What is attribution modelling worth to your business, asks Simon McEvoy, planning director, Tangent Snowball.
Measuring the current zeitgeist for marketing buzzwords, it seems you can’t attend a meeting at the moment without someone bringing up accurate ‘attribution modelling’ (disclaimer - I am well guilty of this myself and am a regular winner of buzzword bingo).
Of course, like most buzzwords, attribution modelling is actually a very fancy term for something pretty simple - measuring the ROI of different points of the path to purchase and trying to assign a weighting to them to better understand their value. It’s a discipline borne out of resistance to the ‘last click fallacy’ (only crediting the last click before purchase with 100% of the ROI) and is a vital way of understanding consumers better, assigning budgets more evenly and creating more focused marketing campaigns.
But how do you do it, and what is it worth to your business?
The reality is you’re probably doing some of it already. A recent Adobe report showed that 26% of marketers are carrying out some level of attribution modelling, with a further 28% doing it but only assigning a ‘last click’ attribution approach.
Attribution activity might include taking into account how many emails a customer has opened and read, even though they eventually bought from a PPC Google link. Or measuring how engaged a customer is with your Facebook page, even though they bought later from a mobile app. What we are trying to do is discover how a consumer interacts with the brand across multiple channels, devices and media and how each one influences the final decision to purchase.
Much of the desire to measure attribution more effectively has come from our new-found ability to do it more accurately, thanks to online tracking and data management.
Tools such as Adobe Marketing Cloud, Google Multi-channel Analytics and IBM Marketing Optimisation Suite are able to give us a more rounded picture of consumer behaviour - ads clicked, social media engagement, emails opened, web visits, loyalty card purchases - and help us string this together to form a more complete picture of a consumer’s path to purchase.
Of course, every consumer journey is slightly different, so using one set of tools is unlikely to give us the full picture. Successful attribution modelling will probably require a variety of off-the-shelf and bespoke tools, as well as aggregate disparate data sources such as ePos, email and DM responses.
However, even a simple attribution model can throw up fascinating results. One luxury brand we worked with was unsure about the value of its email campaigns, as they didn’t tend to result in many online sales. When we combined its email data with its in-store purchase activity (tracked by asking for email addresses at the point of purchase) we realised that it was far more common for email recipients to go into a store and purchase after receiving an email, due to the products’ high price points.
The reality was, email was adding millions of incremental revenue to the brand and was arguably its most profitable marketing channel.
So attribution modelling is great and we should all head off and get started with it straight away, right? Unfortunately it’s not that simple.
Many businesses are simply not prepared to deal with the impact or investment in attribution modelling. You need to ask some serious questions, particularly about how you will use the insight. If you discover that a large portion of your budget is funding activity that is not really making a measurable impact on sales, will your business have the courage to cut it? What if that was an advertising ‘staple’ such as TV spots?
Further challenges exist around accurate measurement and assigning weightings to activity, too. It can be hard to assess how much offline activity is affecting online sales and vice versa, as sometimes the data just isn’t available or is too patchy. These types of obstacles can make businesses reluctant to invest in measuring attribution.
However, despite this I believe it is essential that we try to create better ways to attribute value to different points on the customer journey, even if it’s not perfect at first. Consumers no longer take a linear journey to purchase; it’s a messy web, not a neat funnel.
In order to meet their need for truly integrated marketing and commerce we need to understand how they are behaving and what’s making a difference to their decisions, something that can only be achieved from modelling their journey and understanding what each action is worth.
What’s more, it can make a big difference to the profitability of future campaigns, with the same Adobe paper reporting that 89% of respondents had reported a minor or major benefit to their organisation from engaging in attribution modelling. In short, with a nod to Jon Wanamaker, it can take us one step closer to discovering ‘which half’ of the marketing budget is wasted.