Introducing PerfectRec

Finding exactly the right product should be easy. The problem is, it’s not.

Spammy search results and fake reviews make it nearly impossible to get smart, honest recommendations online. 

If you know what you’re doing, good info is still out there. People in the know ditched Google and Amazon for product reviews years ago in favor of Reddit. But most people don’t know about Reddit, let alone how to do a Google site search, and even if they do know, finding the exact info they’re looking for can be a time-consuming slog. And anyway, they shouldn’t have to know — all they want to do is get a product recommendation!

So we built something different. It’s not a list of generic recommendations for generic people. It’s a decision engine that combines the real-life knowledge of top subject matter experts with the power of machine learning to find the best products for you. 

We hope it’s the best of all possible worlds — the human expertise you can still find in some corners of the internet, but in a totally accessible format that requires no tech-savvy at all. We want it to be like asking your nerdiest, most obsessive friend what computer to buy. Let us be your obsessive friend. 

There’s a reason this doesn’t already exist: it’s really hard. That’s also why we’re doing it. Between us, we have PhDs from Stanford, UC-Berkeley, and the University of Michigan. We’re economists and machine learning researchers. Our Chief Science Officer was the first head of data scientist at Etsy. Our technical advisor teaches Machine Learning at Berkeley. And our CEO has a PhD in Economics with a focus on ecommerce and pricing. We like hard problems. We believe an extremely accurate personalized recommendation engine should exist, so 

Welcome to PerfectRec.

How It Works

We scoured the internet (and the world) to find top product experts in unlikely places —  the kind of people who run product-specific subreddits and consider TV hardware “a hobby.” One of our phone experts owns 33 different Androids, just for fun.

Then, we put them to work doing what they do best: analyzing each product in fanatical detail. Our experts compile, review, and verify product stats by hand — we don’t rely on scraping. They also filter out the junk. We don’t include every single phone available in our calculations. If a model doesn’t meet our expert quality thresholds, we’re not going to recommend it to you. Finally, they figure out what really matters in a product: what are the specs you should actually care about? What separates a good laptop from a bad one? What’s genuinely exciting, and what’s manufacturer jargon? Is a TV truly bright enough for a sunny room, or is the manufacturer lying about its nits?

The experts then turn their analyses over to our Recommendations Team — a whole different band of obsessives. They comb through the expert data, translate it into a series of simple, straight-forward questions anyone can answer, and then use the latest machine learning methods to find the best options on the market for your particular needs. It’s more than just a filtering system: arriving at actually useful recommendations means we have to be able to balance contradictory needs — what’s the best product when you want feature A, but you also want mutually exclusive feature B? Our recommendation model lets us take all your needs into account, even when they conflict. We take your preferences seriously, but we also tell you when we think you should consider being just a little bit flexible — and why. In other words, we’re able to capture the nuanced advice you’d get if you were talking with an expert one-on-one.

Our Plan

We’re starting with phones. The reasoning is pretty simple: they are a huge category. People buy them every few years, and every time, there’s a ton of new information to parse. Phones have an annual product cycle, so the market is perpetually in flux, and most importantly, people use them for 5 hours every day — which means they can give us feedback on our recommendations. Are they as good as we think? How can we make them better?

From there, we’ll expand to TVs and laptops, and then begin branching into other categories: cars, home appliances, e-bikes, credit cards, and beyond. Eventually, we want to be your first stop before you shop.