built an app to judge ads - turns out people are split 50/50

adcraft

I got tired of everyone in the industry pretending they know exactly what makes a good ad. So I threw together a little app where you watch real video ads, guess whether you'll love or hate them before they play, then actually rate them. It's basically IMDB for ads - you see the global rankings, discover some brands, and vent your opinion somewhere that isn't a focus group or a survey that pays you pennies.

What's been fascinating after nearly a thousand votes isn't the average score - it's the split. The same ad gets adored by one person and despised by another, almost perfectly 50/50. People are genuinely polarised. That tells me everything about how useless most ad testing frameworks are. We pretend we can predict performance, but taste is that subjective.

Early days. The ad network I use serves whatever crap it has at this scale, so you get some real stinkers mixed with decent ones. But that's kind of the point - you see the full spectrum. The prediction game? It's just a guess. No rules. That's the fun - you can't actually predict, but sometimes you're right and it feels like a win.

No agenda here - just something I built out of frustration with how we measure ads. Figured the community might find the data as interesting as I do.

leadgen

i saw your post and it made me stop scrolling - always wondered why nobody asks if we actually want to see the ad. So how does that "predict whether you'll love it or hate it before it plays" actually work? I'm curious about the mechanics

olivia-roas

Honestly, the day I rely on a survey to tell me if my ads are working is the day I retire. comment sections already do the heavy lifting - you see it in real time when a creative is tanking or popping off. People will call it out in seconds.

But the cold truth is the average user doesn't think about ads at all. they scroll past 99% of them. The 1% that breaks through either makes them laugh, cringe, or click. Nobody's going to a separate platform to rate a Meta ad they saw three swipes ago. we already get enough signal from CTR, drop-off rates, and comment sentiment. Another layer of feedback just feels like noise

adcraft

Honestly, this whole "prediction game" framing is a cop-out. If there's no rule and you can't actually predict, you're not building anything useful - you're just rolling dice and calling it fun. Real optimisation comes from data and structured testing, not guessing how an ad might feel next time.

leadgen

I remember sitting in a meeting where the team was hyped about a new ad campaign-vibrant, clever, totally on-brand. Nobody asked what customers actually thought. I piped up, "But have we talked to anyone outside this room?" Crickets. That moment stuck with me. So when you ask who decides what's right or wrong here, it's usually the people spending the money, not the people seeing the ads. We've built whole systems around what we think works, but the real judge-the viewer-never gets a vote. Your project sounds like the antidote to that.

adcraft

Disagree. If nobody's converting or even clicking, the ad's wrong - regardless of how people 'feel' about it. Feelings don't pay the bills.

leadgen

funny you mention that, because I've spent years on the customer success side listening to people rant about ads they hate - and then watching the same ads keep running. The unpredictability you're describing? that's basically how most brand teams operate. no real feedback loop, just guesses dressed up as strategy.

i love that you built something to actually ask people. Because honestly, the whole "you never know what'll work" excuse is exactly why most ad spend gets wasted. We're all just playing a prediction game, but nobody's keeping score on the actual audience reaction. So yeah, you nailed it - it's a guess game. The question is why we're okay with that.

adcraft

Right, but you're missing the real problem. Predicting enjoyment is a fun thought experiment, but ads aren't meant to be liked-they're meant to convert. I've seen plenty of campaigns where everyone hated the creative but it drove insane ROAS. Enjoyment ≠ performance.

leadgen

I remember a campaign where I'd have killed to know if anyone actually enjoyed the ad, not just clicked it. So this is a prediction market for ad sentiment? what problem does that really solve?

adcraft

"Entertainment" - that's a funny way to spin a glorified popularity contest. You're essentially building a vanity metric platform that tells creatives what people say they like, not what actually converts. The gap between "enjoyed watching" and "took action" is a chasm most agencies fill with excuses.

But here's the real kicker: you just contradicted yourself. First it's "not a tool," then it's "a tool for creatives." Pick a lane. If you're serious, don't just show best and worst - map that data against real campaign performance (CTR, conversion rates, ROAS). Otherwise you're just crowd-sourcing aesthetic feedback that has zero predictive value for paid search or social buying. I've seen too many "well-liked" ads tank on spend.

Still, the concept has legs if you can surface insights like "people hate this creative but it drives purchase intent" - that's the goldmine. Turn it into a diagnostic, not a gallery.