LONDON: Econometric models might be a useful strategic tool, but they are not suited for giving marketers actionable advice on media choice while campaigns are running, Vodafone's head of insight has suggested.

Speaking at an event organized by Thinkbox, the commercial TV trade body, David Still said Vodafone was instead employing pharma-style controlled tests in order to inform spend across channels. (Warc subscribers can read more in an exclusive report from the event: Swapping econometrics for test-and-learn: Vodafone's new media strategy.)

"These [econometric] models are great with providing us with the big picture, getting loads of data over a long time," Still argued. "They are the foundation of our marketing plans. But too many people use the models to answer questions that they weren't designed to answer."

Instead, Vodafone has taken a cue from the pharmaceutical industry's testing of new drugs, using 'exposed' and 'control' groups, serving different marketing spends per channel. They ran different sets of tests and experiments in each region, switching them each week.

"We took all of the media channels and ran different permutations of spend against each other," Still added. "Then we put sales against our spend. These are real world sales – nothing was modelled."

While the results of these tests are confidential, Still said the overall results of the project had informed Vodafone's business decisions. "I now know the causal impact of each media channel – for each level of spend, what the channel will deliver in terms of connections," he said.

And, in future, Vodafone is likely to extend the projects overseas, and use the controlled tests for other issues such as pricing elasticity.

"The typical conversation used to be 'how much I think I can deliver'. Now, it is 'here's how much I can deliver'," Still said. "We can now give the business two choices – get more volume for the same spend, or the same volume for less spend."

Data sourced from Warc