Platform-Reported Averages Are a Hypothesis, Not a Result
Reddit's Max campaigns went GA with 17% lower cost and 27% more conversions reported. Real numbers. Still not a result for your account.
Luka JovanovićReddit2 min read
Key takeaways
An average hides its distribution
Nobody outside the platform sees how many advertisers gained 40% and how many gained nothing, or worse. The average is real and tells you nothing about you.
The tool still closes a real gap
Reddit's manual campaign management has always demanded more attention than small teams have. An automated layer is a genuinely different product.
Test it like a hypothesis
Small, contained, on a budget you can lose, against a campaign you understand well enough to notice the difference between smarter and merely different.
17% lower cost. 27% more conversions. Those are Reddit's own reported averages for its AI-powered Max campaigns, which went generally available to every advertiser on September 4, after a longer beta earlier in the year (MediaPost; Social Media Today).
Good numbers. Worth noting. Not, on their own, a reason to move real budget yet.
What an average leaves out
Here's my actual issue with numbers like this, and it has nothing to do with Reddit specifically. Every ad platform reports its automation wins the same way: an average, across its own advertiser base, measured against its own definition of a "standard" campaign. Nobody outside the platform sees the distribution behind that average, how many advertisers saw 40% gains and how many saw nothing, or worse. An average is a real number. It's also the easiest kind of real number to build a confident headline around without telling you anything about your specific account.
That's not a reason to ignore the tool. Reddit's automated targeting, creative selection, and budget allocation close a real gap. Manual campaign management on Reddit has historically demanded more ongoing attention than a lot of small teams have to spare, and an automated layer removing that overhead is a genuinely different product than what existed two years ago.
How to test it
It is a reason to test it the way you'd test any platform's self-reported win: small, contained, on a budget you're fine losing entirely, against a campaign you already understand well enough to notice if the automation is doing something actually smarter than you would have, or just doing something different and calling it smarter.
Platform-reported averages are a hypothesis about your account. Not a result. Test it like one.
Sources
Written byLuka Jovanović
Seen it go differently?
Tell me. I'd rather be corrected than quoted.
