This panel discussion took place at February's iMedia Brand Summit in Florida. Doug Weaver, President, Upstream Group, moderated. Panelists were Gerard Broussard, Senior Partner and Director of Media Analytics for MOne, and John Nardone, Executive Vice President - Product Development and Marketing at Marketing Management Analytics (MMA).
Monday, the group defined Marketing Mix Modeling. Yesterday, they contemplated interactive's effect when it gets the smallest portion of the spending Here's more:
Doug Weaver: Okay. We're going to go to audience questions in just a minute.
I have one more topic I want to bring up, but if you have a question please get Rebecca's attention or Rick's attention. I'm going to ask the $64,000,000 question now. You've both had experience with the internet as a part of the total marketing mix. You've seen a lot of modeling where the internet was at least a part of that program. Does the internet work within the marketing mix, and how? Let's start with you, Gerard.
Gerard Broussard: Well, I will give you that example that I spoke about earlier. With this financial advertiser we saw that it was driving a large part of the advertising effect, along with television, because there was about a 40 percent investment. Now, stripping away all other media -- we did have one example recently of one of our clients just advertising only on the internet. And we ran a regression on the sales of the product, which was by the way, a high consideration product, which took maybe a couple of sessions going back and forth to find out more information about it. And we found that there was a direct correlation between the internet advertising and the driving of sales. With a lag effect, which means there's a delay in advertising of about four weeks for this product.
Weaver: While you're doing explanations, can you go back to regression for a second, help me out with that.
Broussard: Yeah, regression is simply putting two or more streams of information together. Or we have one variable called the dependent variable, it could be sales. And then we have all the other things like: media spend, pricing, promotion, seasonality that are tracked for several -- or actually a multitude -- of weeks, over a couple of years, in many cases. And then we attempt to see if there's relationship between those variables and the sales.
Weaver: Got it.
Broussard: In this case, we had a 12-week window. And there was no other media in market. It was just internet advertising. And there was a direct relationship between the internet advertising and the sales of the product. So, we know that it's working. Now, let's say if they layered TV over that, it might obscure the internet's effect or it might make the internet more powerful. It's a hard question to answer.
Weaver: So, just to go back a second, I want to make sure that I've got all the facts straight here: In this instance, we're not talking about a direct response kind of campaign, but you're simply saying that in the real world, sales went up by …
Broussard: Driven …
Weaver: … in correlation to how the internet spending was?
Broussard: That's right. They were driven to retail outlets where there's demand for the product.
Weaver: Okay. So the online advertising/offline sales model panned out in that case?
Broussard: Yes.
Weaver: So in that isolated instance, we could say that the internet improved sales.
John, what else does the internet do within the marketing mix and does it work?
John Narbone: Well …
Weaver: Who does it work for?
Narbone: Well, the first answer is yes, the internet works for marketing, like any other media works. And in some cases, in some situations, with some copy, for some industries it works better; and in some cases, not as well. The thing about marketing mix -- assuming a way for a moment, enough spending to pop -- is that it does put all the media on an equal footing. And it's not unusual to do a marketing mix analysis where print comes out as not working particularly well for a client. And sometimes internet comes out as not working particularly well. Sometimes it comes out working really well.
We have a tendency, because we're very internet-centric certainly in this audience, to say, well, if it didn't pop in that marketing mix, then it's the methodology; it's not giving it a fair shake. Well, if print didn't pop, nobody would be saying, well the methodology didn't give print a fair shake. Our collective learning across doing this for lots of clients is that sometimes the internet is a great driver, and sometimes it's not. And if it's not, we need to do the due diligence and the research to say, "Was the creative not very good?", "Were we not targeting correctly?" or "Does it really not fit into the consumer's purchase process in a way that's really going to move the needle?" Sometimes it works; sometimes it doesn't.
Weaver: Sometimes the answer is that the internet may be working but we can't see it because there's not enough spending …
Narbone: Absolutely, that's a possibility.
Weaver: Within the pipe …
Broussard: Yes.
That doesn't mean it's not working. In fact, I eluded to before that some of the work that's been done by Yahoo! -- which is absent of other media effects -- definitely isolated sales effects for the internet, which I thought was pretty conclusive about it. It also depends on what the objectives are for the internet and offline media. If you've got an offline media objective that's consistent with the internet objective, there might be a better chance at measuring the internet's effect than when the two are doing different things.
Weaver: Yes, exactly.
Got a question in the audience?
Audience Question: [Inaudible question about methodology]
Broussard: The methodology definitely began in packaged goods. Actually Kraft pioneered the methodology and it spread to a lot of packaged goods marketers from there. But today, our business is about 60 percent outside of CPG and it's concentrated in retail, pharmaceuticals and financial services. And the reason why it's concentrated in those industries is because those industries tend to have good data. So if you think about retail, we can take the data from their point of sales systems and that gives us a very rich data set to model. For pharmaceuticals, prescriptions are tracked by regulation. Every time a doctor writes a prescription that's logged, so we've got a very good data stream from that. But the methodology is applicable for any industry as long as you've got a good data set to model.
Weaver: So retail, finance and pharma?
Broussard: … and pharma tends to have very good data.
Weaver: Which all happen to be very hot categories in the interactive space and industries that are …
Broussard: Anything that's trackable, where there's a lot of customer information, you're going to have a good investment on your return.
Weaver: So even though this practice began with CPG …
Broussard: Yeah.
Weaver: … which arguably has some of the lowest customer information flow ...
Broussard: Yes.
Weaver: … it's modeled that way. That's interesting.
Broussard: Actually CPG, because you have Nielson and IRI, those provide a pretty good basis for modeling.
Weaver: Gotcha.
Another question over here?
Chris Theodoros: Hi. I'm Chris Theodoros with Google. I wanted to ask both of you about your thoughts on the implications for Ad-ID, the ANA and AAAA's initiative to identify advertising by unique identification. Could that be helpful in the cross media evaluation of advertising? Or if you have any opinions about the project.
Broussard: I would say anything that can provide a direct link between the consumer ad experience and a consumer reaction would be desirable. So, not only is Adlink important, but things like the nature of Apollo …
Weaver: Well, can we…? Let’s stop a second here, because as we all know, Chris is a step ahead of most of us here. So can you go back for a second, let’s talk about Ad-ID, Adlink, Apollo. Can we just … you guys quickly just spell out what these initiatives are so that we can kind of catch everybody up, and then go to the answer.
Broussard: Chris, you want to do that?
Theodoros: I'll have Barbara with the ANA do that.
Barbara Bacci Mirque: Hi, I’m Barbara Bacci Mirque with the ANA and we along with the AAAA's are driving Ad-ID. And really, in its format, it's just a unique identifier for one's advertising asset that can be used in any medium, from television to online to floor tiles. That's the very simplest reason for it. We took the old ASCHI system and converted it into something that's digital. So as we get into some of these new media that we're learning about, that Joe talked about, you can track your advertising asset. It will facilitate other things that may do some of the consumer linkages. It's just simply that unique identifier that can be tracked digitally.
Weaver: So Barbara, this is an ID that lives with the creative, essentially, digitally?
Bacci Mirque: Yes, with the advertising asset. So if you're getting into things such as addressability, et cetera, and you're doing 10 or 15 variations, it's a simple way to keep track and make sure that the right advertising asset is run, at the right place: if it's online, television, radio, et cetera, et cetera.
Weaver: Okay. So that's Ad-ID. Apollo?
Broussard: Apollo is an initiative, most hardily pushed by Procter & Gamble, to link single source data, which means basically, that you're trying to link advertising exposure with consumer action, meaning sales. And that would involve getting a panel of people, you know, in the United States, where you know your ad exposure and then somehow you can tell what impact it had on them. Then you can close the loop between advertising and sales. (Editor's Note: Read an op-ed by Dave Smith where he examines Project Apollo.)
Weaver: Okay. Now back to Chris' question.
Narbone: The simple answer regarding Ad-ID -- anything that gives us better data is a friend of marketing mix modeling. And just from an operational standpoint, Ad-ID will make our job easier because we have to go through by hand and look at all of the spots and sort through what the key messages are and essentially make them into the data. And that's actually a pretty time-consuming process. So an Ad-ID system would make that job a lot easier for our analysts and make the process go more smoothly.
Broussard: Building on John's point, there's not really a lot of work done -- or there's some done -- on creative effectiveness in market mix modeling. It could provide some key learning by identifying which creative worked and which did not in market mix modeling so they could be better the next time they're out in the marketplace.
Weaver: So Pepsi would know whether the …
Broussard: Which execution …
Weaver: … the truck or the Tom Brady commercial …
Broussard: Right.
Weaver: … potentially worked better …
Broussard: Yeah
Weaver: … stimulated the effect, even though those two commercials might be in the marketplace at roughly the same time.
Broussard: Uh-huh. Correct.
Weaver: Okay, great.
Tomorrow: Wrapping it up with more audience questions.
