Bridging the gap between the strategy of attribution and tactics of personalisation

Let’s face it your average marketer these days is much more data literate than ten years ago. They must be, given the rising profile and utility of data to support and guide their decision making from planning and targeting their marketing, to evaluating it effectively in the pursuit of maximising ROI - all whilst under greater scrutiny from senior management.

And attribution is now front and centre of this shift towards measurement and accountability. Most of the progress has been made within the relatively self-contained digital ecosystems where marketing can be ‘measured’ by establishing the origin of web traffic from organic, to paid to display, affiliates and social for example.

The self-contained nature of these systems, and the prevalence of incorporated packages such as Google Analytics has led to a reliance on simplistic measures of attribution such as last and first click that disregard the value of other marketing touchpoints contribution towards a sale.

Even later enhancements that seek to distribute attribution over a wider path to purchase have the same fundamental flaw. That flaw is one of subjectivity - whether a brand measures attribution on first or last click, or weights attribution across all impacted touch-points is down to judgement, educated judgement granted, but judgement at best and speculation at worst. The approach is open to all sorts of mis-representation ranging from a poor understanding of how marketing drives sales, through to expediency, and knowing self-justification.

But attribution is very complex phenomenon anyway – representing it accurately is as much about how marketing touchpoints interact in terms of frequency, and sequencing as it is about how they are weighted.

Best practice attribution takes in to account the impact of all channels working together to deliver a desired outcome - most commonly, a sale. To calculate this, marketers need to use mathematics to model these complex relationships. The approach requires four things:

1. Build up a picture of how individuals behave across all channels and along the path to purchase. This should be stored in an analytical data mart (effectively a hub/data warehouse) at individual level to show a Single Customer View for each customer or prospect.

2. Understand the order, frequency and timing of those marketing touchpoints because the way these factors interact impacts marketing effectiveness

3. Build mathematical models that predict how these touchpoints work in tandem to deliver the outcome of interest (usually sales).

4. Use these models to allocate the sales value across the touchpoints in an unbiased way to create a true picture of marketing attribution.

These highly individual- approaches to attribution, that are built ‘bottom-up’ from data on consumer’s behaviour, are also opening brand new opportunities to personalise communications to individuals. This effectively bridges the gap between the more strategic view of attribution (i.e. how marketing budget is best deployed across channels), and a more tactical communication-based approach aimed at customers.

The detail locked up in these data marts, built to underpin data driven attribution, can be used to also drive next best actions. For example, consider those individuals that you know from the data, are part way down the path to purchase but haven’t yet purchased. Wouldn’t it be useful to know who to target next through what channel to maximise conversion to sale?

Next best action is all about context, not just who should receive what offer but when. These data sets provide that context, identifying where individuals are in their journey and how they compare to similar people on similar journeys who have converted. The models used for attribution can be used to test scenarios comprising different ‘next best’ marketing interventions and identify which are most likely to lead to a successful outcome. These ‘decisions’ can be executed via the appropriate channel, optimising utilisation of marketing budgets to ensure it’s target at the right combination of individual and channel.

This approach explicitly bridges the gap between attribution and personalisation, using the same data asset to build models of next best action for any consumer part way down the path to purchase, identifying what offer to make, and through what specific channel, taking into account everything we know about them at that point in time and, crucially where they are along the path to purchase.

This ‘action optimisation’ approach bridges the gap, extends the value of the investment made in attribution, but opens a whole new opportunity for the marketer to optimise revenue generation.