GOMO: Building an MTA model that cares about the journey, not just the outcome

Client: Singtel

GOMO is a sub-brand of Singtel, Singapore’s largest telecommunications network. As a digital, no-contract MVNO (mobile virtual network operator) it is great for recruiting a youth audience into the Singtel portfolio, thereby future-proofing Singtel. Given its digital nature, a significant amount of GOMO investment is in always-on performance marketing, seeking to improve CPA and conversions. 

 Singtel’s existing attribution model was based on a last touch model (common practice in the digital world), which gives 100% of a conversion’s credit the last click or impression, which is often search or retargeting. This model disregards the multi-dimensional nature of consumer purchase behaviour.

 Our challenge was to develop a custom multi-touch attribution (MTA) model that would help increase GOMO sign ups whilst lowering the CPA through always on data-driven performance marketing.

 Hence, we based our MTA engine on the Markov Chain model, known for its use in advanced sequence mapping predictions, such as weather forecasting and stock market analytics. To use the Markov Chain model to fuel media decisions, we needed to feed it with hundreds of thousands of consumer journeys to analyse ad sequence, ad interactions, user outcome and investment. From this, each ad type could be allocated a ‘transition probability’, i.e. the probability of moving a user to through the funnel and ultimately converting.

 With the model, the analytics and the dashboard, we were able to pinpoint ideal combinations and high performing ad types, leading better allocation of investment, higher conversions and lower CPAs.