Digital attribution or Econometric modelling – What’s the best way to connect cause to effect in digital marketing?

0
Digital attribution, econometric modelling, or a hybrid of the two approaches?

What’s at stake for marketers today?

The return on investment from broadcast channels such as TV and print is long-established and well-proven. The analysis of cause and effect in marketing is best done using econometric modelling. Many advertisers have significantly increased their investment in digital advertising in recent years, and yet few truly know how digital compares in efficiency with broadcast channels. It’s important to get this right, because getting the optimal channel mix can improve media ROI by upwards of 20%.

Digital attribution modelling has become popular in the marketing industry as the digital investment has increased – among advertisers, of course, and promoted by providers of digital attribution services. While digital attribution can identify the relative performance of different digital channels, there are several ways in which it fails to tell the full story. Chief among these are the often arbitrary rules applied to attribution – from first click to the last click – and the fact that digital attribution cannot account for non-digital and competitor marketing activity.

To get a truly rounded view of cause and effect, advertisers need to combine the best of top-down econometric modelling with the bottom-up granularity of digital attribution. This approach – known as Multi-channel attribution – is the best way to get a total view of what’s causing what, of what’s having the most impact, and what’s wasting precious marketing resources.

What is Multi-channel attribution?

Multi-channel attribution is a branch of data analytics and modelling that enables marketers to understand the impact that campaigns, channel choices, and activities have on their customers’ behaviour. Thanks to its ability to identify, isolate, and attribute causes to effects, Multi-channel attribution can show how a particular combination of activities (or stimuli) generates specific outcomes (or responses).

As marketing for almost every organization involves the use of multiple channels, attribution modelling is designed to inform marketers which combination of inputs is driving which responses, and so to rate and rank their relative value. And because it can make sense of the impact of each input alone and in combination with others – for instance, the impact of TV advertising on search – marketers can then use this intelligence to guide marketing investment decisions.

When should you carry out Multi-channel attribution?

Optimizing the marketing mix to deliver the best possible return on investment is an ongoing process of testing and learning. Multi-channel attribution should be used periodically to deliver a total view of the impact of a campaign or channel mix on customer behavior. The growth of digital marketing and channel choices has given advertisers more consumer touchpoints, more data, and more opportunities to fine-tune customer journeys than ever before. But it’s important that Multi-channel attribution modelling factors in the contribution of all marketing activity, not just digital.

Who uses Multi-channel attribution?

  • Marketers looking to understand the dynamics underpinning their customers’ journeys.
  • Marketers looking to optimize their marketing investment by spending more on those activities that have the most profitable impact on customer behavior.
  • Marketers looking to minimize wastage and enhance the return of investment.

Has digital marketing made Multi-channel attribution more complex?

Without doubt. More channels, more opportunities to influence customers, and more site specific and cookie-level data all make attribution modelling more complex – though by no means impossible. One of the biggest challenges for marketers is knowing what type of Multi-channel attribution model to invest in. Digital attribution has a role, but it is only one chapter of the story.

With the growth and explosion in digital, a large number of digital attribution models and services have been brought to the market – by digital agencies, by media agencies, and by specialist digital attribution product and service vendors. The trouble is, the clear majority of these digital-only attribution services do not take account of the impact of non-digital media and other factors on customer journeys. In this way, they overstate the contribution of digital by ignoring the impact that non-digital media and other inputs have on the customer journey.

What is the traditional approach to Multi-channel attribution?

Digital attribution modelling. This is good at understanding the relative contribution of different digital channels to customer behavior, but cannot account for non-digital media and other inputs. This gives a false impression of the relative and absolute value of digital media and their effect on the customer journey.

Consider the case of online price comparison websites for financial services products (insurance, credit cards, utilities) or online travel retail aggregators. In many mature advertising markets, these businesses are among the biggest investors in TV advertising, a channel choice designed to build and sustain brand and to influence search behavior. Any attribution model that ignores the impact of TV for one of these online-only businesses will give an inaccurate account of what is causing what. It will fail to give the total views of causes and effects.

What’s the trouble with traditional approaches?

There are four main problems with many of the established digital attribution models.

  1. They can only account for the impact of digital media channels and inputs. The only ever give a partial view, not a total view.
  2. They’re rule-based and don’t let the data decide what inputs have had what impact. Rules used include first-click, last-click, linear credit, position-based, and time decay.
  3. They are ‘black box’ solutions that require marketers to buy into a proprietary but unexplained and unaccountable methodology. Advertisers are increasingly demanding ‘clear box’ solutions.
  4. They are compromised by the vested self-interest of those selling them, who can often also be involved in selling media inventory.

How can you improve on traditional approaches?

Best practice in Multi-channel attribution is to use a combination of top-down, traditional econometric modelling, and bottom-up, ‘clear box’ digital attribution modelling.

Traditional econometric modelling scales all factors stimulating sales, including offline and online media, price, promotion and distribution. To complement this, ‘clear box’ digital attribution modelling provides correct scales to all digital media impacts in detail, by search keywords and by site and format for digital display.

This approach has been shown to give a truly holistic understanding of what’s working and what needs attention when looking to optimize performance across the customer journey. It’s what we at Ebiquity call Total View Attribution.

How can I find out more about Total View Attribution?

Download our Viewpoint paper: Understanding Total View AttributionClick the image below for your complimentary copy.

 

About Author

Patrik Sahlin is Principal Consultant in Ebiquity’s International Effectiveness practice. Before joining Ebiquity, he headed development of a digital attribution offering and provided digital media consulting services for Publicis Group’s clients at Ninah Consulting. In recent months, Patrik has led the design of Ebiquity’s Total View Attribution service.

Comments are closed.