Ilya Lichtenstein is Co-Founder of Mixrank, an online advertising, competitive data and analytics company. Lichtenstein discussed his company’s offering last week with AdExchanger as well as today’s release of a new API (read more) designed to “enable any enterprise customer complete access to our entire dataset of historical data going back nine months or so and let them finely segment any part of any advertiser’s campaign,” according to Lichtenstein.
AdExchanger: What happened with MixRank coming out of last September’s Y Combinator event? Any surprises?
IL: We had a lot of people talk to us as well as advertisers sign up for our service – we have over 5,000 advertisers using it now. The biggest thing that surprised me is that big brands and creative advertisers are getting more interested in performance. When we were starting out, we were going to focus our product on just performance because e‑commerce advertisers are interested in conversions, tracking everything, seeing where they can cut their costs and get the most cost efficient traffic.
But, in the world of big brand advertising there’s a shift towards being more metrics‑oriented and advertising becoming more trackable – rather than a marketer having a $50 million branding budget, spending it on branding and hoping that it increases sales.
Mixrank has focused a lot more on engagement and targeting specific audiences that are likely to respond well to our messaging and convert. That conversion, by the way, doesn’t have to be a sale. Obviously, for a brand, it could be entering your email address and signing up for something, or it could be a social action like – a “like” on Facebook.
What problem does Mixrank solve?
The “big picture” problem that we’re trying to solve is that the market for display advertising is inefficient. Our goal is to help all advertisers run successful display campaigns – and successful means that it hits the performance that I was talking about. What we’ve learned is that depending on who the advertiser is, they have different approaches to using a product like this or different media planning strategies.
We started by focusing on a self‑serve, easy‑to‑use product. Initially, we wanted to help small businesses get into display and do for display what Ad Words did for search.
But candidly, what we learned is that while some businesses are a good fit, a lot of businesses are not. There’s a mistrust of display, and there’s been a lot of focus on search. So, we started moving up market to build a more sophisticated enterprise product.
Another thing that’s been surprising to us is the different ways that people are interested in using our data. We thought it would just be media buyers that are doing research, but as it turns out, we’ve had ad networks use us who had been doing all this manually.
Market research firms and even financial institutions have signed up. For instance, people at hedge funds are signing up, because they think they can use the data that they get from monitoring changes in advertising campaigns to predict financial performance. That’s been interesting.
The big thing that we’re working on now is what we’re going to release next week (Released today, 3/12. Read more.): our API. That API will essentially enable any enterprise customer complete access to our entire dataset of historical data going back nine months or so and let them finely segment any part of any advertiser’s campaign.
Can you talk about a use case with the API?
For example, you can use data filters and say, “What are the ads that Coca‑Cola was running two months ago that were getting the highest impression share?” Or, “What are the highest‑trafficked sites that my competitors were appearing on for the past two weeks straight?” So all these specific queries pull out of the dataset we gather from millions of ads that we collect. You can pinpoint a strategy or extract broader trends.
The two key touch points that we’re looking to focus on are: either enabling you to go back in time and look at any campaign changes; or, conversely, look at broader trends and see a particular market or industry, such as what the biggest advertisers are doing. Users can do queries, look at the biggest changes, see what tests went on in the past, which ones were successful or not.
That’s of our big push. And in the process, it is something that a lot of startups go through and learn – where they start off as a more simple, straightforward service, and then realize that in order to really scale you have to become a platform that other people are building upon and imitating. That’s why we’re rolling out the API.
How defensible is your product? Couldn’t Google take a look and suddenly crank this out?
The data itself is not particularly defensible, and it’s never been something that we saw as defensible, other than the time it takes to collect it. If you decided to work on a product that does this now, given nine months and enough resources, you could collect this data. There are, of course, companies like Nielsen or comScore that we consider competitors that have been doing it at a small scale for years.
The interesting and defensible part is what you do with the data. That’s where we excel, and that’s what we’re building out right now. We use a machine‑learning algorithm to derive insights from the data. Instead of just showing you the creatives, placements or whatever they’re running on, we can use all of our historical data and build a model on top of that. We can show them the most effective creatives and then the least effective ones.
What we’re doing is analogous to hedge funds and the financial models that they create for trying to predict stock performance. We’re doing the same thing to predict ad performance. It’s similar, in that everyone can see how a stock or ad moves or changes, and how long it’s running. The hard part is fitting a model to that data and trying to effectively analyze it, and go from just correlation to establishing causality.
You’ve heard the phrase, “big data” no doubt. Certainly a lot of the venture capital community is obsessed with “big data.” What do you make of it?
There’s no question that the volume of data that’s out there is exploding right now. There’s more and more analytics as well as open‑source technology that enables us to capture more data than ever before. It’s incredibly valuable.
The real challenge, and the next generation of big data, if you will, is evolving toward making sense of it all and derive insights. That’s a lot harder than just saying, “We’re a big data company and we’re collecting all this data.” There’s a ton on VC interest in this, and there are a lot of startups that are starting to play in this space. It’s just starting to heat up, and it may very well be that big data will be the defining industry of 2012, just like cloud technology has been the defining trend of 2011 or 2010.
The challenge is, one, aggregating enough data to get meaningful insight out of it but, two, actually making sure that you don’t get lost among all the noise. You have to be scientific and analytical about it and not rush to conclusions. Right now, we’re in such early days that it’s hard to predict where the value will come from. That’s the big challenge facing the industry right now, because it can’t all be valuable so you have to extract the nuggets of gold and the valuable insights out of it.
Mixrank is in Silicon Valley, of course. Any thoughts about moving closer to Madison Avenue? Or is the “distance factor” overblown?
Definitely, we’ve given that some thought. In the early days, when we were first meeting with Y Combinator’s Paul Graham, he said, “Well, you’re an advertising company so how about you move to New York?” We’ve been resistant to that, just because we wanted to define the vision and plan for the company in a specific way. At our core, we want to be a technology company, and differentiate on technology and product, versus being an ad network that might be more of a sales‑driven company.
Not that there’s great challenge on the East Coast, but a lot of the people that want to work in pure technology, and that want to work for a startup, especially a big data startup, they’re concentrated in Silicon Valley. A lot of the smart data guys on the East Coast might be more driven to finance or trading. So that was a decision that we made very early on – that we wanted technology to be the focus, rather than sales. And personally, I love living on the West Coast. I love the weather and everything else.
How many people is Mixrank up to today? Any plans on additional funding?
We just raised a $1.5 million seed round in December. We had some awesome investors from funds like 500 Startups to people like Mark Cuban – and advertising industry people as well. We’re not looking for funding immediately. We’re still a small team, four people right now. We’ll probably grow to six or seven in the next month or two. Right now, we’re heads‑down, building the next generation of the product, and building out the team. That is our focus at this point.
Looking to the next 12 to 18 months, any milestones in particular that you would like to see your company accomplish?
The next 12 months is exciting for us, because we’re going to start ramping up development. Up to this point, what we’ve been doing is just collecting all the data. In the next 12 months, we’re going to be focusing on making sense of it all and trying to identify the important patterns and build something that is valuable for bigger advertisers and agencies.
We need to transform our product into something that drives ROI and create a real benefit. That’s probably something we’ll build in the next year. We’ll be able to offer much more sophisticated and nuanced recommendations to advertisers that will show them exactly how they need to target their campaigns. We’ll generate their media plan and, potentially, cut out the waste and inefficiency of the current display market.
By John Ebbert