Executive Vice President, Americas
Pat LaPointe, executive vice president of the Americas for MarketShare, spoke with eMarketer’s Lauren Fisher about how the company uses an extensive amount of first- and third-party data to build client models that provide insight into current and future cross-platform ad performance.
eMarketer: Tell me how MarketShare helps media buyers measure their cross-platform ad campaigns.
Pat LaPointe: We provide a combination of software and services that allows us to capture all of the historical data around what media and advertising was delivered, what was seen and what was paid for, or some combination of those things. From there, we put this ad performance into context alongside everything else going on in the marketplace such as weather, economic issues, competitive activity, changes in distribution patterns or product launches.
In doing this, we help clients identify the real relationships between each ad channel and also illustrate how the whole component of the marketing mix is influencing bottom-line business results.
eMarketer: How has your approach to cross-platform ad measurement changed over the last year or so?
LaPointe: In 2010, our models were averaging about 20 variables, but today, it’s in the hundreds and even thousands. Many of the new touchpoints and channels—especially in digital—have an inherent data capture associated with them. As a result, most of our clients are sending us terabytes of data at this point. Three years ago, we were in the gigabytes range.
“Most of our clients are sending us terabytes of data at this point. Three years ago, we were in the gigabytes range.”
When you think about the fact that in order to accurately represent that analytically you have to have models that test each point-to-point relationship, you can imagine we’re now dealing with an almost exponential increase in the number of models that have to be created.
We’ve also had to make big investments in building our clients an even greater number of filters through which to see their campaigns. There’s never any one, single truth in terms of what’s going on in a marketplace, so companies need filters they can apply to see slightly different versions of the truth.
eMarketer: Can you walk me through an example of how a marketer would use this data to make an informed advertising or marketing decision?
LaPointe: Prior to Hurricane Sandy, we had a lot of clients who were waiting out the election season and not spending advertising dollars on television due to inventory constraints, high prices and the fact that a lot of advertisers feel the TV environment is poisoned during the election season. They don’t want to have a positive ad sitting between two negative political ads, so a lot of advertisers just pull their money, even though the lack of advertising usually has a negative impact on their sales.
So right after Election Day, many were eager to charge back in and run advertising. When Hurricane Sandy hit, it seriously affected the advertising model. We have a lot of clients who operate extensively in the Philadelphia to Boston corridor, so naturally we went into the model and simulated what the impact of the hurricane would be on buying behaviors over the next 30 days to give them an idea of what this would mean for their ad spending.
Given the outcome, we advised a lot of them to pull money out of mass media activities and refocus it instead on some of their more customer-specific digital and direct marketing tactics. This allowed them to continue to engage those people who were still in consumption mode vs. those who were largely in survival mode.
Now, you don’t need a model to tell you that’s a good idea, but what the model did tell us was exactly how much money we should move from X to Y and what we should expect the impact on sales to be as we move it.
“When Hurricane Sandy hit … we advised clients to pull money out of mass media activities and refocus it instead on some of their more customer-specific digital and direct marketing tactics.”
eMarketer: Is there such a thing as collecting too much data?
LaPointe: There’s a tipping point where you go from being comprehensive to being impractical, yes. However, there’s also a threshold beneath which you’re not looking at enough variables for the analysis to be credible.
In the old days, people used to build media mix models for which they basically just input the media spend and impressions and ran regression analysis against sales.
That does a nice job of telling you which media seems to have a better effect on sales than other media, but it does nothing to give you a holistic perspective on how marketing’s overall impact is helping the business. It doesn’t take into account other outside influences like product launches or sales compensation or social media, etc. In order to be really comprehensive in managing that ecosystem, we need to take into account whatever we can get our hands on.
A longer version of this interview is available to eMarketer corporate subscribers only. If you’d like to learn more about becoming a corporate subscriber, click here.
Check out today’s other articles, “Worldwide Ad Spend Grows Steadily, Bucking Economic Slowdowns” and “Apps Most Popular for UK Mobile Search.”