Thursday, February 23, 2012

Leveraging the analytics advantage

Resistance to using analytics in marketing is decreasing, says Michael Wolfe.
Social media is ‘word of mouth on steroids' but cannot replace mass media, notes analytics veteran Dr Michael Wolfe

Between 2007 and 2009, Starbucks fell upon hard times. The economic recession in the US seriously hurt the business. The company ended up closing a number of its stores. Chairman Howard Schultz returned as CEO in January 2008. The rest is oft-cited history.

Until around that phase, the brand hardly had a marketing strategy related to advertising and media spends, recounts Wolfe, Founder Director of Bottom Line Analytics and Senior Director at The Worth, of the BBDO network. The analytics veteran, who has just joined the advisory board of Bangalore-based Rainman Consulting, was speaking to BrandLine on his maiden visit to India last week.

“The primary objective at the point was to reset the company and inject innovation. The belief until then was that if you build a store, people would turn up. That had to change. That's when they hired BBDO,” notes Wolfe.

The agency did a project for Starbucks, and built models on new products being tested. One of the new products – instant coffee Via – was an exploration beyond Starbucks' core competency. A business model with no marketing spends was compared to one that involved significant marketing spends. It did meet with some resistance, notes Wolfe.

They went ahead nevertheless, to show that marketing does work, and figure out which part of the marketing spend delivered. Based on this project, Via was launched nationally in the last quarter of 2009. For the first quarter in 18 months, there was an increase in same store sales and profitability.

“The lifts in Via sales were similar to what we came up with in the modelling exercise. It was a pivotal turning point in the business. With those types of engagements, our job is basically to quantify impact and RoI on marketing expenditure, and build accountability into those investments,” Wolfe explains.


Landing in Mumbai after a two-day seminar on marketing mix models in Kuala Lumpur, Wolfe is confident that the opportunity in this area of analytics is significant in markets such as India. Simply put, accountability is key with marketing and sales being a large expense head – resistance to applying analytics to marketing expenses is constantly decreasing, notes Wolfe.

Perhaps understandably during the recession, the demand for analytics professionals exceeded the supply, notes Wolfe. The analytics practice at BBDO saw increase in demand and revenues through the 2008 slowdown, with companies wanting more from each invested buck.

“It is becoming increasingly apparent that a CMO who does not have a background or appreciation of analytics does not succeed. This is different from five or six years ago, when the CMO was primarily the shepherd of the development of advertising. CEOs and CFOs are demanding and getting more accountability from the marketing organisation,” he explains.

In the course of his career that began in 1979, he has worked with FMCG majors such as Kellogg, Kraft Foods and the Coca-Cola Company. He was involved with setting up Kraft's analytics practice in the '90s, and has just come back from a project for Coca-Cola in Pakistan. The consumer packaged goods (CPG) sector lends itself better to analytics, agrees the veteran, thanks to the ready availability of sales data.

“In the case of B2B marketers where you have shipment data but not actual sales to end users, it's tough to do predictive modelling. But there is opportunity outside of CPG – in retail, hospitality, cosmetics and beauty, BFSI and even casinos,” says Wolfe.


Understanding which part of the marketing budget is delivering returns seemed utopian at some point. While it can be done now, thanks to analytics, what is the margin of error, but before that, where do companies go wrong?

Wolfe explains, “Predictive modelling can be done at best for six months, usually for a quarter. Understanding and quantifying RoI through a one-off modelling exercise is useful, but to leave it at that would be a mistake. There are always surprises in the market place.”

To reduce error margins, Wolfe advocates due diligence through statistical diagnosis. The model needs to have data of the last five years on a brand on board - minus the last quarter. The model is tested for how it forecasts for these three months. If a model is unable to get a 98 per cent level of accuracy, it is not accepted for application. Where such data is not available, other mechanisms to reduce error with lesser stacks of data are adopted.

Customisations by clusters, sometimes at very local levels – as in the case of auto dealerships – are needed to ensure accuracy, he adds.


Media continues to be a key part of the marketing mix. But within media, a lot has changed in the last few years.

Word of mouth (WOM) remains key, notes the analytics veteran, who terms social media as ‘WOM on steroids' – in terms of their capacity to generate awareness. Citing examples of Toyota's recalls in the US and video service Netflix whose increase in prices alienated a segment of customers, he underlines the immediacy of the impact on social media.

“Negative reviews have far greater absolute impact than positive reviews. We have developed a metric of social media that has 70 to 90 per cent correlation to sales,” he adds.
This involves taking words used on social media and developing numbers and thereafter metrics around them.

And for all the press of moving significant part of budgets from mass media to digital and social, there are examples to show that experiments haven't worked, explains Wolfe.
“Some companies have tried to do this in an extremely swift manner. When Pepsico tried this last year in some markets and made a significant move to digital, sales went down. They have since reverted to mass media,” he notes.

For other companies moving in this direction, he foresees ‘similar hard lessons' in store.
“Digital still doesn't have the reach of mass media,” is his measured explanation.

No comments: