Ken Mallon, Dynamic Logic
Ken Mallon
Senior Vice President-Custom Solutions and Ad Effectiveness Consulting
Dynamic Logic
Dynamic Logic’s nearly decade-old MarketNorms database has become the de facto standard among marketers, agencies and publishers looking to measure online advertising and marketing effectiveness across multiple consumer products and services categories. The company also offers consulting services and custom research.
As more marketers seek to prove the return on investment of online marketing, Ken Mallon, as head of the company’s custom solutions team, is on the front lines advising clients. Prior to his current role, he worked in product development at Yahoo! and created its data-mining group.
eMarketer: What are the most effective methods of online brand measurement?
Ken Mallon: Obviously I’m biased because Dynamic Logic does online brand measurement. There really aren’t a lot of different methods. The method that we use, which is real-time Web intercept within the media footprint where we identify control and exposed groups, is pretty much the standard.
We track who sees and who doesn’t see the ads, so that enables us to see who’s exposed and who isn’t. We have the control group, we have the exposed group and then we do a live Web intercept where we intercept them as they’re searching the Web within the media footprint. We invite them to take a survey and we ask branding questions.
There’s also the panel-based method. The main limitation of using a panel is that you have a much smaller sample size, so you need to have very large campaigns in order to find people within a panel who are exposed to it.
Because we’re intercepting people when they’re actually surfing the Web, we sample proportionally the people who do more Web surfing than the ones who don’t. Whereas if you’re doing a panel, you’re sampling people who maybe went to those Websites, but you may or may not be taking into account how much they were on those pages.
eMarketer: In what case is a panel-based method best?
Mr. Mallon: If you do a very large campaign and measure it after the fact, then a panel is pretty much the only choice you have.
eMarketer: Is all audience measurement inherently flawed? Do existing methods actually reflect that consumer watched a streaming video or TV show or saw an ad on a Webpage or in a magazine?
Mr. Mallon: When people talk about online measurement, they’re really talking about two different things—audience measurement is different from ad effectiveness measurement. Audience measurement is really the Nielsens and comScores of the world telling you what the traffic is to a particular site and what the profiles are. That area is very hotly debated.
In the audience metrics world, online ad effectiveness is a hot topic. The central debate is about the projectability of the sample. It’s hard to have a representative sample of online browsing. The reason is that people use the Web from many locations—they might use it from home, work or they might use shared systems at universities and public libraries. There are different estimation schemes for determining the number of unique users.
eMarketer: What is the solution to resolving that particular dilemma?
Mr. Mallon: Dynamic Logic tries to stay out of the fray on that. We’re really interested in measuring ad effectiveness independently of how big the audience size is. The two main sources of online audience estimation—comScore and Nielsen—have similar numbers.
We have found that our clients—big publishers, agencies and marketers—typically subscribe to both. We’re interested in measuring ad effectiveness independently of how big the audience size is.
eMarketer: How does a marketer successfully tie online advertising to offline sales?
Mr. Mallon: I used to work at Yahoo! and while I was there, we developed a system called Yahoo! Consumer Direct, a partnership with AC Nielsen. It created a joint panel of people on Yahoo! who were also on the AC Homescan Panel.
They record everything they purchase after they shop. We measured everything from actual exposure to actual purchase. It was limited to Yahoo! and it only analyzed grocery items.
The other method is market mix modeling. Market mix modeling is econometric modeling to measure the impact of GRPs or impressions delivered against certain forms of media. Then you collect aggregate sales data and do modeling exercises.
Typically, online advertising, because its share is so small, is very difficult to pick up with this technique. But many people still try to do it, particularly with campaigns that have a high online reach.
Dynamic Logic is working on a system that we think is going to be a big improvement—I can’t really talk about it until it gets released. It will be beta tested this year and we hope to release it early next year.
This is a system that would be limited to grocery items. The system won’t rely on a non-random subset of the population of grocery shoppers or people who go to certain stores for certain things and who happen to have a frequent shopper card.
We’ve developed a method to track through to what we call “stated sales.” So it’s not actually picking up a transaction, but we’re asking people, “Did you buy this?” That’s a response to the fact that recall-based tracking doesn’t always work well for grocery items because people often don’t remember the specific brands they purchased.
eMarketer: In what client categories are you testing this system?
Mr. Mallon: We’re testing it in the automotive category with two clients and doing some in the consumer packaged goods category. Certain food brands seem to resonate with people. They remember buying Kraft Mac & Cheese versus another brand or a store brand.
eMarketer: What is the single biggest problem with online brand measurement today?
Mr. Mallon: Recruiting people for the brand impact studies. We probably execute about 1,000 of these projects a year.
Safecount [a WPP company] is trying to make improvements in the online recruitment space to facilitate live data collection from the Web. It’s an industrywide initiative. We want to create better recruitment practices for people to participate in online surveys and panels.
eMarketer: Looking at the entire consumer purchasing funnel, from awareness to final purchase, where is the measurement model most broken? What pieces of the puzzle are missing?
Mr. Mallon: The hardest piece is as you get down to the bottom. The hardest piece is the last step. It’s real easy to measure awareness. We measure ad awareness, brand awareness, message association and brand purchase intent. The hard part is what happens after someone makes the intent to purchase.
What happens after someone raises their hand and says, “Yeah, I do intend to go to that movie. I do intend to buy that car in the next 90 days.” How does intent translate into actual purchase? The system that we’ve developed is a system that translates that intent into actual stated sales using a longitudinal approach that asks people what they bought. We’re also pursuing partnerships that are going allow us to directly link to transactional-based data.
eMarketer: Do you have a name for the system that you’re testing?
Mr. Mallon: We don’t have a formal name for it yet but we’ve referred to it as a conversion index, because we’re basically trying to index translating a purchase intent to an actual purchase.
eMarketer: What are the measurement challenges with respect to online video? How can they be solved?
Mr. Mallon: Video represents two main kinds of challenges: One is the tracking from a technical standpoint. The other thing is there’s a proliferation of different video formats.
We’ve worked scalable ways to track online display ads and Flash ads and other rich media, but video is technically different. There are special ways to track the exposure.
With the different video formats, you’ve got things like ticker formats, in-page units and preroll. And you have proprietary video players. Some of the video campaigns have 100% share. If you show up on a certain page you’re going see that video ad. And that makes it difficult to get control samples.
We’re mostly in the business of measuring the branding impact of online video. Clients are also very interested in interaction rates not only for video but all rich media formats. They want to know if someone expanded the ad or not. For video, did they turn the sound on? Did they stop it? Did they rewind? There’s a lot of different interaction and engagement types of metrics.
eMarketer: What is/are the most important metrics for online video?
Mr. Mallon: You’ve got reach-based metrics—the only ones used in TV. In the TV world, it’s mainly about reach estimation. In online, it’s how many times was a video seen and by how many unique users. We’re measuring the branding impact of ads wherever they occur. It’s not just a reach issue. It’s about impact. Most TV ads are measured by unique user reach estimation.
eMarketer: Is the Internet destined to be viewed mainly as a direct response medium?
Mr. Mallon: Absolutely not. I’m seeing big advertisers increasing their spend online and it’s largely for branding. My view is that the Internet is like a magazine that’s very well-targeted. People are consuming content that they enjoy and they’re seeing ads.
The performing ads online tend to be ones that have an almost magazine feel. They look good. The more direct response-oriented online ads that have flashing lights and “buy now” or “click here” perform poorly in terms of branding.
eMarketer: What is the potential game-changer for online brand measurement?
Mr. Mallon: There’s are two pieces—metrics and new platforms. People want the online behavioral data and offline sales incorporated. We know that and we’re working on it. We have custom tools for that, but we’re working toward more scalable solutions.
The other piece is that the platform itself is changing—look at all the user-generated content, blogs and video. So the game-changer is really understanding and measuring next-generation media platforms. We’re working with companies that measure word-of-mouth and blogs so that we can combine that with our branding data.
eMarketer: Should the industry be creating a standardized online brand measurement system?
Mr. Mallon: I don’t think so. I’m not in favor of a standard because I think there are different needs for different clients in different circumstances. Competition is good. And we think that our system for measuring brand impact is the standard. Competition pushes us to do new things and to take into account the impact on brands of social networks, mobile and other new media and to analyze behavior.
