Convertro Tackles Attribution Challenge Through Television’s Lens

convertro-zwellingTV has a tenacious grip on media budgets, and so it makes sense to place the boob tube at the center of attribution models. Anyway, that’s where attribution pure play Convertro is placing its bets.

“Clients want to know how to spend money better, where to spend money to make money,” CEO Jeff Zwelling tells AdExchanger. “Convertro’s dashboard will tell you to move 6% of your television budget from BBC to A&E in order to realize a 4% increase in profit. We do a lot of things, but at the end of the day, that’s what we really do.”

Zwelling spoke with AdExchanger recently about optimizing TV budgets, the falling costs of offline data partnerships, and Convertro’s accelerated marketing efforts.

AdExchanger:  Catch us up on the Convertro platform. Where are you today, and what are the priorities?

JEFF ZWELLING: We’ve been working on this for three years with no sales and marketing. We just focused on making a better mousetrap. The original idea — to be frank, we didn’t even know it was called attribution — was spawned out of another company that I founded and sold to private equity. The underlying principle was how to find more profitable marketing vehicles. Attribution is just a subset of what we really do, which is identify opportunities in marketing for greater returns on your investment. We’ve built a lot of technology that is critical to achieving that goal and has a profound, material effect on the business.

We weren’t really interested in pretty dashboards, eye candy, or shiny new objects. What we’ve learned is that the most important component for our customers was the ability to understand the relationship of television to digital and its impact on revenue and sales.

A lot of our investment has gone into becoming experts in the media measurements of television and building models that inform our customers on how to buy television better, especially since users are increasingly interacting not only across television and digital but across multiple devices prior to making any kind of purchase.

Who’s your target customer?

Our target customer is someone who spends millions on acquiring customers across television and digital. Typically, we focus on people who have direct revenue events. Someone called us direct marketers, but that word is a little misleading because most companies today are, in fact, direct marketers; they’re all trying to drive a customer to transact with them, and they spend a lot of money to get there. In most of our target there’s a considered purchase, meaning that people don’t pull the trigger right away, either because the average order value is not extremely low and/or there’s some consideration to be done.

For example, let’s say I’m in market for my vacation plans. I have to talk to relatives and make sure they’re all going to be in the right part of Florida at that time, and I have to find the hotel. My interactions with multiple travel vendors has taken a month and frankly, as far as I know, only technology capable of understanding those multiple interactions across multiple devices will eventually lead me to pull the trigger and book those hotels and flights for my family.

Talk about your data sources for the offline or TV side. Who are the most important partners, and what are the most important sources of data for online to offline attribution modeling?

Online to offline attribution modeling typically refers to conversions that occurred off the Internet, like in stores and car dealerships. TV side is literally offline to online – how to measure people’s response to TV and its impact on sales.

We have several national retailers whose majority sales occur in retail stores. We draw up on partnerships with companies like LiveRamp, and we also have a proprietary partnership with someone who provides us the data on what cars are bought every night. We actually know which cookie bought which car for about 40% of the new cars purchased every month. This requires the client to collect an email address and/or a name with an address, which many of the clients do now as part of in-store promotions. Without us getting involved with the actual PII, we’re able to work with a vendor like LiveRamp to correlate that email address or that physical address to the online Convertro cookie, and then we see all the marketing events that drove the customer into the store to buy.

Are data relationships getting more expensive?  

The data relationships are getting cheaper. Companies can see in-store correlations with multiple startups. They recognize a business model and so they’re all very hungry for our client’s business. Once we have something that needs to be custom-built, it’s increasingly much easier to get.

Processing time is expensive. When a company wants to partner with us or is pitching me for whatever reason, the first question I ask is, “Are you on Amazon?” Invariably, the answer is yes. The second question is, “What’s your Amazon bill? ” If I don’t hear a number that’s $50,000 to $100,000 a month, they’re not really doing big data. The CPUs necessary to process big data are expensive. It’s easy to say you’re doing big data, but you know you’re telling the truth when you have monthly CPU rental fees of $70,000 to $100,000.

Could you talk about the execution platforms and how you integrate with them to help clients modify their media plans? Are there ways that you can help plug into demand-side platforms for real-time bidding?

We know what people should buy and how much they should spend on it. We can either provide that insight in a human-readable form or in a machine-readable form. Output in a machine-readable form sits in persistent RAM level data and it can be tied into any big drive. Today we tie into Invite Media, MediaMath, Marine, Google, x+1, Commission Junction, Linkshare, and DoubleClick DFA.

Currently there is no television-buying platform, but we’re really excited about the people building technologies in that space. Being able to effectively measure and optimize television is incredibly groundbreaking, and that’s what we really focus on. The digital attribution is an important component because most people who respond to TV engage in digital paid media, and most of our clients have done probably 60/40.

Where are you at with funding?  

We recently announced Series B. We originally raised in 2009 from Jeremy Levine of Bessemer Venture Partners. Levine is also responsible for LinkedIn, Yelp and Pinterest. We also have a bunch of high-powered investors like Chris Dixon and Bill Wise, ad-tech luminaries. We raised from another round that’s completely being deployed against sales and marketing. Up until this year, I was actually thinking that it was too early to invest in marketing for our category, but I no longer believe that. Now I think that the opportunity is huge, and most of my day is spent figuring how to use that capital efficiently.

We’re working on a distributional relationship in Brazil and signing up for one conference a month right now. We’ve come out of our reclusive shell.

What changed your mind? What made you think it’s a good time to start marketing?

Incredible amount of inbound leads from huge companies. RFIs and RFQs flying around left and right, lots of dollars suddenly being deployed against us.

What do clients really want from an attribution play?

Our clients, like I said, are not really shiny-object customers. They don’t want just another slide in their deck. They are very focused. They want to know how to spend money better, where to spend money to make money. Convertro’s pure and core strength is insight. Convertro’s dashboard will tell you to move 6% of your television budget from BBC to A&E in order to realize a 4% increase in profit. We do a lot of things, but at the end of the day, that’s what we really do. That’s what our customers want – to know exactly what to do and be provided with the data.

Convertro’s algorithms determine the biggest economic opportunities for a company. We rank them by their potential upside or profit. We post them to the dashboard in order of their value to the customer. We have algorithms that see what clients actually did and if the prediction matched the outcome, and whether the client acted on a recommendation or not.

What else?

I think in order to do this right – to achieve certain technical milestones – you have to be able to deal with several problems in a privacy-compliant way. If you’re getting beyond “last million paid” as your primary attribution model, you better make sure you know what the earlier marketing events are. If you use simple cookies, you’re going to run into lots of data issues.

Also, cross-device is very important to TV, especially because most of the response to TV is increasingly occurring on tablets and smartphones. Conversion is still occurring on laptops and PCs. Being able to tie together and correlate devices and users enables you to know what the effectiveness of your TV is.

People who can’t measure the effectiveness of their work keep doing it, I think, because they’re afraid of what they may find. It’s like a buy where no one’s really measuring you, so you don’t know if you’re right or wrong. It’s a lot easier to be wrong if no one knows about it. If you actually focus on what you can measure, you will know how effective you are, which determines whether or not the client sticks with you.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!