Multi Touch Attribution vs. Marketing Mix Modelling

In a rapidly changing environment where the impending death of third-party cookies, and other changes like iOS14.5, are impacting the way that marketing activity is going to be planned and measured, the focus is on analytics. And, in particular, attribution solutions that are fit for purpose what is a changed marketing landscape.

In this blog, we consider the relative merits of Multi Touch Attribution vs. Marketing Mix Modelling, including:

• A quick overview of attribution modelling
• What is Multi-Touch Attribution?
• What is Marketing Mix Modelling?
• Which approach is likely to be right for you?

A quick overview of attribution modelling

Attribution modelling is the practice of identifying and evaluating the marketing touchpoints that contribute to a users’ buying journey – and working out the extent to which each of these has contributed to conversion.

Here is a quick summary of the types of basic attribution models which are available:

• First-Click – attributes 100% of the credit to the first customer touchpoint
• Last-Click – gives 100% of the credit to the last customer touchpoint
• Linear – applies equal value to each touchpoint
• U-shaped – favours touchpoints early and late in the customer journey
• Time Decay – assumes that touchpoints closer to conversion are more influential
• W-shaped – an extension of the U-shaped model but with additional weighting for the opportunity stage

Find out more on these models here. But now on to the detail of the Marketing Mix Modelling vs attribution question…

What is Multi-Touch Attribution?

Multi-touch Attribution is concerned with unravelling the complexity of customer journeys which are increasingly being undertaken across a wide range of digital devices such as laptops, mobile and tablets. And touching a wide array of media from display advertising right down to retargeting activity at the lower end of the funnel.

The aim is to identify the incrementality and conversion impact of each, and every, touchpoint on the customer journey. In order to properly attribute the influence that channels, campaigns and even individual creative is having on conversion and revenues.

Although buying journeys are increasingly being undertaken on digital devices a sound approach to Multi-Touch Attribution will also integrate and attribute data for offline media. To ensure that the impact of top of funnel activity like brand building is given credit for the impact it is having.

By its very nature, this type of modelling is able to deliver, as close as possible, real-time data insight.

Pros of Multi-Touch Attribution

• Overcomes limitations in single-click approaches – avoids the mistake of heavily weighting attribution to lower funnel touchpoints while ignoring key interactions higher up the funnel.

• Goes right to the heart of the need for identifying incrementality – by looking at each and every interaction that the customer has on their buying journey and applying credit for conversion.

• Enables you to get granular in your analysis – because it is less cumbersome to set up it enables you to, relatively quickly, dig deeply into what is really happening in digital channels, campaigns and creatives and adjust them to impact ROI.

Cons of Multi-Touch Attribution

• Is focused on measuring clicks alone – and ignores the impact of offline media. This is a criticism which is often levelled at Multi-Touch Attribution approaches and it is worth listing here as it could be the case depending on the attribution solution you choose.
  
  However, as we alluded to above, a good solution is going to allow you to ingest data from offline sources and ‘Walled Gardens’ and stitch that into the user journey. In fact, that is the approach that our own attribution solution, Corvidae, takes.

• Applies arbitrary weightings to conversion impact – some of the criticism levelled at Multi-Touch Attribution centres around the use of arbitrary weightings which are decided upon and then applied to specific touchpoints.
  
  The inference being how do you decide on the weightings and the potential for bias. In manual implementations this is something to be aware of but it is possible to work around the issue. For example, in Corvidae the issue is removed through the use of AI and Machine learning to assess incrementality and conversion impact.

What is Marketing Mix Modelling?

Marketing Mix Modelling, or Econometrics, is an approach that goes as far back as the 1950s but rose to popularity in the 1980s.

It involves using statistical techniques to attempt to identify the incrementality in large data sets using mathematical techniques.

Steeped firmly in the traditional marketing model which aims to consider the optimum mix of price, product, promotion and place it found its roots in the retail sector.

It’s essentially a top-down approach to assessing the impact of more traditional media – such as TV, print, radio and sales promotion efforts – on sales and ROI.

Inputs to the statistical calculation come from a wide range of sources that includes:

• sales and marketing data
• costs
• revenue
• competitive data

And also external inputs such as economic and other relevant customer behavioural data.

It is about analysing this historical data and projecting forward to assess whether current levels of marketing spend – and the shape of the marketing mix – is likely to deliver the level of sales and revenue set out in the business plan.

So, what are the respective benefits and drawbacks of Media Mix Modelling?

Pros of Marketing Mix Modelling

• Works well in more traditional marketing channel analysis – and is a good potential fit for marketing approaches that combine media like TV, sales promotion, outdoor display and radio.

• Can highlight broad trends – because of the global nature of the analysis provided this approach can surface broad trends that exist right across your media and channels mix.

Cons of Marketing Mix Modelling

• It is costly and resource intensive – this is a fairly complex approach that is founded on establishing statistical validity. Which calls for high volumes of data that has to be managed and refreshed on a more or less ongoing basis.

• Lack of real-time data – unlike Multi-Touch Attribution which is able to deliver data in real-time, Marketing Mix Modelling is typically tied to longer term reporting schedules on a quarterly or annual basis, for example.

• Conversion weighted too heavily – by its very nature Marketing Mix Modelling can result in an over-emphasis on conversions which means that your top of funnel activity, such as brand and awareness building efforts, can be undervalued.

Which approach is right for me?

There is no ‘one-size fits all’ response to the Multi Touch Attribution vs. Marketing Mix Modelling debate.

To know which is right for you, it largely depends on what you are looking to achieve. And it is entirely possible that you might run both types in tandem to satisfy different analytics use cases.

Media Mix Modelling may well be what you need if you are looking strategically at overall performance right across your marketing activities.

But if you are looking for the type of day-to-day analysis that enables you to identify exactly what is and isn’t working across channel, media and individual creative then Multi-Touch Attribution is the clear winner.

In an environment where changes like the impending death of third-party cookies, the fall-out from iOS14.5 and the potential illegality of Google Analytics are being felt there has never been a better time to re-assess how you measure marketing effectiveness.

If you are looking at this then don’t forget to dig into the types of cookieless solutions that can futureproof your attribution analytics and provide granular, visit level attribution data that enables you to tie the impact of your marketing activity directly to revenue and ROI.

A Buyer’s Guide to Selecting the Right Attribution Solution

If you’re looking to find out more about how to find an attribution model and solution to meet your needs, download our buyer’s guide to find out:

• What is attribution and why is it so important?
• 4 clear signs you might need a better attribution solution
• Helpful tips for choosing the right solution for your needs