How to Tackle Attribution Analysis

attribution analysisMarketers are abuzz over attribution analysis, which provides insight into how their ad channels are working—individually and together—to accomplish online marketing goals. Alan Osetek, president of digital marketing agency Resolution Media, an Omnicom Media Group company, says few marketers understand what attribution analysis is. He spoke with eMarketer’s Lauren Fisher to share his insights on attribution analysis and modeling, particularly for search and display.

eMarketer: Before we jump into speaking about attribution analysis for search and display, can you share with me some observations about how marketers are currently using search and display?

Alan Osetek: In the last 12 to 18 months, there have certainly [been] a variety of new ways of looking at display as a behavioral demographic unit, from a search retargeting perspective or by using cookie-based information to figure out where people have been, what they are doing and what they’re in market for.

We see marketers integrating the two in a couple of different ways. One is by looking at real-time bidding inventory, or trading-desk inventory, to take advantage of looking at the interaction of search and display together, from a cookie perspective.

The second is what I call the beginning stages of search and display attribution, where marketers are building large-scale data warehouses of consumer information and habits. Once you collect and analyze this information, you’re then able to target these consumers with offers and promotions, particularly when they’re in market for products and services.

eMarketer: For marketers who are looking to implement attribution analysis or an attribution model, where do you suggest they start?

Alan Osetek: A lot of times, people start to build an attribution model designed to be a complex mechanism for understanding how to optimize across all channels.

 

“It’s better to start small and build from there. Pick channels like search and display, and pick a very specific business problem.”

 

It’s better to start small and build from there. Pick channels like search and display, and pick a very specific business problem. From there, you can layer in organic data and after that, maybe site-side or customer data.

eMarketer: How do marketers determine appropriate attribution weightings for each of their channels?

Osetek: A lot of times, companies say they’re doing attribution analysis, but they’re just assigning equal weighting to channels, 50/50 or 33/33/33, and that’s not the right approach.

But, if you’re able to create a model where you use all the online data available across channels, over time, historical data can help the model learn what each weighting should be.

 

“Each model differs per client. So, there’s no real right or wrong percentage of attribution for each channel, it’s really building the model based on historical data to see what works and what doesn’t.”

 

And weightings can be assigned based on the ability of the model to look at the acquisition data to see what worked and didn’t work. Each model differs per client. So, there’s no real right or wrong percentage of attribution for each channel, it’s really building the model based on historical data to see what works and what doesn’t.

eMarketer: Do you think attribution analysis has gotten easier for marketers over the past few years?

Osetek: Absolutely. First, there are more companies that specialize in helping marketers to extract data. Second, more people have opened their eyes to data sources out there like Google, Yahoo! and others. Third, more digital marketers know the complexity of these models and know how to work on these types of problems.

Even though it’s gotten easier, the main problem that still remains—and I think it’s one that won’t go away for quite some time—is just being able to get at those datasets, because they usually reside in multiple places within a company. So, if you’re trying to pull site-side data, you may have to get that information from the IT group.

Then you have to get the display, the search and the customer data. In most large corporations, getting access to that data is difficult. Other departments aren’t always willing to give it to you. If you are able to get the data, it’s often clunky, cumbersome and not normalized across channels.

It requires a lot of upfront work to collect and normalize the data, but once you do that and build the model, it gets easier.

 

“Another challenge with attribution is that, even though everyone is interested in it, not everyone is willing to pay for it.”

 

Another challenge with attribution is that, even though everyone is interested in it, not everyone is willing to pay for it. Solving a business problem using attribution may take 4 to 12 months, so attribution carries more of a long-term ROI that’s not always easy to get budgeting approval for.

Post from emarketer.com

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s