Telstra turns to data to reduce customer churn

By on
Telstra turns to data to reduce customer churn

How does your telco decide when it’s time for a chat?

Page 1 of 2  |  Single page

Armed with up to three years of historic data and a suite of analytics tools, Telstra is making sure that it does not waste a cent of its marketing budget in order to attract customers - and keep them - on its books.

The telco has overhauled the way that it uses its enviable data stores to identify the most efficient way of treating an account.

In front of a room of marketing professionals yesterday, general manager of product and marketing analytics Rowan Drummond demonstrated how Telstra’s new ‘uplift’ modelling method can be used, as just one example, to divide customers into a quadrant based on their propensity to churn.

The predictive analytics reveal which customers are likely to remain loyal with Telstra no matter what, which customers never even considers choices in the market (and can thus be more likely to churn if the topic is raised), those who have already decided irreversibly to dump the provider, and those who are shaky – but still persuadable – and thus deserving of a friendly phone call.

“What uplift modelling does is separate the sure-things from the persuadables. In different response models it is quite easy to confuse those two and then you spend resources contacting people you didn’t need to,” he said.

Interrogating every dollar

Uplift modelling is the latest in a series of data analytics experiments Telstra has embarked on.

For the past 12 months, Drummond's marketing analytics team have been putting together a refined business process and tool set that will allow the CMO group to analyse - right down to the most granular level - the return on investment they are receiving from every cent of their spend.

About 18 months ago the team dumped a lengthy customer satisfaction survey in favour of a two-question net promoter questionnaire.

Without explicit questions to drive a deeper understanding of customer motivations, it became the data analytics team’s remit to build a new and more accurate view of the customer.

“Marketing mix modelling needs very granular data over a very long period of time.

“We have a minimum of two years worth of data now, and what we are doing is layering in all of the factors that affect the view of our company,” Drummond said.

The kinds of data that his team are calling upon include call centre waiting times, repeat home visits to connect phone lines, the retail footprint in a particular region, competitor marketing and even third-party product launches.

“If an iPhone launches there will be a spike no matter what we are doing and we do not want to attribute that to our own marketing,” he said.

By reverse engineering these layers of events based on past activation and deactivation data, Drummond’s team has come up with a model of what investment produces what result which they can then map onto the near future.

“Like most marketing teams there is always pressure on us to reduce our costs. With marketing mix modelling we can actually show what the return would be if we added 10 percent to that budget.

“The way we do that is via the modelling layered on top of a business understanding of what is going to happen over the next 12-15 months.

“We know roughly where new products are going to launch from suppliers, we have a reasonable sense of what our competitors might do, and we can make macroeconomic predictions,” he said.

“We want to invest into areas where we think we can change that advocacy,” Drummond said. “We want to create the best possible outcome from that investment.”

What technologies is Drummond's team using and what role does IT play in analytics? Read on for more...

Next Page 1 2 Single page
Got a news tip for our journalists? Share it with us anonymously here.
Copyright © . All rights reserved.

Most Read Articles

Log In

  |  Forgot your password?