Market Mix Vs Multi-touch Attribution Model: Which One and When?

Marketing Mix Modelling (MMM) and Attribution Modelling or Multi-Touch Attribution (MTA) are two well-known approaches used to quantify the impact of marketing and advertising activities on modelled KPIs (such us sales, subscription or other types of conversions).

One can find lot of re-sources online (including our blog section) that discuss strengths, limitations, and key differences between MMM and MTA.

In MMM, modellers regress a set of independent variables (that capture all marketing indicators such as price, promotions, media advertising and external factors) on the modelled KPI. It is based on well-established statistical theory. The technique allows disentangling the contributions of individual factors and quantifying the impact of bigger marketing activities on the modelled metric.

The models, typically built from weekly time series, are suitable for holistic measurements that provide input into high-level recommendations. Their granularity is not sufficient for detailed recommendations or ‘on-the-fly’ optimisation.

MTA, on the other hand, enables measuring the impact of individual digital touch points on a path to conversion. Web-analytics data can be incorporated into MTA model to quantify the contribution of own assets to conversions.

Their key differences determine the main trade-offs between the two methods.

MMM vs MTA: Key Differences

The length of Modelled Period

First of all, MMM is built from 2-3 years of weekly data while MTA is often built from a much shorter period of 1 week – 1 month of individual or cookie-based digital journeys.

Incremental vs Overall Impact of Advertising

Second, MMM quantifies the impact of advertising above the base line, i.e. above the base level of modelled sales or other KPI that has built up over time. In this respect MMM better reflects above-the- line, incremental impact of advertising. The media activities or channels are seen as effective only if it has potential to raise the KPI level above its base line. MTA does not differentiate between base and incremental conversions.

Holistic Evaluation of All contributing Factors vs Detailed Evaluation of Digital

Third, MMM’s key strength lies in its holistic approach and the ability to quantify the contribution of all marketing as well as external factors. MTA often does not consider off-line media and other external factors not captured at an individual level.

An Entire Customer Journey vs Short-Term (Last Interaction) Analysis

Fourth, MMM approach is closest to what would be de facto the last touchpoint or time decay method in MTA. For each incremental sale (or another conversion), MMM gives credit to those media activities that took place in the same modelled period or up to several periods earlier. If the purchase cycle is short and transactional, the marketing effectiveness measurement can be quite accurate.

The data-driven MTA method takes into account the entire customer journey. It has much stronger potential to fully explain what was driving each respective conversion. For categories that have longer and more complex purchase cycles, it is essential to incorporate granular measurement of customer journey. Good understanding of customers and what drives them will be critical for their long-term marketing strategy.

Figure 1 depicts MMM decomposition of the main drivers of the modelled time series of bookings. MTA modelling is generally done on a subset of conversions included in MMM but it models them in a highly disaggregated manner. In MMM, an incremental booking made in a certain period will be potentially attributed to the campaigns that took place in the same period or several periods before the booking. MMM quantifies advertising contribution through aggregating incremental sales that took place on top of the base level MTA measures the contribution of all touchpoints and campaigns a customer was (or might have been) exposed to on their purchase journey. The value of each touchpoint will be quantified by MTA algorithm.

Which Method to Implement?

The choice of the primary measurement method will be driven by

• the brand presence online or offline (such as in stores)
• the proportion of media budget invested in online media (or other directly trackable channels) or offline media channels
• access to data – that can be at a user level or at an aggregated level
• It requires all data available at the same level of granularity
• All data need to be transformed into format suitable for the attribution model

The businesses that distribute and market through both online and offline channels will benefit from implementing both approaches and their integration

With increasing share of media budgets diverted into cost-effective, innovative ways of marketing in digital space and the need for marketing agility tip the scales in favour of models that can be actioned quickly, more frequently and do not require long periods of data collection.

Is There a Scope for a Single Measurement Framework?

Marketing decisions can benefit from the combinations of both MMM and MTA. Some providers of analytical services offer such product.

There are some challenges in this undertaking:

Marketers who collect individual level data can include offline touchpoints into MTA (even if only less accurate measurements of offline mass media such as TV or Outdoor advertising are available). Similar holds for external factors. Bringing all the information into the MTA framework, such model can provide both granular insights and a holistic view of customer journey and its main triggers right before conversion.

While this approach is ideal, data availability poses a significant constrain for many marketers. Figure 2 illustrates well this point. Different sources of data are often collected at different levels of granularity. These can be a cookie level, individual level and other metrics that cannot be attributed to a specific individual.

Source: Merkle’s CRM innovation summit presentation (2014)

Sydney, March 2018