Start with a coveted set of luxury products not normally found online, such as designer samples and overstocks, direct from 2000 partner brands. Offer them in limited quantities, discounted up to 60 percent – but only to an exclusive audience and only for a limited time.
The flash sale recipe pioneered by Gilt Groupe has made the retailer an online phenomena. In just over three years, the roster of registered members has grown from 15,000 to millions, most of them invited by an existing member.
“Our members consider Gilt to be entertainment and sport, not just a regular retail website,” says Tamara Gruzbarg, Senior Director of Analytics and Research.
The data challenge
With millions of website visits a month, 1.4 million members making purchases from eight distinct shopping tabs, and 11,000 customer care contacts a month, the Gilt Groupe has a lot of customer data: demographic data from site registrations; browsing and shopping history gathered from the website, mobile apps and transactions; browsing patterns, and much more.
From the analytics perspective, the amount of information that is theoretically available to us is unprecedented,” said Gruzbarg. “But that doesn’t mean it is readily available. When you have so much data available, you’re bound to have certain questions and challenges.
Analytics in action
Gilt has an analytics solution composed of advanced data analysis, graphics and reporting enabling business users to manipulate, manage, store, analyze, visualize and report on data – all from a single environment. As a result, Gilt better understands its customers and can do a better job engaging them with marketing messages and merchandise most appealing to them.
For organizations looking to drive business decisions with advanced analytics, Gruzbarg shares five key pieces of advice:
- Link the data to the business results you want to achieve. “It is important to know how this information would be used to move the business forward,” says Gruzbarg. “Out of all the myriad metrics and statistics available, what are the key metrics that really explain the business? What are the insights that would tell the story through the numbers and tell you what to do next?”
- Understand the advantages and limitations of the data sources available to you. Some data elements will have more predictive power than others, under different circumstances, Gruzbarg notes. “For example, transactional data has very high predictive power, but by definition, we only have this information for people that actually placed an order. We still want to detect the preferences of members who have not yet placed an order – and we want to determine how they are likely to respond to something new, such as our new gourmet food sales. So, what can be used as a substitute for the missing information? Consider what degree of accuracy – or rather, inaccuracy – is acceptable.” Data in the model doesn’t necessarily have to be perfect, as long as it has predictive power.
- Test and measure. “You will always be asked the question about incremental gains, how a particular initiative moved the needle,” says Gruzbarg. “So it is important to have appropriate test designs for all of your analytics initiatives, for accurate and insightful measurement of results. How did it drive additional demand, incremental dollars or improve something, without any additional efforts on the marketing side? To answer these questions, having an appropriate test design up front, to be able to measure and report the results, is the key.”
- Put analytic insights into the hands of decision-makers. Executives, marketing and merchandising staff have a dashboard, updated weekly, with 55 to 60 different metrics. The dashboard takes about a half day to run and the reports are automatically dropped into Excel spreadsheets. Because of the weekly updates, forecasts and planning are done on a weekly basis – something that wasn’t possible before we had our analytics solution.
- Start early to grab the quick wins. “It is never too early to start with analytics, even if you don’t have full-blown capabilities right away,” said Gruzbarg. “Simple segmentation based on one or two key variables, implemented at the right time, could go a long way in helping to move the business forward. Those initial insights can then be incorporated into future comprehensive strategies.”