Lenovo’s big-data journey

“We warehouse over 250 terabytes of customer data on our big-data platform, and when you’re talking about managing these huge volumes of data, your traditional SQL or NoSQL databases just won’t cut it,” shared Lenovo’s global business intelligence director, Ashish Braganza during a one-on-one interview at the recent Adobe Marketing Summit. “Lenovo can’t afford not to invest in analytics.”

In the consumer-electronics space, Lenovo faces strong competitors with near-unlimited financial resources. The likes of Samsung and Apple have the ability to spend enough on marketing to saturate any market. “We can’t outspend our competition dollar for dollar and neither do we intend to," Braganza said. "Hence we need to outfox them. We have to be more intelligent than the competition by leveraging the power of analytics to strategically deploy assets in market.”

While Lenovo’s business intelligence practice goes back around eight years, its data maturity needs didn’t hit critical levels until around two years ago, when Braganza was brought in from his previous role at cable-TV provider Comcast to help build the PC-manufacturer’s analytics division.

In most companies, noted Braganza, data analytics is regarded as a cost. “If there’s a dollar to be invested they’d put it towards marketing or sales,” he said. However, as businesses start to mature and market share starts to plateau, they begin to realize that they’ve tapped out their market potential and need to get smarter. Lenovo, fortunately, understands this and is in a position to make investments in analytics, he added.

Ashish Braganza

Braganza’s team is structured around three main areas: predictive and prescriptive analytics, personalisation and multi-channel analytics. It currently runs mainly out of the US thanks to the complexity of datasets available and the high competition it faces in the marketplace, but more recently, Lenovo has launched a social-analytics arm in Singapore.

“My team focuses on marketing analytics to drive real-time measurement and optimisation of our campaigns, whilst our team in Singapore mines data from social and voice-of-customer applications like surveys and forums to provide Lenovo product teams with prioritisation on the development of new product features that customers value," he said. "We share datasets across multiple analytics teams at Lenovo, as we use one big-data platform. But each team has their own area of focus. The Singapore team has been created to focus on those areas that have not been getting much scope.”

The way Lenovo’s analytics team functions and the return on investment it represents for the company are probably better expressed through a case study. Around a year and a half ago, the company was getting a touch frustrated with its US-based retail site, Lenovo.com.

Although it had been determined that in the US, the PC firm’s target should be under-25s, sales coming through the site were largely from the 35-plus group. Yet marketing programmes were driving the target audience to the site. They just weren’t buying.

“That meant there was a dissonance in the type of audience we intended to reach based on our stated brand goals—the upper funnel—versus the audience we saw at the moment of conversion—bottom funnel—which made us revisit the marketing execution strategy to bridge this disconnect between stated goals and actual,” said Braganza.

An issue was the lack of information coming into the site. When visited by an anonymous user, all Lenovo knew was the clicks and hits. So the analytics team engaged third-party data providers so that within the US at least, every customer that landed on the homepage would arrive with around 14,000 data points.

The first step, though, was to listen. Now that data was pouring in, the team needed to start looking for data sets and clumps, modelling group profiles for audiences. This was incorporated with retail data and the team arrived at around eight modelling groups, profiles of their top customers and what they were most likely to be interested in.

Next step was to tailor the marketing, the message and the products proffered on the site to suit the profile of the visiting customer. “If the person visiting the site looks like any of the modelling groups, they see a targeted creative instead of being bombarded with our entire inventory of products,” Braganza said.

This change in strategy resulted in a $15 return for every dollar spent, he said.

“By customising and targeting to get the right message in front of the right audience at the right time, we drove up our conversion rates by 76 per cent, and revenue per visit grew 59 per cent.”

Getting to this stage takes time. But brands overwhelmed by the notion of going over decades worth of data can relax on that point. Most data is obsolete after around three to four years. “In the consumer-electronics space, even 30 days is a long time for us, so a max of two years is relevant data,” he said.