Radar is honestly one of our favourite things to show people. In a sentence: it's your ad performance evaluator — it looks across your ads from the last 30 days, grades them with our AI algorithm (trained on $9 billion in real ad spend), pulls out your winners and helps you scale them. Think of it as an extra brain sitting alongside you, telling you what's working, what to scale, and what's worth another go — so you're not staring at a spreadsheet trying to work out where to take things next.
How Radar grades your ads
Radar scores each ad across four things:
Hook — does it stop the scroll?
Retention — do people keep watching?
Click-through rate — does it make people click?
Conversion — does it actually make money, against your goal?
We show these as letter grades rather than a wall of numbers and percentages — because everyone understands letter grades at a glance, whether or not you live and breathe media buying. Each ad gets a grade on each of the four, plus an overall grade.
Reading one is intuitive: if an ad's hook and retention are a B but its click-through and conversion are a D, you can see straight away that that's what's holding it back — and where to focus to turn it into an A.
Two quick things worth knowing: static ads are graded on click-through and conversion only (hook and retention are video things), and Radar grades each ad separately — so the same video running as two different ads will be graded twice.
Winners — and where the real gold is
Radar sorts your ads into tabs so you know what to do, not just what scored what:
Winners — your strongest ads. These are the ones to scale.
High iteration potential — this is where the real gold is: ads that are so close, where the hook was a little weak, or the conversion, and a small tweak could turn them into your next winner.
Iteration candidates — worth a look; good signs, just not enough spend behind them yet to be sure.
Most people assume the winners are the point. They're lovely to have — but it's the high iteration potential tab we'd steer you to first, because that's where Radar really earns its keep.
Setting it up
The first time you open Radar, you'll set your benchmarks so it's grading against your goals.
Timeframe — 30 days is the usual starting point.
Conversion metric — how you measure success (ROAS, CPA, or a custom metric like cost per trial started).
Your goal — the target for that metric (for example, the most you want to pay for a trial or purchase).
Radar looks back over your existing ads to grade them, then keeps updating as you launch new ones.
You'll find it under Analytics → Radar. You'll need a connected ad account with some ads and conversion data for it to grade, and it works across Meta and TikTok.
Turning a grade into your next ad
This is where Radar really earns its keep. Open an ad and click Scale under the recommendation, and Radar writes you a proper, detailed recommendation — a fresh angle, new hooks, ways to attack the same persona:
For a video, it writes you a brief (it watches every single frame, which is why it takes a minute or two — well worth the wait).
For a static, it steers you to generate new versions with the fixes applied.
And these aren't vanilla, half-a-page AI answers — they're specific, telling you exactly what to change and why. That's the difference the training makes: Radar comes at your creative like an expert, not a generic chatbot.
You'll also see reference ads it pulls in — winning and trending ads from across the library — there to show you a format, hook or editing style worth borrowing, so they won't always be from your own industry.
From there, use the recommendation as your jumping-off point for the next round, and launch straight to Meta when you're ready. (Your last few generations are saved in Radar, so you can always come back to them.)
Two ways to work
You don't have to do it all in Radar — whichever suits how you like to work:
In Radar — click Scale on a winner or an iteration candidate for the AI recommendation.
With Raya — ask her about the same ad in your own words, or set up an automation so your winners and iteration ideas land in Slack once a week (see Automate your reports with Raya).
It's the same underlying brain either way — just pick the workflow you prefer.
A note on credits
Radar grading your ads is included — look around as much as you like. When you ask it for a recommendation, you'll see what that costs before you run it (just hover), and you can always review your usage in your consumption report. There's more in Understand & manage AI credits.
"My Radar looks empty"
If Radar isn't showing anything, it's almost always one of two things: your ad account is still syncing (check the "synced … ago" note in the top corner), or there simply aren't any winners in your chosen timeframe yet. Give the sync a moment, or widen the timeframe, and they'll appear. If it still looks off, just shout.
And as always, if you'd like a hand setting your benchmarks or reading your first set of grades, just shout — happy to walk through it with you.