Methodology
Don't spend until the campaign and the conversion have both been proven.
Product runs a closed loop on your catalog and channels: simulate the paid audience, apply the change, prove it with a live experiment, then scale what survived. Campaigns drive the visit. Simulations convert it. Experiments prove both.
How we decide
Simulate. Apply. Prove. Scale.
Guesses don’t get budget. Product simulates the audiences your campaigns already target, applies the listing and site changes you Accept, and runs a live experiment on the campaign and the destination. You Accept again to scale what survived.
01
Simulate the paid audience
A cohort of agents matched to your Sponsored Display, DSP, and search segments tries the listing or the ad, so you learn whether those people click and convert before you put real spend behind the campaign.
02
Apply the change
Recommendations land in Inbox. Accept writes listing title and copy, opens a pull request on a connected storefront repo, or ships a campaign. You keep the final say when money or a live listing moves.
03
Prove it live
Then Product runs a live experiment on the campaign and the destination, with a hypothesis and success criteria you can read.
04
Scale what survived
When the run ends, you get a conclusion: what moved, what didn’t, and what to scale next. Budget follows that conclusion when you Accept, not a hunch.
Experiment types
The designs Product can run on live spend.
Before/after is the default: change one thing, then measure against the baseline. Switchback is for a live listing price that cannot split traffic.
Before/after
Applies a change to one scope and compares results to the pre-run baseline. The default when you want to know if the change made money.
Switchback
Turns the change on and off over time on a single campaign. Used when you cannot split traffic, such as a live listing price.
Stop guessing. Start the loop.
Start free. Connect a catalog and channel, then prove the campaign and the conversion.