Percolate: ‘A Curation Engine That Tells You What to Blog [Or Talk!] About’

Yesterday, Betabeat friend and neighbor Kat Stoeffel told you about Counterparties.com, a new Reuters blog that, in short, teaches you “to read like Felix Salmon.” The site, which features the most relevant and talked about articles from Mr. Salmon’s Twitter and Google Reader, is powered by Percolate, a seven-person East Village startup co-founded by Noah Brier, former head of strategic planning at Barbarian Group, and Federated Media vet James Gross.

Betabeat talked to Mr. Brier about why Percolate hasn’t tapped the local froth in the venture market, whether the Barbarian offices are coming down with startup fever, and why no one looks at Twitter anymore.

When did you know you wanted to launch your own startup?

I left Barbarian Group in January. I had actually built another product in the past, BrandTags, which I sold last year, so I had sort of been in the middle of it. I met up with James Gross my co-founder and we got really excited about what I had built. I had a lot of faith in him as someone to start a company with.

How did you meet James?

We got introduced by mutual friends. He used to run publishing a company called Federated Media and I started a thing called LikeMind, a coffee meetup in the West Village, and a number of other cities.

So is there a case of startup fever going around Barbarian Group?

I think there’s a lot of tinkerers there. A lot of really smart people. If you looked at some of the side project that have come out of there like Cinder. I think they also just did a food truck app. It’s a great place to work.

Why did you think there was a need for something like Percolate?

It was originally a filtering engine. It sucked it in and filtered down and then sent you an email with your top five links. Where we got really excited was the second part of it, the publishing. Counterparties is a pretty good example for that. The idea is that Percolate is really a curation platform, we have a Ph.D in mathematics on staff and we find interesting stuff with the intent of prompting you to react.

Did you think the existing curation engines were missing something?

I just don’t think that we saw this happening somewhere. When you look around the web, there are very few people doing what you do. Most of us are just finding interesting links and adding a sentence or two and pushing them out. So this felt like something. We’ve been thinking of it as curation engine that tells you what to blog about.

How did you end up hooking up with Felix?

Felix was an early power user on Percolate and got excited about it and brought us in at Reuters. They’re licensing our API, they’ve got a Counterparties account, filled with all Felix’s sources, and they comment on it and it gets pushed out.

If it’s an engine that tells you what to blog about, why is Reuters your first media partner?

We totally meant for this to be a consumer platform. Ultimately we’re seeing that brands and media companies have the same problem that consumers have, which is sort of like: What should I talk about?

Your brand clients pay $10,000 to $20,000 in licensing fees a month. Does Reuters have the same arrangement?

Yeah, roughly. We don’t talk specifically about the numbers for any of the clients. But it’s in the same ballpark.

Most startups go for the consumers first and then try to bring on paying clients based on consumer adoption. You went the opposite route.

We’ve used the money from our brand licensing deals to fund the company. We have not taken any venture capital up to this point. So having a business model for us was a pretty natural thing. At the end of the day the product that the brands and media companies are using is the exact same product the consumers are using.

VC money is so easy to come by these days. Why did you opt against it?

I don’t know that there’s any specific reason. We felt good about the product we had for brands. At some point we may raise some venture capital–or maybe we won’t, but at least we have the choice. We’ve been able to support ourselves to this point, we have seven people who work for us now, so we have a pretty reasonably-sized company. We’re not opposed to venture capital. For us and the partners we have in place, it was the right decision.

Where are you working?

We’re on Bond Street and Bowery. We’re in an office with the folks from PSFK, the trends website.  We just rent some desks in here.

Right now Percolate pulls from your Twitter and RSS. But how do you determine which stories are the most popular?

So it’s not actually “most popular.” The algorithm works across about 15 different factors. Popularity is certainly one of them. So it looks at how many people in your network were linking to that thing. But it’s also looking at recency, you know, how old is that link. So if you go on now and two hours from now, there will be probably different links on the top of your brew–because we’re basically saying links lose value over time. Then there’s some secret factors that we keep under our hat. We also try to calculate authority–how important do we think this source is to you, which we try to figure out in a bunch of different ways. The system also learns over time. So that’s a big part of the publishing piece. When you comment on something, when you tag something, you’re also training the algorithm on what you like most.

Is Percolate is going to be free for consumers when you open it up to the public?

Yup. It’s free for them now and it will continue to be free. We’re in alpha now. We’re building out a couple key features and testing out the product and refining designs and stuff like that. But when we open up in the not-too-distant future it will be free for consumers. It’s free for brands to sign on as well. Really, the paid license for the API piece is the one piece that there’s a charge for.

Part of the reason for building Percolate must have been overload from too much information. So do you envision Percolate being a substitute for Twitter, or another site you have to check?

A little of both. With something like Twitter–more and more it’s becoming embedded. When you look at the iOS integration, it only becomes more of that. I don’t think even think about checking it directly anymore.