First Steps in Building Effective Marketing Attribution Models

Having an effective marketing attribution model is one of the holy grails of modern marketing. With the need to (1) have access to, and (2) aggregate multi-channel multi-screen data it is a daunting task, and that’s not the place to start.

Marketing attribution can be daunting – start simple but not too simple

The most common solution to the daunting complexity is to use simple models such as “first touch” and “last touch” and only in relation to online. That’s simple but it breaks Einstein’s advice to devolve problems to be “as simple as possible but not too simple”. It’s too simple.

But let’s step back and clarify “marketing attribution”. It is the ability to allocate the costs and credit for business outcomes to touchpoints on the customer journey. Those touchpoints are often aggregated into channels and campaigns. A terminology that I like is that each touchpoint generates a “marketing signal” – the signals are the result of marketing activity.

A “model” is the method you choose to do the allocaton. Hence the  “first touch” model assigns all the costs and value to the first point at which you can identify the customer, and “last touch” assigns all of the costs and credit for a conversion to the source of traffic for the session on which the conversion happened.

Both of those models are badly flawed in being very crude, and give poor insights into business value, revenue and costs outcomes across the customer journey. That’s tremendously important because we want to optimise the customer journey and we cannot do that without assigning metrics to the touchpoints.

What’s needed are multi-touchpoint models and ultimately algorithmic models which use data analytics to dynamically assign values to touchpoints.

How can you get started?

You can’t get started in marketing attribution by trying to boil the ocean. Leave non-digital and unintegrated digital aside, and focus collecting marketing signals from just the web and social. That’s enough, and it is doable with current enterprise digital customer experience platforms e.g. Sprinklr, and with the right choice of platform you can build up attribution performance as you develop your digital tracking maturity.

Clearly, you need a platform that will bring together inbound and outbound social, paid advertising, and web tracking data, and be able to have that tagged and categorised systematically and mapped to user profiles. You’ll need to also add social listening insights and social customer care data into the mix, along with CRM integration.

If that sounds like a tall order, it is! Make no mistake it is multi-disciplinary project, requires careful enablement planning and rigourous implementation, and a willingness to experiment along the way. That’s why it needs to be a collaborative project between us and your experts in marketing, sales, social media, social customer care, ecommerce, digital and your agency.

There is no one “right answer” for the best marketing attribution model. However, it is all perfectly doable and is being done by leading brands.

Technology basics

While the technology choice is not the driver of success, the wrong choice or choices will almost certainly lead to failure. This list of basic requirements needs to be ticked off. The capability to:

• Triage high volumes of inbound conversations and route them to appropriate handlers; Consistently evaluate keywords and search terms to ensure all product, services, brand, and other key issues are captured from social channels and across the Web;

• Create and tag profiles with information based on each customer’s history of social interaction, stated interests, owned products, and other social account profile properties;

• Merge social accounts into a single customer profile to allow for a unified view of conversations; Group, cluster, and uniquely label social profiles together into specific segments (influencers, competitors, analysts, press, etc.);

• Automatically add users to audience segments based on any available distinguishing criteria (e.g., geography, gender, language, keywords, hashtags);

• Deploy paid advertising to boost high-performing content, amplify campaign reach, and extend marketing impact;

• Track performance and generate reports that help teams synchronise paid, owned, earned initiatives, EDM, listening insights and competitive benchmarking  across the organisation as campaigns unfold in real-time.

With those capabilities you have the basic technology enablers to be able aggregate marketing signals into insights which will enable you to iteratively improve a multi-touch attribution model.