Interview with Matt Wise from Q Interactive
Adrian Bye: Okay, so today I'm here with Matt Wise who is the CEO of Q Interactive, and I've known through a friend I guess for a little while and he came to an event of mine in New York. Matt, thanks a lot for being here today.
Matt Wise: My pleasure, Adrian.
Adrian Bye: Maybe you can tell us a little bit about who you are and a little bit about Q Interactive. I don't actually know much about Q Interactive apart from the fact that you run Cool Savings.
Matt Wise: Sure, that's an excellent place to start.
We do own and operate Cool Savings and as a matter of fact, the company used to be called Cool Savings. We were publicly traded. We went public in 2000 right as the bubble burst. I think our stock price went up about 10% before plummeting about 90%. So, for all those of you who invested early on, I wasn't part of the company, so don't blame me.
But anyway, the business is actually an interesting one because its origin was in coupons and that's what Cool Savings was all about. But over time, what occurred was we were getting paid $0.06 a print per coupon and players like Uncle Ben's came to us and said, "This printing coupons is great and we'll pay you $0.06 for each one, but we really want to talk to our consumers. We want to open a dialogue with them, so if you could instead of printing a coupon for them, get them to sign up for a newsletter and raise their hand for permission to talk to them, we'd love that and we'd pay you $2 a person." So for us back in our offices, it's a time we were a little slow and we said, "$0.06-$2, $0.06-$2 – I wonder which one we should do?"
So luckily by 2004, we figured out that this lead generation thing was really the way to go and advertisers really wanted to start that dialogue with consumers and they wanted us or a firm like us to figure out who were the right folks to talk to and who was actually going to convert, as opposed to them just going out and buying general banners.
Thus was born the concept of a lead gen network, so we have been doing lead generation on the site since about '99. By 2004, we had inverted the model and made it available to other publishers, and the frame of reference when we were a public company, we were doing about $30 million in revenue on Cool Savings. In the fourth quarter of 2004, we did about $5 million in our newly-formed lead generation ad network and then in 2005, we did about $37 million on the network and about $30 million on our historical Cool Savings.
So in the span of 18 months, we'd shifted the entire business model from a Cool Savings centric business to an ad network-centric business and that has continued to grow. The lead generation business and the publisher business continues to grow, and we have expanded outside of just lead generation into e-mail.
All of this is founded upon a simple concept which is the best way to match consumers and advertising we had found is to use very rich data for the targeting. That may sound like a bit cliché and a bit boring than what everybody else does, but the reality is since we sit behind registration usually while we're doing e-mail or we have access to the registration information, we know a tremendous amount about the consumer. We know their name. We know their address. We know their age. From that, we can buy extra own databases, so we can figure out creditworthiness and a variety of other things, and we use all of that to come up with advanced targeting algorithms to figure out the right offer for this consumer at given time.
Then on the fly, we take all of our predictive modeling and we test that in the open marketplace, so it runs against the network, and hundreds of thousands of people come in, and we're constantly saying, "Is this predictive model predicting consumer activity accurately? Are people really taking this particular offer the most?" so it's constantly reaffirmed.
Adrian Bye: So, what happened to Cool Savings? I mean, the original idea was that it's coupons and that people can sign up and get coupons. So, you just grabbed it, headed towards the lead generation business and naturally being focused on the coupon stuff, is that how it works?
Matt Wise: Yes, for all intents and purposes the Cool Savings still exists. The site still exists and we still have coupons on there, but we actually sold the coupon technology about a year ago to News America. So although we still run what is one of the largest, if not the largest coupon destination site, we don't power the coupons – we outsource that aspect of it.
Adrian Bye: When you say network, are you a co-registration network generating leads?
Matt Wise: Co-registration… Yes, most people would think of us as a co-registration network. The reason we call ourselves a lead generation network as opposed to a co-registration network is people can't sign up. The old model of co-registration was you sign up for one thing and you're automatically signed up for two or three more. We never allow a consumer to sign up for offers simultaneously. They always have to go through an additional step, read more about a particular offer, confirm if they're really interested in that offer. But for the clarity and simplicity for folks when they think about it – yes, we're usually a post-registration or a co-registration player.
Adrian Bye: Do you allow pre-checked leads so you always force them to go through at least one more step?
Matt Wise: We do allow pro-checked leads, but whenever you have a pre-checked lead, they ask to go through one more step, so we view the pre-check as a recommendation. So if there's 20 offers on a page, five of them may be pre-checked. Those are our recommendations. If a person hits "Submit" at the bottom of that page, they have to read more about each one of those offers.
If we have what's called an easy opt in where it's just a one-step which is about I think 0.5% of our revenue, we don't recommend it to advertisers, but if someone really wants to drive slightly less qualified volume we do allow that, but we don't allow it to be pre-checked. Meaning if it's only a one step opt in, they have to actively check it on their own.
Adrian Bye: So let's say I'm a user going through your site and I turn up to see five offers pre-checked, and I hit "Submit", what does it mean by I have to read more? Does that mean I have to take an extra step to say, yes, I do want all those five offers?
Matt Wise: Yes, so what happens specifically is let's say you picked the Wal-Mart offer, the JCPenney offer and this Disney Music Club offer. So, you pick those three, you press "Submit". What will happen is for each one of those, a new page will load and we'll say, here's all the information that we've got on you, that we're planning to pass to Disney. We always ask another custom question, so the consumer has to interact with the ad, so we know it's a bona fide person and then person has to hit "Submit" again. There's also a skip button there, so they can say, "Oh, now that I understand what I'm signing up for, I don't want to do this," because some of them have expenses like Better Homes and Gardens Magazine. You may get the magazine in the mail and you have to pay for it eventually. So, we want to make sure that's all clearly disclosed to the consumer before they press "Submit".
Adrian Bye: Obviously that kind of stuff in the past has been quite problematic with the pre-checked and you go through someone's registration past and you end up with 10 zillion e-mails coming in everyday. With what you're doing now, is that avoiding that kind of stuff of people getting overwhelmed and never wanting to come back to whatever the site was again?
Matt Wise: Right, if done appropriately, it's not bad for the consumer at all because a consumer always knows what they're signing up for. Two things can go awry though. Yes, it avoids the old…which is why we hate to use the word co-registration. It avoids the old situation where you came in and signed up for one site and all of a sudden, you got signed up for 10 other things, and you didn't know it. So, a consumer always knows. It's impossible actually to get through and say, "I didn't know I signed up for the Gap," because not only did you get the little pre-checked or unchecked box that you had to check, but then you got this fuller page which you had to interact with. You had to confirm all your information. You had to answer another question and hit "Submit" again and then actually, for complicated ones where we have a telephone call following it or something like that. We actually do a triple opt in where after you press "Submit" we'll put a little warning box up that says, "Hey did you really read this closely? Someone is going to give you a call. Are you sure you want a call?" and they have to press "Submit" again. So rarely do we run into consumer problems where they signed up for too many things.
What we do find however is there are some less scrupulous players in the field that when you sign up for a site, they take your data and they resell it to five or six other players. So you've only signed up for one site, but now you're getting mail from six other players and you don't know how or why you're getting this information, and the reason is because your data is being resold.
Adrian Bye: Did you agree with the terms and conditions when you went through to sign up?
Matt Wise: Exactly.
Adrian Bye: So, what's your policy on that kind of stuff?
Matt Wise: So, our policy is that permission should be clear and conspicuous, meaning it should be on the page in one of two formats – either what we call an active opt in which is there's a checkbox and it's not checked, so you're probably going to get a 20% conversion there or what we call a passive opt in where we pre-checked the opt in box, but it clearly says, "Hey, you're opting in for mail." You can state right there if you're opting in for mail from let's say Cool Savings or you're opting in for mail from Cool Savings and other parties. So, the consumer can say, "I don't want that," and can clearly uncheck it, and we think every time a piece of mail is sent, it should say, "Hey, you signed up on (for instance) Cool Savings. That's why you're getting this piece of mail," and the consumer should always be able to with one click say, "Well, I don't want to get any other mail from Cool Savings or anybody else that you gave permission to at that time."
If you do all that, you put all the power in the consumer's hands and it still allows for direct marketers and promotional marketers to run a robust business. It just enables the consumer to have a choice or whether they want to opt into it and then two, easily have an ability to get out of it.
Adrian Bye: The way I've heard one guy who does a lot of e-mails talk about is when you get some e-mail addresses like through co-registration, you're getting those e-mail addresses at the same time as a bunch of other guys so you got to send them the more mail sooner until they stop using that e-mail address.
Matt Wise: Right, so that's what the unscrupulous players do.
Adrian Bye: Yes.
Matt Wise: There is a race literally to the minute to get out mail and the most common practice is something we call brand spawning. So for instance, you sign up for a site and this is most prolific in the free giveaway sites, so you sign up for free iPods. Five minutes later literally, you get a piece of mail from greatstuff.com. You have no idea and someone owns the URL. This is for example purposes only, but you get something from greatstuff.com. You never signed up for greatstuff.com. You have no idea how they got your e-mail address and what happened was the free iPod folks sold your name and now they're baiting it. So, the unscrupulous players will sell that to 5, 10, even 20 players and all of them race to get the mail out because they know in days or a few weeks that mailbox is going to have literally thousands of pieces of mail in it and it's going to die or it's going to become unproductive. So they want to get in as quickly as possible before it becomes unproductive. But again, the players who participate in that are not good players. I mean, that occurs out there, but that is definitely the low-end players who are doing that.
Adrian Bye: Yes and actually, I'll just add in as a note, like in the free iPod example, I actually interviewed and he was there that the free iPod guy was there and our cocktail, and they were doing some e-mail stuff in the beginning. But, he's overall concept of what he wanted to do with the free iPod program was to create a brand and so they stopped doing any kind of e-mail follow up fairly shortly into the program.
Matt Wise: Yes and so, if you want to create a brand, that e-mail practice will kill you…
Adrian Bye: Yes, alright.
Matt Wise: …and the Gratis folks did that. I mean, they did a fantastic job. They were one of the early pioneers in this and they did a fantastic job at it for a long time.
Adrian Bye: Yes, right.
Basically, the way you started out was with the coupons network and then people signing up for coupons. You then had a company saying, "Hey, coupons are nice, but we want leads and we'll pay more for leads," so you put the registration part behind the sign up process for Cool Savings. You then started to see that hey, this is working really well – this co-registration stuff or lead generation stuff and so we'll start to make this the focus of our business and then you start taking out that lead generation process and taking that to other businesses, so that they can also generate leads for you. Is that a correct evolution?
Matt Wise: That's a precise evolution, yes. The one thing that's, I step in there which you didn't mentioned, but it's an important one is not only did we find out that lead generation was a good idea, but when we had so many advertisers coming to us who wanted lead generation, we ended up with the very simple problem of what advertiser should we put up. Thus was born the beginning of our yield management system and so right now, doing lead generation is a relatively simple concept on the surface. Wal-Mart wants leads. You put the Wal-Mart offer up. Someone fills it out. You send the data feed to Wal-Mart.
Where it gets complicated is predicting what consumers actually want Wal-Mart and when the Wal-Mart offer is going to yield higher than let's say, the Disney offer or the Target offer or the Mortgage offer, whatever else it may be. So that's where you get the higher end players of building very complex algorithms and mathematical equations to predict that and then there's a whole other business on the backend of that of how do I confirm that this person was bona fide. Not only in the user experience we try to do that, but we do it in the background by cleaning up addresses, checking on addresses, preparing the data to be sent over to the client, etc.
On the surface, we all look similar because we all arrive at a page and there's a form there, but everything from how that form was presented and then what happens in the background – there's a tremendous amount of work that occurs and so, with both tools we built up along the way also.
Adrian Bye: So, what kind of eCPM can people see for their site by adding your registration up?
Matt Wise: It stretches anywhere. I mean…
Adrian Bye: I'm sorry. I don't know this model that well, so I don't know what the monetization time, whether that's eCPM or something else.
Matt Wise: No, eCPM is what we use and so it depends on the implementation, so we have people at the low end who are doing what we would call the softest touch possible. So, they may be presenting only six offers, no pre-checks and they'll exclude some of the more hardcore direct marketing players who pay high premiums. So on those, you may be talking about on a page $75 eCPM which for most people if they get a $75 eCPM on a banner impression would be fantastic. That's at the very low end.
At the high end, our best performers who have the most aggressive paths where you've to go through a lot of work to get through quite frankly, probably peak out at I would say, $1,800 eCPM, so it's $1.80 a person. So, there's a large span in between. Our average is around $500 eCPM. I think it's about $520.
Adrian Bye: So, from the point of view of the user like you're going through to do something and then you're getting put through one of these hard registration processes that you end up being worth $1.80 a person, I guess that does roll back out to the advertiser. I mean, it seems astonishing that, that works. I mean, what is in the user's mind like, why are they saying, "I've got to complete this next offer. I have to fill this stuff in."
Matt Wise: Right, so the key is and what we demand from all of our publishers is that the consumer never feels they have to fill it is. This is why this industry's a little complicated. There are unscrupulous players who insinuate to the user that they have to fill something out in order to get to the next page. So we control and demand of all our publishers that they never put any language that may infer, insinuate or anything else that an offer has to be taken because otherwise, you start to degrade the quality leads you send them. But the reason these folks can generate such a high eCPM is in most of the types that eCPMs are generated on promotional sites, is they attract consumers who like to sign up for stuff.
I know that may sound crazy and many folks think that high end sites like About.com is one of our partners and wow, signing up for a newsletter there is going to be a higher quality individual than somewhere else. The reality is those are good people. However, the folks who come to promotional sites really like to sign up for stuff. It's just in their nature, so when they come to the site they see in our offers – many of them are very good. There's discounts. There's newsletters of interest for everything under the sun. There's discounts on magazines. So, they end up looking through there, and finding a lot of stuff they like and signing up for a bunch of it, and again to generate $1.80 a person, although from a $1,800 eCPM sounds egregious, about 40% to 50% of the people who come to our page find something they're interested in.
So, most banners have a click through of 2/10 of a percent. We have about a 40% conversion rate, meaning people come to the page, find something they like, fill out the form, press "Submit". So if we're getting on average on that site $1 a conversion, well that's already generating $0.50. So, if they convert on three offers, they sign up for just a few of them, then they've got $1.50 or $1,500 eCPM. Then if you take that a step further that some of the leads that we generate – there's an education lead or mortgage lead – these highly qualified leads, those people are paying $30, $40, $50 a lead, so you don't have to sign up a heck of a lot of those people to make the needle really move.
Adrian Bye: While this is all going on, you have your process optimized, so that it's always putting the best, most revenue generating leads first, so that it's to the publisher's interest and also to your interest, and therefore potentially the consumer's interest because they're showing the offers where I get, well, I don't know where that's necessary and there'll be offers they can spend the most money, but maybe the offer's not the most relevant for them.
Matt Wise: Right, so the algorithm works much like Google in the fact that who gets shown at the top of the page is actually a combination of propensity to click and convert, so how attractive is the offer and then the rate paid by the advertiser. So, the two of those together result in yield and then the yield is not just across all consumers, so it's not like we say, "On this site, this is the best performing offer." We say, "In front of this consumer on this site is the best performing offer.
So, a woman aged 23 is going to see a radically different set of offers than a man aged 55. A 23-year old woman may see a lot of Pampers, Wal-Mart, JCPenney, etc, Disney Book Club because we're aiming at their propensity to have children, etc. and an older male may have Mortgage Refinance or Automotive Loan or Retirement or A or P or who knows what that matches up with them.
Adrian Bye: So, an interesting case is…are you familiar with the social networking site called tagged.com…
Matt Wise: Yes. I mean, I'm generally familiar with that. I can't say I'm part of it.
Adrian Bye: Yes, it's more for teens. Alexa ranked 120 using viral marketing pretty effectively, generating about 300,000 sign ups per day – that's sign ups for the site. So, Greg's the CEO. He put co-registration on his sign up process and that was sort of a big breakthrough I think that helped bring the site to being profitable. Is that the sort of site you like to work with and then do you use the data points you're collecting on users to help target those better? I mean, how do you work with a site like Tagged?
Matt Wise: Yes, social networking we think is a fantastic spot just because consumers are providing information and there are people who are willing to provide information. This is one of the keys to our targeting and the success of lead generation – you'd find people who are willing to pass over that information.
One of the challenges we have in social networking is the youth aspect of it. If we go too far down the youth level, we can't monetize people under 13, so there's a breakpoint there and there's just a breakpoint in how much revenue you can generate off of the 13 to 21-year old – the group. It's tremendously valuable from an impression and display arena, meaning there are many advertisers who want to influence that group, but the price points paid by many of the advertisers in lead generation are higher for people who have higher incomes and have the ability to convert, so they have credit cards, etc.
Do we like social networking – you bet, it's a great segment and for a player like that, most of the stuff is automated, it's going to pick up particular offers, but we may work on some creative implementations and push the more youth-oriented offers up the spectrum, so that they can get in front of their users.
Adrian Bye: So, what is the ideal kind of traffic that you want most? Is it women aged 30 to 50?
Matt Wise: It's actually, yes, very close to that. It is women probably aged just a little younger than that – probably 25 to 50, and the reality is we like all traffic. I know that sounds a little ubiquitous, but the selection of offers we have – about 1,500 advertisers a year running through the system, so we have offers that appeal to all the different segment groups, whether they're retirees, young people, old people, etc. But, those who monetize the absolute best are middle-aged women or not even – a little younger than middle-aged women. They just shop more online, they buy more online, they're more willing to sign up for things online, etc.
Adrian Bye: So, when one of these goes onto a site, does it affect the branding of the site? I mean, the people look at that and say, "Man, this site used to be really good and now they're wanting me to get a free gift card," or "…they want me to fill in my information, so I can be called back by (whoever)…" Does that actually have a detrimental effect on sites as they add your lead generation path in?
Matt Wise: Not if they do it right and what I mean by that is there are a tremendous number of ways to handle the execution of the advertising, so if you handle the advertising with the consumer in mind that they can always get out of it easily, then it's not a problem. It's just like an interstitial on the New York Times. When you get that interstitial, as long as the skip button is there, no one really minds. If the ad is of interest, they look at it. If it's not of interest, they click on "Skip" and they get by real quick and get to their article.
So the same thing with us – we put up 10 offers in between someone signs up and the user experiences, "Thanks for signing up. We have these 20 offers we thought you might be interested in also." So right there, what happens is: (1) you've positioned it as a positive to the consumer that we found something of interest; (2) as long as you work with a player like us or another high end player, those offers that show up actually will be of interest. There'll be something likely on the page… Again, half the people convert, so it's likely there's going to be something on the page that's of interest to them. But if there's not – if they have zero desire to interact or nothing on the page is of interest, as long as with a single click they can say, skip and move on, then we find there to be zero interference with or bad impressions with the consumer and we reflect on that in two ways: (1) we can track how many people abandon the site in our advertising. So we say that is not the optimal user experience and we can see as publishers choose to get more aggressive – put more ads up, put more pages up that attrition rate gets higher. Then secondarily, we work with publishers, who often pass us back backend statistics, so they'll say, "Hey, we did an A-B split test where 50% of our users went through your path and 50% didn't and we saw that there was a 10% degradation of interactions of that consumer once they passed through your path." So, depending on how big that degradation is we can see if we have affected the user experience or the brand experience on the backend.
Now, a little degradation is okay because the reality is a consumer only has a limited amount of time to interact with a site. Many sites have a four minute interaction time. So, if they're on my pages for a minute and a half of that or two minutes and they're filling out a couple of offers then they get to the core website. Well they probably have less time to interact with the core website which is okay for some folks because let's say they earn $1 off of my registration or even $0.50 and then when they go to their regular website, they maybe peruse through 15 pages on average and each one of those has a $3 eCPM. So they only earned $0.15 on the backend, but they earn $0.50 off of us, maybe we burn 10% of their time, so they only made $0.14 or $0.13 on the background, but they still made $0.50 off of us, so it's a huge gain.
So, that's what we try to keep an eye on. If you're not scrupulous about it, if you're not careful about it though, yes, you can definitely stack up a bunch of ads, take away the skip button and make it very difficult for the consumer to move on and then the experience becomes one that is substantially less positive.
Adrian Bye: So, one of my friends is Drew Curtis who runs FARK.com. Have you ever heard of that site?
Matt Wise: FARK.com?
Adrian Bye: Yes.
Matt Wise: Yes, I just saw him out at the Venture Summit last week.
Adrian Bye: Oh, really?
Matt Wise: I didn't talk with him personally, but he was up on a panel. He seems like a great guy.
Adrian Bye: Yes, he's a good guy. Yes and he's come to a couple of my events. I've talked with him about this sort of stuff on his site and his kind of view point is… He's pretty anti-advertising anyway, he's more of a tech guy really. His point is, look, my guys – they're Linux users or general US. They're male, 18 to I guess 40 or 50 I think he talked about, they're using Firefox, they're running ad blockers and they complain about advertising and all that kind of stuff. How would you work with someone like Drew?
Matt Wise: We definitely would work with Drew and the implementation we would use given his user-base would be slightly lighter, meaning we'd make sure we didn't put too many barriers up in front of the consumer. But again, the ads that are shown will be shown on an optional format, meaning they can say, hey, something is of interest on this page or not. For the FARK.com folks I actually think it would be a good user experience because the nature of FARK.com is coming here to peruse through a bunch of interesting new sites and funny stories – I'm going to pick up the ones I want to read about.
So, our advertising would present it in the same, exact way. Here are 20 ads and offers that we thought you might be interested in. Peruse through, pick out the one you want. If you don't like any, keep moving on. So the paradigm works perfect for this site.
Adrian Bye: Would you put those in between the articles like when someone clicks on an article or are you suggesting…because I've talked to him about that and he said, "That would create like a revolution," he felt. Would it be when a user signs up for the site or would it be some ads on the site in which case he's mentioned that while he gets a ton of traffic, the click-throughs on ads tend to be fairly low.
Matt Wise: Yes, so I would do two things, one…maybe three…I would do registration. You can't bang up people every single time they're on the site with this interruptive flow or every time they click on an article. It's kind of like an interstitial. If every time you clicked, you got another interstitial, you'd get annoyed, but if you came to the site and once every time you're on the site or once an hour or whatever it may be you got an interstitial, that'd be fine in my opinion. I don't think many users would have a problem with that and they all know they're getting free content, so the message can be, "Hey, check out these offers from our sponsors. (1) We've not just picked random ones – we picked ones we think you're interested in; (2) you understand we're advertising driven, so this is how we make our money; and (3) if you don't like it, no problem – just click ‘Close' or click ‘Skip' and you can go back to reading your articles." So, we would probably recommend they do it on registration and then at some intermittent time, maybe every 10th article or something like that they'd get one of these and I'd probably use the equivalent of an interstitial which there were maybe 10 offers on there and they got to pick from one of the 10. If they didn't want to do it, they press "Skip" and keep moving on, especially for FARK.com, I would lead with a message of why I was putting this up that, "Hey, I'm not putting this up to annoy you. I'm putting this up because we're an advertising-driven site."
Adrian Bye: Yes, he'd have a revolution on his hands because they're a loud and noisy group, so they'd be complaining in incredible amount, but at the end of the day, it's whether they stop using the site or whether they continue using it and it generates revenue.
His average user does come to the site and clicks on two news articles when they come and I guess they come everyday or every couple of days, so maybe you'd suggest like once every five click-throughs they get shown something.
Matt Wise: Yes, something like that, so they wouldn't necessarily see it everyday. They'd see it every couple of days.
Adrian Bye: In something like that, how much revenue do you think that would generate per showing?
Matt Wise: Much less, because when you get someone fresh from registration, there's where you're going to get your big dollars. Your intermittent tags later on –you might be down on $20 eCPM or something like that, and on the registration on a FARK.com, I would guess he would be up around…we'd probably do a light implementation, so maybe a $350 eCPM.
So I mean, for a player like that who does large volumes… I mean, some of our players, we write enormous checks for because they have large volumes and if you're registering a lot of people I mean, there's just enormous money to be made. Most sites don't have a large registration-base, most sites are registering hundreds a day. Maybe if they're fairly large, 1,000 a day. But some of these social network sites – some of these viral sites are registering in the tens of thousands a day, and when you're in the tens of thousands and you're getting paid $0.50 a registration, that's an awful lot of money everyday – day in and day out.
Adrian Bye: I don't think Drew is generating a lot of daily registration. He's generating a lot of daily visitors and then some of them sign up for stuff, and he hasn't pushed that side of it but his daily uniques is one million to two million.
Matt Wise: Yes, he's got a huge general reach. I don't know what his stats on registration are – you said he doesn't push it. I'm not sure exactly what volume he's getting there.
Adrian Bye: I don't think the actual registration is that high. I don't think he's really pushed the people's incentive to go ahead in register.
Matt Wise: I see a revenue opportunity for him. I know he lives a comfortable lifestyle, so I don't know if he even wants to delve into it, but there's definitely money to be made there.
Adrian Bye: Can you give like a crazy, out of the ballpark figure on what that could generate for him per year?
Matt Wise: I mean, let's say he was doing 10,000 registrations a day and we were generating $0.50 a registration, that's $5,000 a day. That's $1.5 million a year roughly. Not bad.
Adrian Bye: Do you think that's conservative?
Matt Wise: That would be an average. That would be an average eCPM conversion. Now, the 10,000 registrations a day might be outlandishly high for him or might be low.
Adrian Bye: You mean that's user registration, as in users signing up to his side or is that in showing these users as they click through news articles?
Matt Wise: Right, I'd say that's user registration. On the click-throughs it would be a much lower eCPM, although the frequency would be higher. I'd hate to even hazard a guess, but maybe there's sort of $500,000 in that for him.
Adrian Bye: So, if he was to fix up his way users sign up to the site and really encourage them to register, and them put them through this registration process, that's $1.5 million a year potentially there and then putting them through occasionally advertising in a way doesn't frequently run too much, it could be another $500,000, so he could add on another $2 million in profit through his site every year.
Matt Wise: Yes, tell him to call me. It's humorous that you bring up Drew because he was a panel of five folks and he's a person that I've got on my list that I want to give a ring to in the next couple of weeks, just to chat about just this. So, Drew, if you're listening, I'll be calling you.
Adrian Bye: He'll be reading the interview. He's on my list.
Alright, so in the beginning when you were talking, you talked about targeting and predictive models, was that what you were referring to the kinds of offer they show you when they're going through the kind of process?
Matt Wise: Yes, we're doing this in two ends, so one is just what consumers have the propensity to pick, so they're just looking down the list and we're looking at them saying, hey, this is a female, 35, they have a higher likelihood of taking the Better Homes and Gardens offer than they do the Sears Sighting offer. However, we also use advanced predictive modeling or backend conversions, so a continuity club may come to us and say, "We're going to share back files with you of who actually converts." So, the first thing the targeting engine does is look at consumers and say, "Who has the propensity to convert for the Book Club?" Then the Book Club passes back a file and says, "Hey, you sent us 10,000 people, 8,000 of them converted and 5,000 of them were fantastic converters," so then we feed that back into the predictive models. So, when we look at a consumer, we say, "Not only are you probably interested in this offer, but you're interested enough that you're probably going to convert well for the advertiser."
Adrian Bye: You're doing score-carding.
Matt Wise: Yes, that's one way to put it.
Adrian Bye: Who do you work with? Who generates the scorecards, is that Fair Isaac or do you work with someone else?
Matt Wise: No, we do it all ourselves. We do that. We also work with some offline data companies, one of which is a company called Alliant who aggregates pay up rates from people like National Geographic, BMG Music Club and others and then they create custom model and scores people, so we have a propensity for people to pay their bills. So, we can apply that to the algorithm but the rest of the algorithms are built. We have a staff of about half a dozen statisticians and mathematicians, so all they do is build the models and advance the targeting.
Adrian Bye: Was that something like you introduced one day to the next, like it would make a big difference when you started doing it?
Matt Wise: So, it's a constant iterative process and what I mean by that is, at first it was just, hey, put different offers in front of men and females and then step-by-step it got more and more complicated to now, it's so complicated, sometimes you wonder where offers are or why offers are showing up. But the targeting engine is smarter than the rest of us in figuring out what people will likely want.
So each time we make an advance, add another data point on or something like that, it raises yields anywhere from usually 1% to 5%, depending on how advanced the targeting element is. It's hard, once you get up here, once you get the easy strikes out of the way, it's hard to continue to find mathematical ways to increase yield. But the specific one I just referenced which was with Alliant – been using offline scoring. Scoring of consumers and consumer groups on offline behavior was actually a big mover because many of those players – many of those magazine companies couldn't make online lead generation work because too many people were signing up and saying, "Yes, put me on the list and then send me a bill," and they wouldn't pay for their magazine once it arrives, so there'd be bad pay up rates. For a magazine or a continuity club, those are the death knoll because they have a huge expense going out the door, not just hey, the consumer signed up and I paid you $6 to get this consumer, but then I put a magazine in the mail and I mailed the magazine, so now, I'm in for another $3 or $4. So the person never pays their bill – I'm really out a lot.
So, our ability to score those consumers and predict this group of people has an 80% likelihood to pay their bill, these people – 90%, these people – 70%. What decile do you want to buy to? That was a huge move in the continuity in the magazine arena. For many players, they just couldn't do business without it.
Adrian Bye: So, that's all in-house, you just find your own statistician. You don't have any like, magic formula or like, what data points are you bringing in? Are you like, looking up things on each user to determine stuff, like do you access any like Equifax databases or things in real time to make sure things work or is it just based on the data that you have?
Matt Wise: It's twofold. We do buy data from Alliant and some other players, not all of which we disclose. Most of it is done behind the scenes prior to the consumer's arrival, so we take consumers and we can match the likelihood of activity, etc. before they arrive. But, we do some things on the fly, like we do behavioral targeting. If a person and our behavior is not they surf somewhere, but they actually took an offer. So maybe they took a fitness offer and therefore, we're going to show them a health food offer next because we know the propensity for them to take one after the other is very high. So, we'll do that stuff on the fly. We'll do a lot of the pay up rates and a lot more of the advanced mathematics behind the scenes, so that when they arrive, we can immediately peg them to a particular user group. But, we can predict ethnicity. I can predict household income, I can predict likelihood to own a home versus rent – a whole plethora of things.
Adrian Bye: Alright, is it the main part of your business, like are there other aspects of your business we haven't talked about that we should?
Matt Wise: Lead generation is the primary aspect, but we use the same algorithms, the same approach in e-mail. So for instance, for a publisher, just like we monetize their registration path, we can also monetize their e-mail and we use the same logic of let's take a look at all the consumers – let's figure out what the best offer to match that consumer is on a given day and the reason that's different from most other players is most players say, "Hey, I'm running (whoever it may be – Allrecipes)…and therefore, I'm going to sell my entire list of Betty Crocker today – all (whatever it is) 3 million people."
Well the reality is, we can see because we see hundreds and actually thousands of offers that mailing Betty Crocker isn't the best thing to everybody, so one sub-segment of that list is going to like Betty Crocker. Well, we can predict that another sub-segment of that list is actually more in the market for Wal-Mart today and that more people will click, convert and generate more yield on Wal-Mart than it will on Betty Crocker. So what we do is when we take a list, we score that entire list the night before and say, "Okay, if I have a list of 2 million, 50,000 get this piece of mail, 50,000 get this piece of mail, 150,000 will get this other piece of mail," etc. and every night, we'd mail all those out, we'd track the conversions. All that becomes intelligence back into the machine to figure out, okay what are we going to send out on the next day, and so…
Adrian Bye: …and how on Earth do you figure out a user is likely to do something when all you have is their e-mail address or do you have more data?
Matt Wise: Well usually, we have more data because when they go through these sites, we ask for the registration data, so we ask for the full dataset, but for some subset of those people, the only thing we have is e-mail, so for those people we have to wait until they've taken an action before we can put them somewhere.
So, if all we know is they're on the Allrecipes site, then my data level is very small and I start sending them mail that I think affects that or appeals to that segment or that contextual site. But as soon as they take an offer, then I start to see some behavioral history behind that consumer and I say, ha, they're starting to skew towards this behavioral group or that behavioral group, and I put them in there and send more like e-mails like that to them.
So, those are the two primary business models and then we're just starting to dabble in banners and the same concept is we're starting to target banners based on this de-breached data we have. So the idea is if you come to the site, you register and if you use Q, we have the registration data. If they went through the registration path, we have the behavioral data from registration. If you're using us for e-mail, we have responses in e-mail and then as they surf around the site, I know it's Joan Smith. I know she's just taken a mortgage, she's just taken a car loan offer and she passed up on the Sears Sighting offer, so now what banner ads should I put up in front of her? So, the amount of…
Adrian Bye: Is that able to generate enough volume? I mean all the different behavioral approach that I hear always have problems with volume.
Matt Wise: Right, so when we buy in the open network, we have the same challenge. When I try to buy in ValueClick or AD.COM or something like that, I have to run across my cookies, so volume is very difficult.
But, where we're a little different is if we handle the registration path for you, then obviously 100% of people who came through registered. All the people who registered I should say, we have data on. So there, the penetration and the overlap is tremendously high as opposed to hey, I collected random behavioral data across the web. Let's see how many of those people run across your site. So, it's two very different approaches.
So, I can't do volume. If Wal-Mart, which they do, comes to me and says, "Hey, I want to buy banners," I can't go out on the Net and say, "Wow, I can find you 50 million people that have these behavioral characteristics." But if you run your own website and you say, "Hey, I want to target people on my website and I want to differentiate based on income, based on even ethnicity, if you wanted, but gender, whatever else. I can do that to 80%, probably, of the people who registered on your site because I'll have all the data on them.
Now, some are going to wipe cookies, so I have to wait until I see them again, so I can recognize them, but outside of that, you don't have a huge overlap.
Adrian Bye: So does that mean that you then want to serve all the ads on someone site? Are you becoming like a Zito? How does that work?
Matt Wise: Yes, we would like to serve the ads on a person's site and then what we'd like to do is say, "Look, if you can sell the ads in at a higher rate than we can, you fill the inventory. If we can find ads better than you can which we think over time we'll be able to because our targeting algorithms will constantly match the right offer with the right consumer, then we take the spot." So, it's a relatively low risk for the publisher because the publisher gets to say, "Look, I've got my own sales force. We're doing a good job. We're generating let's say, an average of a $4 eCPM." We'll come in and say, "Okay, let us have a shot in there, too, and if we can generate higher than $4, we'll take the space. If we can't generate higher than $4, you keep the space." So, there's very little risk for the publisher and hopefully put large potential upside for them as a lot of different advertisers start to come in and not run-of-network advertising that they would get from every old network, but the stuff that would show up is very targeted since it's based off this demographic.
Adrian Bye: What are you doing for advertisers? Are you taking people to landing pages or are you putting them through more registration processes?
Matt Wise: Those players – so off the banners, they could to either to a registration page, for a landing page for a particular offer or just over to someone's site.
So, you know… I'm trying to think of another… Colorado Tourism, they want to drive people back to their site, but they may have come to us and said, "We only want people from the surrounding four states." So most of the time, a contextual site would say, "Well, I can't do that." We, since we have the registration data would say, "No problem. We'll target this ad only to the people in the four states surrounding Colorado." Even a better example, we take offers where people have franchises, and they may have 150 locations around the country and they want different ads in front of every one of them. That's a nonstarter for most ad networks. For us, since we have the registration data, we say, "No problem. This offer will go to the folks in Florida, this offer will go to the folks in Southern Texas, etc." and the CPMs they pay, theoretically will be substantially higher. So, the banner network for us – this concept of the banner network is still in its early stage. We're just starting to test it out now, but from what we can see, it's going to be great.
Adrian Bye: Have you heard about what Frank Addante is doing with the Rubicon Project?
Matt Wise: Not specifically, no.
Adrian Bye: I did an interview with him a month or so ago. He's creating like a network to handle stuff like yours. So basically, a publisher like let's say, FARK can come along and say, "Okay, these six guys want to run stuff on my site. I'll put them all into the Rubicon Project and then the Rubicon Project will manage them all for me and then let whichever one is performing the best continue running, ongoing," because as I understand, it's actually a hard problem for the publishers to keep switching all these different advertisers out.
Matt Wise: Yes, it's tremendously difficult, so there's a player, the Rubicon Project or Right Media Exchange are all players that we'd probably work with, and all they're trying to do is facilitate exactly what we're trying to facilitate which is let's take the burden off of the publisher to have to pick a particular ad network. Let's find some vehicle or a particular advertiser. Let's find some vehicle that chooses the best for me on an ongoing basis.
Adrian Bye: Right. What about fraud? Do you have problems with fraud? In the CPA networks, it's a gigantic problem where people are signing up for offers using prepaid cards and a bunch of other techniques, so they get a $20 commission on a $2 sign up. Do you have that sort of problem?
Matt Wise: Not that much and the primary reason is that the way we run our publisher network is we don't have a self-serve tool that slows moms-and-pops in. We've contemplated it but we just have never built it. So, all the publishers we work with, we have specifically hand-chosen, so the likelihood that one of them would commit fraud is low because obviously they're going to get cut off from us. So, we don't have the same challenges as many of the affiliate networks have where a publisher comes in and they don't worry about getting cut off by and affiliate network because they're only doing 100 registrations a day and if they get cut off one affiliate network, they'll hop to the next, and we don't do a lot on the credit card side. Most of our stuff is soft offers, meaning you sign up for a newsletter or whatever else, so there's not a lot of benefit for the random people to go sign up for $0.50 offers. They've got to fill out literally thousands of them before they make any real money.
However, fraud's still an issue and it's fraud on two levels: (1) You worry about individual publishers trying to goose their numbers by just typing in information; and (2) you worry about un-bona fide users, so the users, for whatever reason… Trust me, sometimes I have a hard time defining why consumers do things. But for instance, we've got mortgage offers which are probably some of the most burdensome offers to get through on our registration path because you have to not only fill out your person-identifiable information, but you have to fill out a whole bunch of other information, financial, so a lot of different questions and then we usually use a triple opt in our mortgage offer, so they get a third time and they get popped up a window that says, "Hey, do you really want to get this offer?"
The way people would go through all that work, go through the triple opt in, submit it and then when they get a telephone call from the mortgage provider, the consumer says, "I never signed up for this." Yet, we have the record of what site they were on – the fact that they went through literally an extra 10 questions. So, we try to figure out and define whether people are putting in un-bona fide information. So for whatever reason, they may feel compelled. They may feel, even though we've written on the top of the page that this has no impact on a promotion, they may feel that well, maybe it does have an impact so I'm going fill out something anyway.
So, we build screens on the background to say, okay, is this really a person? Does this person match the address they're at? Is this Fred Flintstone or is it some other commonly-used and constructed fake name. Often times, we don't even correct that on the front form, like if we see someone who says, they're John Smith from 123 Main Street, we let them submit that. We just take it in and we can it in the background. Now unfortunately, some poor John Smith who lives at 123 Main Street signs up for stuff and never gets it, but those are ones we predict are more likely to be fraudulent than real, so we just take them out of the system and if we've rejected them on the front page then all the consumer does is keep getting creative until they find something we haven't discovered yet.
Adrian Bye: So, you must be getting lots of nice registrations to 123 Fake Street which is obviously a fake address.
Matt Wise: We do get lots of different fake ones. What I find interesting is the people aren't all that creative. When people make up addresses, there're probably 50 that are the top 50 that everybody uses.
Adrian Bye: Okay, are there any funny stories of fake information people have put in?
Matt Wise: They are only funny stories from the fact that you try to come up with all the different variations in advanced and yet, someone always finds something, so I'm trying to think of the more humorous ones.
We had Bart Simpson out there for a while. We finally took that out of circulation. But, we just had one the other day that came in that was some fictional character. I can't remember off the top of my head. It was a superhero or whatever and so obviously fictional, but we just hadn't thought to put it in the fraud filter. So now, it's in there.
Adrian Bye: So, if someone ended up buying leads from Darth Vader, that's when you keep finding those and you have to keep adding those to your…
Matt Wise: Exactly, we have some advertisers who say, "Do you really believe Darth Vader is a real consumer?" I'm like, "Whoops, sorry about that."
Adrian Bye: Alright, so actually, I have one more question for you.
Matt Wise: Yes.
Adrian Bye: The Web 2.0 guys – I mean, they're all out there doing flashy Ajax screens – all sorts of really slick stuff and not making any money. Actually, I was talking with Drew about this just the other day, that there's three business models that I'm noticing on the Internet. There is websites that have a lot of traffic but don't make much money. There's a spectrum of aggressive to not aggressive to very aggressive direct marketers that really work hard to monetize users and sometimes annoy people, but some do it less, like you guys and then there's another end of the spectrum where they do it very aggressively.
Matt Wise: Yes.
Adrian Bye: Those are the two that I can think of and the third that Drew added in was then there's venture back startups that have a lot of money and they're going to run out soon. I'm talking to a lot of people. Those are the models I keep seeing.
I'm interested in how you think about these Web 2.0 companies. I mean, some of them are generating a lot of traffic with their really slick Ajax stuff, one of them being Baidu. Are you familiar with Baidu?
Matt Wise: Yes.
Adrian Bye: I'm trying to get some contacts in there and haven't managed to work it out yet, but I'm impressed with the amount of traffic and the stuff those guys are doing, and they have basically, no monetization on their site. I mean, how do you talk with guys like that, that just don't seem to want to touch any of this stuff?
Matt Wise: Well, so ultimately, everybody is going to want to monetize at some point or another and YouTube's a more popular example maybe that brings to point that what they were doing was aggregating audience and what the web is a little behind in is the ability to monetize the passive audience, and people crap on television all the time and people talk about how great the search engines are, but the reality is search engines are just… Google and Yahoo right now are just the giant yellow book. That's all it is. Consumers know they want a BMW, so they go there to find a BMW and they get the lead, and yes, they make billions of dollars, but those billions of dollars actually aren't all that much. I mean, that's billions of dollars – let's not pick on it, but if you look at overall US advertising spending, it's like $300 billion a year, and overall, television dwarfs the web and the reason is because television found a way – a compelling way to monetize media sites.
So, folks like YouTube, folks like Baidu, folks like even FARK for instance are challenged with the fact that the medium – the Internet hasn't come up with a good way to allow advertisers to talk to consumers without interrupting the flow too much.
So, lead generation is one of those where hey, we're finding a way for consumers to actually raise their hand and say, "I want to be marketed to in the future," but where the real money is going to be in the future actually is not lead generation. It will always be a great niche, but the real money is going to be in, how do I find a way in which I can embed advertising like a commercial on media sites. So, I think right now…it happened in '99 and it's happening again…people are aggregating audience together, feeling like eventually, someone's going to solve this problem for me. So, if I get a huge following – if I'm like Facebook or I'm like MySpace, that eventually someone's going to come up with a paradigm to allow me to monetize these folks, and I think that's probably going to be a combination of video and display, so when I say display, it's going to be displayed on the site, and it's going to have to be a rich media experience, so it's most likely going to be video. But, no one's come up with the right standards and the right implementation now that consumers have accepted in the same way they accepted 30 second advertisements. It will happen. It didn't happen in '99, but it'll happen eventually.
So, I think that's what these Web 2.0 companies are coming up with lots of flashy stuff and lots of cool user experiences are running into is they can aggregate the audience, but until someone solves this display element for them and it becomes commonplace and accepted by consumers, it's going to be difficult for them to monetize. So until then I recommend that they come to me and I'll help them out.
Adrian Bye: So, you don't think your model – the lead generation model will be the one for the long term. You think it's going to be midterm model.
Matt Wise: I think there'll always be a place for lead generation. I mean, the concept that someone goes and raises their hand and says, "Yes, I want to receive more from this advertiser," is tremendously powerful, so there'll be a place for it. But, you don't raise your hand and say, "Yes, I want more for BMW until you've seen a BMW commercial – until you've been persuaded that BMW is the ultimate driving experience and you feel like you want to learn more about it.
So, to do all that, I can't do that in a form. I need some rich media experience where I can show a car driving down the force and I can get a feel for the Italian leather that's going to be on the inside or whatever else. Until I get that rich branding experience, I can't drive desires. All I can do is answer desires when people already know about BMW and I say, "Hey, you want to get a brochure for it?" and they're like, "Oh, yes. I saw a television commercial for that the other day," – that's where we work and quite frankly, that's where all the web works right now with the exception of some of the viral stuff and some of the display stuff. But, generating interest is very difficult and the web has not mastered it yet which is why it remains a $20 billion segment instead of a $100 billion.
Adrian Bye: …and at some point, that's going to be cracked and where it will become a $100 billion segment.
Matt Wise: Yes, I have no doubt about it. People always complain about the fact that consumers spend more time on the Internet, but the Internet's not rewarded for that and I think that's the wrong way to think about it if the reality is people spend more time on the web, but the web hasn't figured out how to present advertisers with the ability to advertise there successfully and I think Drew's reaction to wow, I'll have a user revolution is the challenge. People have now gotten used to the web being in the format it is and advertising being secondary. So, until we sort that out with the consumer and find the right experience that they won't have a revolution about, the web will still be hampered – the web will still get a short on advertising dollars relative to other mediums.
Adrian Bye: Very interesting. I know we're running out of time here. Is there anything else you want to add in closing?
Matt Wise: No, you know I appreciate the call. I think you're organization's a very interesting one, so I look forward to more interactions. Lastly I'd say, just from our perspective, we see this tremendous growth out there, both in the interactive space in general and our lead generation space, but the key is to do it responsibly. So finding that right balance of feeding in advertising, but not disenfranchising the consumer and concerns about user revolutions of FARK and stuff like that, I think consumers have a greater than most people appreciate for advertising as long as it's their option and they can delete it when they want.
Adrian Bye: Very good.
Matt Wise: That's what I'd close on.
Adrian Bye: Thank you very much for your time.
Matt Wise: Adrian, it's very good talking to you. Thank you.