The Challenge of Adopting Conversion Attribution Modeling

I recently had a conversation with a former Atlas colleague, someone who has extensive industry experience. We talked about conversion attribution—what the right conversion modeling looked like, what the right technology would be to support it (yes, my answer is still our own Engagement Mapping). And then he asked: Why do you think most advertisers are still not measuring and optimizing using some sort of attribution model? This question led to a surprisingly great discussion covering a lot of challenges we face from technological, industry, and strategic standpoints.

I have a lot of initial responses to why I think many advertisers have been slow to adopt conversion attribution as the method for optimization. What I think is clear: no one can argue that measuring and optimizing using last ad methodology is flawed. Almost everyone can agree that using some form of conversion attribution is the smartest way to think about optimization. And yet, how many advertisers are actually using conversion attribution as part of their inherent measurement processes across campaigns? Why has it been so difficult to shift the mindset in practice as an industry? Some quick takes below:

1.    Uprooting established success metrics is difficult.

If you’ve been measuring success by last ad, there are CPA goals, key performance indicators (KPIs), benchmarks, etc. that have been well-established. Agencies particularly have a difficult time telling their clients:
a) We’re starting from scratch on data measurement, and no, we can’t compare this year’s holiday CPA to last year’s, because we’re measuring it differently.
b)What we were doing before isn’t “wrong”…we just think this is better. But we’re not sure the attribution model is actually the best…yet.

This conversation is not the easiest to have with a client, and is seen as a big gamble with their data measurement.

2.    There is no standard. Conversion attribution is variable.

Depending on the technology solution and data crunching methods of arriving at success metrics, there are many ways to slice and dice the data to take multiple touch points along the funnel into consideration.

Take Atlas’s Engagement Mapping. What I still bet the farm on with eMapping is the fact that the reporting results are actionable. Our data processes allow for a single number output—a fractionally weighted conversion—based on variable inputs like ad size, frequency, recency, rich media engagement, etc. With a single number—conversions, or CPA, you can replace your existing optimization with a new number that takes all the touch points into consideration. So if you were optimizing off a single number before, you can continue to look at a single number—it’s just not based off last click.

However, take another technology—one inherent to an ad server, or an independent attribution technology vendor. Most of those solutions provide log files of the last 10+ events (impressions or clicks) in a user’s cookie history leading to a conversion. But what are you meant to do with that data? How do you interpret the data, and what meaningful conclusions can you draw? How does a user’s history play into your optimization decisions?

No matter what solution you are using, there is variability in how you analyze, process, and action the data you’re leveraging. Compare this variability with the consistency of the cut and dry former methods: CTR, CPC, and CPA. While the art and science of understanding and implementing conversion attribution can be a differentiator for agencies, without a cut and dry standard, that prospect can be both scary and potentially risky.

3.    The new process has to fit, roughly, into current processes.

There are many ways to optimize campaigns—agencies have their own methods, but even within agencies that methodology could vary by client. In optimizing, how often does one optimize creative, media placement, and channel? Do these optimizations take place concurrently, or separately—even by different specialized teams? Uprooting methodology on top of process is not an easy sell.

4.    And the final, a whopper of a nearly immovable trio: resources, budget, and time.

Let’s be clear—it isn’t just conversion attribution we’re talking about here for straight optimization decisions. What about the other implications? What do you do with the conversion funnel itself, and how do the various touch points contribute to the success of your campaign?
For example, Atlas has a great analytics report on conversion funnel analysis, and we’ve published this whitepaper on the purchase funnel. But how should you action it? If you know that certain sites fall within a certain range in the conversion funnel, would your creative strategy be different? Would your buying behavior change?

Using a conversion attribution model for optimization is just one aspect of looking holistically at the touch points leading to a conversion. But there are very few who have the creative budget, the time and the resources to plan media and creative together along every point in the funnel, across all channels. It’s not that agencies and marketers don’t get this need—it’s often that they are trying to make things work given tight budgets, quick turn-around times, and limited resources.

Adopting conversion attribution can be challenging for agencies and marketers—it’s not for lack of understanding last ad is flawed. In fact, I recently read an article that said adopting the wrong model is worse than last ad (that could probably be argued somehow, but the statement does support slow adoption, as well). In the next post on this topic, I’ll offer some suggestions for ways to take small steps towards adoption—because even while you may not be ready to flip the switch, starting to think differently and becoming strategic about implementation is the first step to getting there.

Thanks,

Lori Goode

Follow us @AtlasAdvertiser | Atlas on Facebook