After 6 years of running ghost kitchens and 4 concepts across the US and Canada, here is what I have learned about what really drives ranking on Uber Eats. Beyond the standard agency advice.
Everybody asks me the same thing: how does the Uber Eats algorithm actually work? The honest answer is that nobody outside of Uber's engineering team knows all of it. But after operating 4 concepts on the platform for 6 years, and obsessively tracking what happens when we change one variable at a time, patterns become very clear.
Here's what I've learned actually moves ranking. Not the stuff agencies tell you. The real stuff.
1. Click rate is the single biggest lever
If I had to pick one metric to optimize above all others, it's click-through rate (CTR). When a customer sees your restaurant in results and clicks on it, the algorithm registers a "positive signal." More clicks = higher rank in the next session.
This is why photos matter so much. It's not about aesthetics. It's about thumb-stopping power on a 380px phone screen. A mediocre photo gets 1.8% CTR. A great photo gets 5%+. That's a 3x signal to the algorithm.
2. Conversion rate matters almost as much
Once a customer clicks into your restaurant, do they actually order? If 100 people click but only 5 order, the algorithm thinks your restaurant is "click bait" and starts suppressing you.
The main conversion killers are:
- Menu prices that don't match the hero image
- No clear bestseller, decision paralysis
- Weak or missing descriptions
- Stockouts on the items people came for
3. Accept time and preparation time
The platform tracks how fast you accept an order after it hits your tablet, and how long it takes you to mark it ready. Both are scored.
Restaurants that consistently accept in under 30 seconds and prep within the estimated time get rewarded. Restaurants that stall get quietly demoted.
Check your "acceptance time" in Uber Eats Manager under Reports. Most operators have never looked at it. If yours is over 60 seconds, you're leaving ranking on the table, before even talking about photos or menu.
4. Ratings and reviews are weighted heavily
A 4.9 rating vs a 4.6 rating is a massive gap on the platform. And contrary to what most people think, response time to negative reviews matters. Uber tracks whether you engage with customer complaints.
Our rule: every single review gets a response within 24 hours. Even the 5-star ones. Especially the 5-star ones, the algorithm likes engaged merchants.
5. The dreaded "new listing boost"
New restaurants get a 30-45 day honeymoon period where the algorithm artificially boosts their visibility. Most operators don't realize this and waste it.
If you're launching a new concept, the first 6 weeks are when you should be hitting the platform hardest with promotions, excellent service, and fast acceptance times. You're building permanent algorithmic credit.
What doesn't matter as much as you think
Being a "Top Eats" partner
The badge is nice but it's a symptom of doing the other things right, not a cause of ranking. You don't "get" Top Eats. You earn it through the metrics above.
Paying for extra marketing
Sponsored listings help visibility but don't permanently change your organic rank. They're rent, not equity.
How old your listing is
We've overtaken restaurants that were on the platform for 5+ years within 60 days. Age is slightly weighted but nothing beats fresh, strong metrics.
The compound effect
Here's the part most agencies won't tell you: these signals compound. A restaurant with great photos gets more clicks → more orders → better metrics → higher rank → more impressions to work with. And the cycle accelerates.
The flip side is also true. Bad signals compound downward. That's why struggling restaurants struggle, not because they did one thing wrong, but because they're stuck in a negative compound loop.
Breaking that loop requires optimizing multiple levers at once. One fix at a time isn't enough.