A new forecast from Forrester Research credits greater online spending by Canadians, lower shipping costs and more selection for the spending increase.
In syncing up product data for online merchandising, the same old, same old is a good thing.
Among all the strides in retailing made over the past decade with web technology, certainly among the most popular is the ability of merchandise managers to control the display of product and promotional content on a web site. Not only can they load and change such content with relative ease, but they can also quickly test customer response and keep tweaking things until customer activity and sales are where they want them.
“There’s nothing like this in the physical retail space, where we can test content against a million people a day and watch the winners among product displays float to the top,” says Patrick Byrne, chairman and CEO of Overstock.com Inc.
But getting to such retailing nirvana carries a big assumption: that the information you’re loading to a web site, presenting to customers and testing for effect on sales is accurate and up to date.
That’s not always the case. And no one knows this better than Byrne. “Two years ago, our data was a mess,” he says. With product data split among several databases and often with multiple versions of product descriptions, it was difficult for Overstock’s customers to search for products and for merchandisers to set up cross-selling displays, he adds.
In 2005, Overstock took painstaking steps to clean up the data on 1 million products and switched to a single data warehouse from Teradata, a division of NCR Corp. That started the company on a course toward having accurate data across its enterprise-and in turn, faster inventory turns, lower marketing costs and more effective merchandising, Byrne says. “Once you have nice, clean data, you can operate your inventory more efficiently,” he says. “We’ve freed up $70-$80 million by being able to carry less inventory.” He adds that having accurate product data helps to keep the right products on hand that people buy.
The challenge that Overstock faced was that of data synchronization-ensuring that retailers and manufacturers, and all the departments within a retailer that have responsibility for product description, are using the same terminology to describe products.
Data synchronization efforts are nothing new, of course. Retailers and suppliers have long struggled with trying to update each other’s product data-a task that can involve tedious hours in meetings and in constant phone, fax and e-mail communications-to ensure that the products a retailer plans to sell are the same products-with the same information on features, dimensions and pricing-that the supplier plans to send or has already delivered to the merchant’s warehouse.
Is that red or sunfire?
Retailers face the same difficulties within their own multiple departments. If new product information received from suppliers isn’t updated accurately and consistently across all departments, merchants run into trouble with promotions or find that online merchandising displays don’t match what’s actually in inventory or that site search engines aren’t properly coordinated with the way data is structured.
“The challenge is to move away from having multiple databases internally,” says Rena Granofsky, president of Toronto-based technology consultants RIT Experts. “That will continue to be a problem until retailers go toward a more centralized database that everyone accesses.”
To produce accurate data across a retail enterprise, experts promote master data management systems that synchronize new data from suppliers, ensuring that all products are tagged with a standard group of identifiers-using the word “red” to describe the color rather than “sunfire,” for example, and using the consistent spelling and abbreviations.
Although proven technology and processes to clean and better organize product data are available, many retailers avoid moving toward a better data management system because they fail to see either the problem at hand or the value in fixing it-“plus it’s boring,” adds Mike Spindler, former CEO of MyWebGrocer who now runs Gladson Interactive, a Lisle, Ill.-based company that cleans up product data for retailers and suppliers. “That’s why not enough retailers do it.”
But technology “is not really a barrier these days,” says Andrew White, analyst with research and advisory firm Gartner Inc. Most data management systems are at least partially web-enabled and, therefore, more capable than they were before of sharing data among multiple applications, he adds. “It depends more on who owns the data,” he says. “But many business people don’t think that data quality is an issue, so they need to become part of the solution.”
One of the recent developments in managing data, from site search technology companies like Endeca Technologies Inc. and Mercado Corp., is based on the use of tools that transform product data into a more useful form for managing and searching for content.
Endeca, partnering with Silver Creek Systems, released in August the DataLens Foundry, a software module for its site search platform that extracts and standardizes product data from one or more databases. Product merchandisers and IT experts design business rules that determine when data in product category files falls outside of accepted parameters.
An obvious example would be if a listing for a men’s knitted shirt appeared in a category of women’s shoes, but rules could also identify more subtle aberrations such as different ways to describe the same type of high heel. By standardizing such terms, merchandisers can more quickly and accurately arrange cross-selling displays, for example, and shoppers will more likely find all the available styles of high heels, says Martin Boyd, vice president of marketing for Silver Creek.
Thanks to web technologies including XML and web services, such tools offer flexibility in how retailers can use them to manage data, says Michael Rice, principal for retail consulting at systems integrator Infosys, which works on Endeca/Silver Creek deployments. A rule might determine, for instance, that a listing for a hammer be automatically extracted from a file of lawnmowers and sent to a tools file. Another rule might send a listing of a hammer priced at $2,000 through an automated alert to merchandise manager’s desktop, calling the manager’s attention to what appears to be an inaccurately priced as well as misplaced item.
The old GIGO rule