Attributes of Attribution: Q&A With Paul Pellman, CEO, Adometry


It’s generally sunk in that the last click shouldn’t be getting all the conversion credit – as digital spend increases, advertisers are yearning for a better understanding of multichannel campaign performance. Properly assigning credit to various ad types has proved to be a task, but one that companies like Adometry – which just released version 2.0 of its ad analytics platform, complete with fancy new attribution tools – have chosen to take on. We caught up with CEO Paul Pellman and demanded some answers about attribution as well as Adometry’s audience verification and tag services.

Why has attribution become such a hot-button topic now?

Because there’s growing interest among advertisers and agencies to understand what truly contributes to the success of online campaigns. They’ve always known that giving all of the credit to the last-click or search is a flawed approach, but they haven’t been able to reliably identify the real drivers. This uncertainty makes media planning a guessing game.

With the scientific approach we’ve developed for our attribution offering, we can model not only which activities are driving results, but can show you how different marketing efforts interact and make real recommendations on what you can do to improve results. It’s like taking the red pill from “The Matrix” and seeing the truth for the first time.

What are the biggest misunderstandings out there regarding multi-touch attribution?

There are two major misunderstandings with regards to multi-touch attribution. First, while buyers recognize the importance of looking past last-click attribution, there isn’t a lot of awareness of the major difference between a fixed attribution model, which arbitrarily assigns fixed credit to each campaign variable, and a scientific model that algorithmically determines the value of each event/variable. The fixed attribution model produces subjective campaign recommendations because inherently they’re just guessing at the value of each channel and campaign variable. The scientific model is completely objective in assigning value to variables, because it uses data-driven algorithms to determine the value of each event. While each model produces drastically different results, they both refer to themselves as “multi-touch models,” which causes further confusion among online marketers.

The second misunderstanding about multi-touch-attribution is the assumption that attribution by itself is enough. Media buying is a complex task, with multiple channels and multiple buys contributing to a desired outcome. Because of the interactions of events, you can’t just find the highest converting traffic source and simply purchase more of that. You also need to use an optimization model that takes into account all the different interactions and finds the optimal mix of media buys.

In the new version of the Adometry Analytics Suite, there are several new advanced attribution features – which are the most exciting? What’s never been done before?

We get really excited about the ability to take the results of our attribution models and look at them in different ways to get entirely new perspectives on how online advertising works. For instance, we’ve applied our modeling to activities like frequency analysis to make frequency capping recommendations. That is, we allow the marketer to see various site overlaps and understand how often to show ads before they start to have little or no further impact on conversions. Or, using that data to find which combinations of ad buys work well together and which are really redundant. It provides a whole new way of looking at results, and helps media planners make decisions with confidence and no more guesswork. That’s exciting to us.

In December, Adometry partnered with Acxiom, Datalogix and Nielsen to give clients access to offline consumer data. How are these partnerships being used to soup up Adometry’s offerings?

Don’t forget TargusINFO as well, and we are continuing to work with more partners to bring the best data into our platform. There are two main ways we are using data from these partners. The first is for basic audience verification. If a client is trying to hit a certain audience, whether it is demographic segmenting, interest-based segmenting, or in-market segmenting, we can show a client which audiences they are hitting across their campaign.

The second is to feed that same audience data into our attribution engine to identify not only which audience segments perform the best, but to see which combinations of buys work for each audience. We also provide financial data to help customers make decisions about the value of the audiences they’re targeting for their campaigns. We find that bringing in audience data is very complementary to our attribution modeling.

Is audience verification for real? Can you give us a quick explanation on how it works?

Audience verification is very real. All media buyers are buying audiences to some degree, whether they are choosing publishers based on their audiences, or using audience targeting. What Adometry provides is a way to monitor what audiences are actually being reached with each online campaign. As we collect ad verification data, we match that with anonymous segment data from our partners to show what types of users were reached and on which sources. When you marry this with ad viewability, our solution provides powerful insights — you can see not only the audiences you’re reaching, but those audiences actually seeing and engaging with the ads.

How is Adometry’s TagScan product different from a tag management system? How can they be used in tandem? How does your partnership with Ad-Juster enhance the product?

TagScan is designed to help publishers, while a tag management system is designed to help advertisers. TagScan is an enterprise-class tag quality monitoring tool designed and built specifically for web site ad operations teams. Whereas tag management systems are designed to help advertisers and marketers ease the deployment of tracking tags on a website, TagScan is used by website publishers to continually test those ad tags and monitor them for ad specification violations, data leakage, malware and more. The two products are also complementary. We have very good relationships with the major tag management systems, and have worked together on several clients.

Our partnership with Ad-Juster is a data integration level partnership that has proven very beneficial for both companies. Ad-Juster does a fantastic job of integrating with publisher ad servers to provide delivery reports that save ad operations professionals a lot of time and are great for monitoring reporting discrepancies. When integrated with TagScan, Ad-Juster can provide TagScan a feed of all ad tags running on a site for testing against site policies, and TagScan can report back performance data to Ad-Juster to help identify the source of discrepancies.