One of the hottest topics in our industry and at a recent panel at SMX West is the reality of multi-device or multi-channel attribution and its impact on conversion optimization. If a user enters your funnel on one device and then picks up the process on another device, what does that do to your CRO efforts? It raises questions around managing the customer experiences, determining when and where cross-platform optimization is appropriate, and what if any impact this evolving reality has on our ability to measure and maximize conversions. Here’s a closer look at multi-device attribution and some preliminary thoughts on how you can work with this in your own efforts to increase conversions in your business.

multi-device attribution and conversion optimization

Image source: Flickr user adactio

What are we really talking about with multi-device attribution, anyways?

Avinash Kaushik wisely points out that there’s a significant amount of confusion when discussing multi-device attribution. He suggests that to some extent, different audiences perceive it uniquely, mostly according to which interpretation matters to their own immediate priorities. Executives and managers at the highest levels may perceive attribution challenges as determining how your online activity drives online profits. (I will follow his recommendation and refer you to this great case study from Google Think Insights on HP’s Online-to-Store conversion tests to help you explore this issue more in depth).

At a more digital level, which is where I will focus today, there are two distinct challenges that we face as marketers  managing the impact of multiple devices on our conversions. The first is the reality that most users today are what Kaushik calls Four Screen People. Four Screen People interact with our brands in multiple digital formats: television, computer (desktop or laptop), tablets, and smartphones. There are three salient challenges with this reality:

  1. How do we control and architect the customer experience for each device AND across devices?
  2. What brand encounters are our customers and prospects having, and which ones are really matter?
  3. With so many options, how do we achieve the right mix (and create the right sequencing) in this potentially fragmented environment?

And then finally, there’s the attribution challenge, as in “where do I allocate credit for driving my actual conversions?”

How do I know what’s driving my conversions in a multi-device world?

A theoretical buying funnel often looks something like this: your prospective customer is searching Google for information about social media management tools. She sees an organic result or perhaps a PPC ad for your brand. She goes to your website, and gives you her email address in exchange for a white paper about how social media tools work. You start nurturing the relationship, by sending regular emails. Eventually, there’s a call to action (maybe a discount) that’s too strong to resist and then you make the sale. This is a buying funnel that’s simple to understand. The process of where she came in, how you moved her along and deepened the relationship, and ultimately made the sale is clear. It’s possible to look at that objectively and optimize any part of that interaction.

Now, consider the funnel through this view for the same product. Your customer is in a meeting and realizes her company needs a social media management tool. She conducts a search on her iPhone and hits a PPC ad for your product. She emails herself the link, which she browses on her tablet over lunch. In the afternoon, she goes back to her laptop and checks out your site more in depth, eventually signing up for your mailing and printing off your white paper to share with her colleagues. She gets your emails over the course of a few weeks, reading them across devices. Eventually, a headline catches her eye one morning when she’s using her partner’s Android phone and she makes the purchase from his phone using her PayPal account. Technically, the same thing happened, except that looking at the situation through the lens of which devices played a role vastly complicates things.

At the highest level, you can still optimize any of the components. Perhaps your insights are actionable – X% of your conversions are mobile, so your sales and checkout process had better be optimized for mobile. But there’s a bigger question in play. You can only spend so much on digital advertising and content deployment. The real benefits of most optimization techniques are incremental, and you need to focus on specific things to capture that. So when you’re looking at the high level overview, you’re really trying to determine which devices and what channels really contributed to the conversion, so that you can attribute credit. Credit translates into future digital spending and how you allocate precious staff time and creative investments.

The trouble with attribution models

There are a number of different attribution models that are used to figure out which digital interaction point “made” the conversion. This list isn’t exhaustive, but rather meant to illustrate some different approaches to thinking about the process. I’ve also listed any issues or advantages to each one. The bottom line, though, is that most of the attribution models that we currently have are insufficient for the complexity of the situations that we’re managing, particularly in a multi-device world. The idea is to get to attribution data and an analytical model that provides sufficient information for an upgrade in the way that we do business.

First click: the first click attribution model looks at which click brought the lead into the funnel and attributes credit there. It’s helpful because it lets you know where you’re being seen, but limits insight on what really drives the final conversion.

Last click: the last click attribution model looks at which click closed the deal. This information is great because you know what converted, but you don’t understand their behavior beforehand or what initially brought them in.

Even attribution: in an even attribution model, you have some insight into the different touch points that are impacting your business. This is better than a single click model, and instead looks at multiple channels throughout the funnel. Rather than determine which ones have more weight, you attribute value to each one equally. In this scenario, you’re getting more insight on which clicks matter but you’re not able to weight and rank them, and in turn derive insights for your spending.

Time decay model attribution: This is a more complex version of the even attribution approach. It takes into account the multiple clicks or touch points during the funnel, and weights that according to recency. Things that happened just before the conversion are attributed more value than those actions that happened in the past. It’s better than an even attribution, but it doesn’t really account for the power of nurturing a lead or quantify the universe of factors that could be influencing the conversion.

Custom or adjustable attribution: Custom or adjustable models are the best. They allow you to use multi-channel behavioral reports to look at the whole lifecycle, and use mathematical modeling and other intelligence that your business gathers to create an algorithm that’s customized to your business and gives you highly personal insights into the right attribution for your funnel.

Closing thoughts

In raising the idea of multi-device and multi-channel attribution, it’s a bit like opening a can of worms. It’s an important issue worthy of examination, even if there are no clear cut answers or silver bullet tools. The thinking is evolving in the direction of multi-channel behavioral reports and using customized algorithms to help you attribute value. The insights derived from this are general enough to help executives make big spending decisions and micro enough to let marketers and analytics professionals tweak variables in the funnel and look at how they convert differently.

Managing this conversation in your own organization probably starts with educating your colleagues on the different types of multi-channel attribute challenges and which ones matter most to your bottom line. It’s then helpful to take a look at where your data and level of understanding is today, and create a roadmap that gets you to a richer analysis. Every incremental improvement will improve your ROI, and get you closer to your conversion ideal.

Is multi-device attribution an issue for you? How are you handling it? Let us know in the comments.