RichRelevance, a provider of e-commerce personalization for retail, today introduced RecLab, a new open-source project designed to spur innovation in retail personalization. RecLab enables academics, researchers and developers to dynamically test and validate their recommendation algorithms in a live e-commerce environment. Traditionally, researchers have had to work with isolated data sets in order to protect sensitive consumer data.
New open-source project offers researchers the ability to test and validate personalization algorithms in a live shopping environment for the first time in e-commerce history
San Francisco, CA – Jan. 31, 2011 — RichRelevance®, the leading provider of dynamic e-commerce personalization for the world’s largest retailers, today introduced RecLab, a new open source project designed to spur innovation in retail personalization. RecLab enables academics, researchers and developers to dynamically test and validate their recommendation algorithms in a live e-commerce environment for the first time ever. Traditionally, researchers have had to work with isolated data sets in order to protect sensitive consumer data. In an industry-first approach, RecLab enables researchers to test and debug algorithms against synthetic data sets, then run their best algorithms against live data on the world’s top retail websites. As an existing customer, Overstock.com is the first retailer to participate in RecLab. Through RecLab, RichRelevance is closing the gap between the research community and the e-commerce industry by facilitating and speeding innovation that brings value to its clients and the industry as a whole.
The initiative was profiled in a recent Fast Company article: “There are many holy grails in online commerce, but one that has frustrated C-level executives and engineers alike is how to produce better recommendation algorithms. Produce better recommendations, and you’ll sell more stuff… [RichRelevance] has come up with a way to speed up the process of finding better math to produce suggestions of things you actually might want to buy.” Read the full article here.
“Every 50 milliseconds a shopper interacts with a RichRelevance personalized recommendation across a network of more than 45 of the world’s largest retailing sites, including Walmart.com, Sears.com, and Overstock.com,” said RichRelevance CEO David Selinger. “Given the pace of e-commerce, new ideas and innovations constantly spring forth from different disciplines, which is why the RecLab research community is so vital. Through this innovation, we’re bringing value to our customers years ahead of when it might surface in research or be filtered through a journal.”
“There are tons of incredibly smart researchers in universities around the world who are clamoring for ways to test their hypotheses and algorithms against actual consumers,” said Darren Vengroff, Chief Scientist at RichRelevance and head of RecLab. “These are the same people who spent years working on the Netflix prize. Now we’re giving them the opportunity to go after actual industry challenges, including one of the most basic problems in retail: will someone buy this or not? We’re letting them take their best shot at coding a solution, testing it, ensuring it works, and, through our secure cloud, allowing it to run in a real retail environment. This is a huge spur to innovation, and we’re already seeing tremendous interest in the machine learning community.”
RecLab Specifics
RecLab is an open source project licensed under the Apache 2 license. It defines all of the key Java interfaces and APIs for interacting with the RecLab environment. It also provides simple implementations of these APIs that allow developers to design, test, and debug their algorithms quickly and efficiently without having to take the time, effort, or expense of setting up a large cluster of their own.
The project supports a wide variety of contextual and behavioral data, both at model build time and at runtime. Code running in RecLab has access to both immediate click-by-click and a wide variety of past shopping behavior. In RecLab, researchers begin with synthetic data sets, derived from probabilistic models, not real shoppers. However, once code has been written, tested, and debugged it can be submitted to run live against a small segment of traffic on a live retail site through the RichRelevance cloud. More information on the environment, as well as a tutorial on building models within the environment, can be found here.
About RichRelevance
RichRelevance powers personalized shopping experiences for the world’s largest and most innovative retail brands, including Wal-Mart, Sears, Overstock.com and others. Founded and led by the e-commerce expert who helped pioneer personalization at Amazon.com, RichRelevance helps retailers increase sales and effectively monetize site traffic by providing the most relevant products, content and offers to shoppers as they switch between web, store and mobile. RichRelevance has delivered more than $1 billion in attributable sales for its clients to date, and is accelerating these results with the introduction of a new form of personalized advertising called shopping media which allows brands to engage shoppers where it matters most – at the point of purchase on the largest retail sites in world. RichRelevance is located in San Francisco, with offices in Seattle and London. For more information, please visit www.richrelevance.com.
RecLab by RichRelevance will connect researchers, developers and academics with the goal of creating better recommendation algorithms for online retailers.
Last fall I had the pleasure of visiting Barcelona, one of my favorite cities on earth. It normally wouldn’t take much beyond the great Catalan food and wine to entice me to visit, but in this case there was an even more compelling reason to visit—to participate in the ACM Recommender Systems conference and spend time digging into deep technical conversations with some of the leading scientists and engineers in the field. I also had a chance to demonstrate Instant Shopper, a neat little demo we put together to illustrate the speed and effectiveness of RichRelevance algorithms that combine search and behavioral data on some of our merchant partners’ sites.
A former engineer at Amazon has come up with a way to speed up the process of finding better math to produce suggestions of things you actually might want to buy.
High Fidelity [hi-fi] typically relates to amplifiers and their capacity in managing the sound output to accurately reproduce the characteristics of the input; but fidelity also equates to loyalty, and prompts the question of how a retailer or brand motivates the customer “inputs” that breed the retention, satisfaction and trust that are the desired “outputs.” How effective is “fidelity” in the digital world today? Online retailers discuss loyalty, attend conferences and spend thousands on communications in search of this holy grail, but does “high fidelity” exist in its raw form today?
Loyalty programmes have long had a significant foothold in the UK, with several retailers leading the way when it comes to making strides into true ‘customer loyalty.’ Tesco has been collecting online and offline data on their customers for some time and regularly sends me vouchers and other very relevant offers via ‘snail mail,’ actively inducing me to reduce total order values or gain more of the same thing [bogofs, etc]. But when I venture online, why is this clear understanding of my habits and tastes so far removed? Going back to high fidelity, perhaps the offline inputs are not being transitioned to my online buying persona, which may depend on my browsing habits, and brands, products and category preferences in order to nurture significant output.
Another major retailer in the UK is Boots plc. As one of very few brands entirely focused on Health & Beauty, their loyalty program is possibly the most advanced with over 16million cardholders (including me), who constantly replenish commodities each month. A YouGov marketing report found that an estimated 70% of Boots’ sales revenue is linked to their loyalty card, indicating that Boots has a very clear understanding of—and investment in—who spends what in its stores. Whilst Tesco will send you vouchers and offers to increase the worth of ‘Clubcard’ points so you can take the family to theme parks or escape to luxurious spa weekends, Boots offers the chance to increase ‘Advantage Card’ points by purchasing at an increasing number of affiliated retailers via Treat Street, a recent expansion to Advantage Cardholders.
While impressive in breadth, both programmes essentially drive you back into their stores to increase their loyalty program opportunity, a clever innovation on trust and satisfaction. These are two examples that I personally invest in heavily; Nectar is another huge loyalty scheme (across several retailers) and there are many others beginning to surface. I will for example find a Starbucks now to utilise yet another loyalty card as opposed to walk into any other coffee brand’s store.
Every retailer possesses endless variety in the product mix, demographic customers (and the buying behaviours they inhabit) and business objectives that govern their viability, so ideas and innovations must be considered to ‘personalise’ each retailer’s ‘personalisation strategy’ and therefore create customer loyalty. Having an agile approach to customisation, and never accepting that one algorithm is right for a single shopper on any given site at any time will be critical as retailers bridge their loyalty programmes from the offline world to the online.
For the time being, in the on-line world especially, the meaning of High Fidelity remains as a song I remember back in the 80’s from the film Fame. Hopefully soon, online retailers will transition the analog value of hi-fi to their customers’ web-based retailing experiences—thereby respecting the shopper.