Setting the right expectations for multi-touch attribution

Despite all the hype, for MTA to be useful its limitations must be clear and clearly communicated.

For many, multi-touch attribution is both exciting and intimidating. The ability to track a sale back to the marketing initiatives that produced it has rightfully garnered a lot of fevered interest. If executed properly, MTA can enable prescriptive and granular media optimizations in near real-time that create significant ROI improvements.

The phrase is thrown about emphatically with terms like Big Data, User Level Analytics, and Real-Time Learnings. All that hype can create confusion, and ultimately, an expectation that can’t be met. Precious little airtime is given to the nuances and pitfalls of implementing MTA. That’s not to say it shouldn’t be activated, but that for it to be its most useful, the limitations and nuances of MTA must be clear and clearly communicated.

Understanding the desktop/mobile divide.
Industry reports and surveys suggest the average US consumer uses 2.3 internet-enabled devices. But MTA datasets usually show fewer than 1.6 devices per user. That’s a significant gap, and savvy clients will be expecting 2.3 because they’ve seen the industry reports and think their cross-device ID is weak.

To be sure, cross-service solutions are not perfect. They are a rapidly-improving component of the ad tech landscape. But in reality, the bigger culprit is that your desktop and mobile ad placements may not be as tightly aligned as you think. They target two similar (based on demographics and psychographics) yet different sets of consumers.

A simple way to dissect this is to create two user groups—Multi-Device users and Single Device users—then profile the media activity of these groups against each other. This will reveal sites that over index for Multi-Device users. These sites may represent a small percentage of the media budget that is causing the gap.  

Integrating offline media is not as simple as advertised.
All MTA sales pitch decks have a slide about the integration of offline media, with a chart showing significant web traffic spikes attributed to TV’s impact. This approach is quite often naïve and breaks down in a few ways.

For brands with massive website traffic, TV airtimes do not consistently create obvious web traffic spikes; without the obvious traffic spikes TV’s impact can’t be seen. In addition, TV cannot drive immediate in-store purchases the same way it drives immediate .com visits. This is critically important for brands less concerned with online metrics and more concerned with in-store sales.

And the approach does not easily scale to other offline channels beyond TV. The other offline channels are not designed to elicit immediate response. As a result, alternative approaches for integrating offline media are often required.

Deploying MTA results via a true DSP integration is easier said than done.
Your DSP has a highly tuned and specific algorithm that adjusts bids on an ongoing basis. While MTA providers will trumpet "Partner Integrations" and delivering MTA results directly to DSPs, it is an entirely different challenge for that DSP to then incorporate those MTA results into the bidding process. If not engaged early in the process, DSPs may crudely apply up-weighting or down-weighting bids based on features of the Attribution analysis, which may not really result in optimal bids.

In general, as analytical outputs become increasingly granular, it is becoming more of challenge for downstream partners to implement them. Any ad tech partner that you’d expect to leverage the MTA outputs may require a change to their scope of work in order to undertake it.

Offline purchases may require special considerations.
E-commerce revenue, digital media ad spend and the overall percent of transactions completed via credit card are all going up over time. These shifts are major components of what makes user level MTA possible as they all generate user level data. But what about cash purchases? These are inherently anonymous and, for some verticals like QSRs, may represent 50 percent of sales. Cash purchasers may have far different media consumption habits than credit card purchasers. Tuning media via MTA outputs could be inadvertently tuning your marketing efforts towards credit card purchasers and away from cash purchasers, which could create longer term issues.

Every aspiring brand will be forced to encounter at least a couple of these issues as they tackle MTA. If you’re not prepared for them, your MTA implementation could stall and ultimately lose traction. Your best bet for success is to go in with your eyes wide open and expect that your MTA provider fully understand your business, how you sell your product and to whom and how you plan and execute your media and marketing activities.

Russell Nuzzo is Global Head of Attribution and Marketing Technologies at Gain Theory.

Start Your Free 30-Day Free Trial

Get the very latest news and insight from Campaign with unrestricted access to, plus get exclusive discounts to Campaign events.

Become a subscriber


Don’t miss your daily fix of breaking news, latest work, advice and commentary.

register free