It’s the age-old question of marketing, to paraphrase retailer John Wanamaker: “Which half of my marketing is working?” Multi-channel attribution is the Holy Grail for marketers wanting to know how different channels are contributing to campaign results.
Simply put, marketing attribution is the practice of using business rules to allocate proportionate credit to each channel’s contribution to particular action. Most marketers either struggle completely or still use a simple, “last-click” attribution model. A stronger approach is “fractional attribution,” where we give some attribution to multiple items, each weighted by its contribution to the mix, and recognize that buyers are created as a result of a number of events in the consumer decision-making journey. Multi-touch, multi-channel, or full-funnel are all forms of fractional attribution that attempt to assess value of each impression and separate correlation of exposure from causality of an event.
The insights from attribution analysis are powerful. Reinvestment in the channels that work – by audience segment or product line – will return higher revenue and improve customer satisfaction. With so much data available, and so many ways for customers and prospects to interact with our brands, an effective model will be dynamic, and keep up with the changing landscape of our testing and marketing campaign optimization. We’ll all be marketing superheroes!
That sounds so great in concept and value, but yet, so few of us actually do it. Why is it so hard?
Usually, our reasons (I won’t say “excuses”!) include some combination of:
- Data. It can be difficult to get “the same” data for all channels
- Return on Investment. A sophisticated analytics solution is not without cost, and so we need to compare the lift in revenue from using two different models. That test is itself an investment. However, there is also an opportunity cost to doing nothing around media mix and customer journey optimization.
- Turf. Attribution is a zero sum game. When one channel is getting a lot of attribution under a last-click model, the owner of those channels may resist change. You will need executive buy in and support to change the shared key performance indicators (KPIs) to make it about customers, not department goals.
- Knowledge. The skillset for developing, testing, and interpreting the results is rich and varied. You may need to supplement your team with outside experts.
- Influence of Air Cover. Direct channels are naturally able to meet the demands of a test and control situation or ongoing model. Mass media, social, PR, and brand efforts (or what many of us call “air cover” contributing to the effectiveness of direct channels) must be indirectly (surveys or customer feedback) or incompletely measured (dedicated 800 numbers). You must take into account the lift from multi-channel marketing, or you will discount the non-direct channels and overstate the direct channels.
Let’s think positively! Let’s say that you overcome and manage all those barriers. What should you actually do to move to a more sophisticated, fractional attribution model? Recently, John Young, the dynamic and scary-smart director of analytics at Epsilon, led a town hall conversation for DMA members (full disclosure: I work for DMA now) and offered these insights.
- Update your expectations. Most of us – if we use attribution at all – are still doing last-click attribution because it’s the easiest to implement and the least likely to generate turf wars internally. However, a better attribution solution gives us not only the relative contribution but also the value of the marketing treatments within the channel, like the offer or the creative. Similarly, attribution models can help identify behavior by segments within channels.
- Bring everyone on board. Get the perspectives of the different stakeholders, especially on the marketing side, he advises. If you want their support, you need their input.
- Tie it to business growth. “How an organization does their marketing and selling really dictates how they need to attribute, taking into consider on- and offline purchasing, customer journey, and fulfillment practices,” Young says.
- Get started and establish a baseline. Consider a “business rules” approach. Let’s say your customer is given exposure to an email, a direct mail piece, and a display ad prior to purchase. In this model, you simply do an equal placement approach where each of those three channels would get a third of the credit. This assumes each channel has an equal contribution and ignores recency and cadence of touches. Many people who move from last-touch models move to this kind of approach and compare to the results of the prior model and then build on a more data-driven model.
- Amplify the approach for higher value. Channel weighting approaches use an assumed level of contribution for each channel and spend. If you can’t measure a channel response, you have to have some way to assign credit. So for example, on mass media or social channels, you can use a dedicated 800 number or landing page, or media mix position to stand in for direct response. In all cases, you also want to factor in any branding or cumulative effects of multi-channel activity – you must factor in that some direct response is a result of non-direct media exposure.
How are you doing attribution today? If you are not, what is stopping you? Let us know in the comments section below.
New Top-Level Domains (TLDs) have become more popular in the last couple of years, so here’s everything you need to know about them.
Sure, some apps are doing personalized push notifications, but what happens when your users are in the app?
Since cloud computing first gained mainstream attention around 2009, its popularity has exploded. Promising increased efficiency, flexibility and cost-effectiveness, it was hailed as the ultimate business solution. But are users seeing the benefits?
The term ‘marketing cloud’ has gained significant traction in the last few years as major software companies have sought to monetise the growing importance of technology for marketing teams.