Ironic is a problematic word. Everybody knows it, and everybody has a strong sense of what it means, but people are perpetually using it to describe situations that are coincidental or just generally strange. See for evidence Alanis Morissette’s 1996 hit, “Ironic,” which includes, by my count, zero ironic events (instead focusing on “rain on your wedding day” and other unfortunate circumstances). Moving past the fact that this misunderstanding is, on its own, an ironic thing, we can consider a similar situation in digital marketing technology.
Attribution (“the process of identifying a set of user actions that contribute in some manner to a desired outcome, and then assigning value to each”, IAB 2013) is a problematic word, one often misused and associated with many broader capability areas. It’s a hot topic now, and has been for a while. In 2015, the Gartner Digital Marketing Hype Cycle had “multi-touch attribution” at the very top of the “peak of inflated expectations,” and since then it has begun the messy slip into the “trough of disillusionment.”
At Ovative Group, we believe much of this disillusionment stems from asking attribution technology to do more than what it’s good at. Marketers should approach attribution as one component of marketing measurement, and set sights on consistent, actionable improvement. In this article, we’ll lay out a few key points of our attribution philosophy and provide guidance for how marketers can make the most of this capability to improve marketing performance.
To start, a few general points:
The goal for attribution should be continuous, incremental improvement in marketing effectiveness. Attribution is only going to do one thing for you—estimate the amount of credit that any given customer experience should get for your selected KPI. It’s not going to deliver for you the optimal media mix; it’s not going to directly optimize bids at a granular level; it’s not going to point you to the ideal customer journey. If these are the goals you hold your attribution capability to, you’re guaranteed to be disappointed. Narrow your focus of what any technology or methodology should deliver, and invest your time and energy in building the ecosystem around this approach to really drive change.
“Attribution” has been conflated with a specific breed of measurement tools, turning it into something to consider investing in, instead of an inherent part of your marketing activities. The concept has been married to a lot of other descriptive words that align to particular types of attribution: multi-touch, multi-channel, algorithmic, rules-based, etc. A desire to build (or, more often, buy) the “best” attribution approach has emerged. This hunt for perfect approach has prevented many organizations from making meaningful measurement progress.
If you report on marketing performance in any fashion, you’re already doing attribution. As such, we’re never starting with a fresh slate in this ecosystem, and it’s important to recognize attribution is already affecting your marketing performance. Improving your methodology can have impact on your business tomorrow. This should be both exciting and nerve-wracking—the good is available for us to take, but the bad is already here.
So what should we do?
Building the better attribution model requires articulating what you’re trying to do, then building the best data set to enable this. There are four steps we’d recommend you take to move down this path:
1) Determine what you’re trying to do. What problem are you trying to solve with “attribution” analytics? Be specific—different tools solve different use cases. For more on this, check out the Ovative white paper on measurement tool alignment. A few common use case/tool combinations include:
- Measure the impact of marketing within the context of merchandising, base, etc.
- Look at Media Mix Modeling
- Identify the optimal allocation of marketing budgets to optimize various KPIs
- Look at Media Mix Modeling + Multi-touch Attribution
- Drive channel optimization at a low level of granularity
- Look at Multi-touch Attribution + Media Platform Data
- Support customer-level testing
- Look at Customer-level data sets and Incrementality Testing
2) Identify & prioritize the data characteristics required by each of these use cases. Do you need your data to be comprehensive or granular? Accurate and stable, or recent and frequent? Individual tools will generally maximize one of these factors at the expense of others
3) Evaluate tools against prioritized capabilities, and choose a focus for each. These tools can, and should, be used in concert—we don’t need one solution to solve all our problems. Allow the tools to shine where they can, and fill gaps with alternate capabilities.
4) Test, test, test. Activate, activate, activate. Most important of all: keep moving forward. You can continuously improve capabilities, but only if you’ve identified what needs to be improved. The best way to develop confidence and know where to prioritize is demonstrate effectiveness in the market.
The hunt for the “best approach” to marketing measurement and attribution has left many in a state of gridlock. To move past this, we need to let go of the hunt for a single optimal analytics provider, and instead focus on building the best stack of tools to support action. By focusing on what we’re trying to achieve, aligning our tools to their strengths, and relentlessly driving a test and action plan, marketers can achieve continuously better measurement and performance results.