Brands have ramped up their digital activities, and most are investing in them. But many are lagging when it comes to attribution, judging by a new digital maturity survey by Adobe.
Of the 735 marketers polled in the U.S. and Europe, 74% employ Web site analytics and 53% use A/B testing for measurement. But you have to scroll far down the list to find attribution modeling and algorithmic attribution — in fact, the latter is used by only 10%.
At the same time, a mere 19% of the respondents rate their firms as “advanced” on the digital maturity scale. Yet 79% say are at the “focused” stage, meaning they use automation and are doing the right things. To learn more, we interviewed Jeff Allen, senior director of product marketing for Adobe Analytics.
The survey reports very low numbers for attribution modeling and algorithmic attribution. How would you explain that?
Attribution modeling is used by 18%, but many plan to use it. And
algorithmic attribution is at the bottom. But again, it’s like diet and exercise. Everyone knows it’s an important thing. One of the big challenges with algorithmic attribution is that it
lets you take your hands off the dials. There’s no bias in the model by someone saying this channel should be favored in some manner. In rules-based attribution, the rules are hard-coded,
opportunistic things that introduce bias into your model.
advertisement
advertisement
So what should brands be doing?
In the past, most attribution plays have been focused on streamlining the media
mix and investing to drive better ROI on your spend. But if you pull the camera back, you see a deeper picture, and the other channels come in. When do I email them, what should I say, and how much do
I invest? What should I do in the retail environment, and what should I say in direct mail? What should be in the mobile app, and how do I drive them to download it?
How do
companies measure the customer journey?
They use statistical modeling to determine the value of any input and follow the customer journey from point A to point B, B being the goal, A being
the point wherever they start. If they’re trying to understand what causes people to fill out the lead form on your Web site, that’s B. And their attribution would look at paid
advertising. But if they stretch B to a purchase, they may have to measure paid, earned and owned channels. There are more channel inputs, and the model is different. It’s a lot harder for most
organizations to decide the contributing factors to this outcome
Are attribution models geared to this?
A lot of the time, the attribution model is asking which channel is
performing better. But how much do we weigh the channel vs. the ad copy itself? If we’re not rigorous in the way we apply the scientific method, it starts to muddy the results.
How good are brands at mobile attribution?
It mirrors what happened with Web sites in the mid-’90s. They were an IT project initially. Then most organizations moved it under the
chief marketing officer. There is now an evolution from the mobile app as a tool to the app as a channel. The mobile apps have been born outside of marketing, and have not had all the business and
digital marketing rigor. But brands are moving quickly to adapt marketing best practices.
Does attribution affect content?
There’s content in every
attribution. You have copy. You have imagery. Did the motion graphic perform better than non-motion graphic? Did video work better? It’s about landing pages, and how we might decorate the retail
environment, or how an outbound salesperson might make an offer.
What would you say in conclusion?
We’re seeing a great desire on the part of brands to drive less by
gut instinct and more by the light of data. And people are recognizing that there is a clear maturity continuum in the use of analytics. The more you stay on the low end of that, the less ROI
potential you have. And further up, you unlock greater degrees of ROI potential.