Bridging the Gap Between Econometrics and Real-Time Marketing
Creating a custom global data-led solution to predict the next acquisition.
In 2019 WU tasked us with driving a 20% increase in global digital customer acquisition on flat media investment.
Like many categories, money transfer is being transformed by technology. The digitisation of money has lowered barriers-to-entry and created an increasingly crowded and competitive space. After decades of being the market leader Western Union (WU) faces a growing challenge from well-funded fintech innovators, such as Xoom, TransferWise, Remitly and World Remit, as well as revitalised traditional competitors, like MoneyGram International and Ria, all competing to carve out a chunk of the rapidly growing market for digital money transfer. Simultaneously, brand tracking showed that awareness of WU’s digital products was stagnant, and worse, that WU was perceived as dusty, traditional and not a brand ‘for me’ amongst important digital-first and high value customer segments.
In 2019 WU tasked us with driving a 20% increase in global digital customer acquisition on flat media investment. Coupled with aggressive competition, an underperforming digital brand and over-reliance on performance media, it was looking like a tough year.
What We Did
The success of Demand Conversion activity is measured on new customer acquisition volumes and efficiency. However, analysis showed that 30- 40% of WU’s 2018 Demand Conversion investment was generating customers at costs exceeding the profit generating ceiling – these customers would never return a profit because we spent too much on acquiring them. We realised that we needed to be ruthlessly focused on identifying and driving out inefficient acquisition, thus freeing up investment for media that could stimulate the top of the funnel. That, however, is easier said than done when you consider WU Digital operate in over 90 markets across 6 core lower funnel Demand Conversion channels - or 540 ‘cells’ for investment.
We needed a way to assess those 540 individual country x channel scenarios, each with uniquely shaped response curves, varying capacity to absorb investment and differing potential ROI.
Our solution was Next Best Decisioning (NBD), a data-led solution that automates the process of systematically identifying the next best place to spend each media dollar, across every market and every channel, in real-time, within profitable cost-per-acquisition limits.
At its heart are machine-learning algorithms that leverage historic performance data, as well as dozens of real-time external factors (including currency FX rates, search demand, competitor activity and seasonality) in order to make investment optimisation decisions.
NBD can adjudicate and predict the response curves of all country-channel combinations simultaneously, for any time period. This means it can allocate budget to the most cost-efficient combinations in the optimal sequence to maximize acquisition, until all budget has been used, or until a predetermined maximum CPA threshold has been reached. It’s also super-accurate - with a Mean Average Percentage Error (MAPE - a statistical measure of predictive accuracy) of only 3.5% we know we can trust NBD’s predictions.