hid harmful comments and replied to others. Targeting, creative, and budget stayed the same.
Park and De FreitasHarvard Business School Working Paper 27-011, 2026
This test covered one brand and used BrandBastion, another vendor: 22 purchases against 15. Both campaigns still earned less than they cost. A longer field test found 16% more registrations per click across about 3.4 million clicks. In lab tests with a made-up ad, the gain came from hiding harmful comments, and replies alone showed no clear effect. Numbers from the paper. These are not Ad Reply results.
How to read these numbers
Hiding drove the gain in the lab.
In lab tests with a made-up ad, the gain came from hiding harmful comments. Replies alone showed no clear effect.
It is not a promise.
The A/B test covered one brand and few purchases: 22 against 15. Both campaigns still earned less than they cost.
It tested another product.
The test used BrandBastion, another vendor. These are not Ad Reply results.
Why Ad Reply still replies first.
A buyer who asks a question wants an answer. The study does not show that replies raise sales by themselves, so treat this as good service, not a proven lift.