Walmart.com’s Improved Search Engine Powered by ‘Social Genome’
On August 30th, Walmart.com rolled out an enhanced "semantic" search engine, named Polaris, for both its e-commerce and m-commerce channels. The feature was developed by a fifteen-engineer team within the @WalmartLabs division, a collection of tech whizzes brought together through the acquisition of a number of tech startups, including Kosmix.
The objective of Polaris is to deliver more meaningful results when Walmart.com shoppers enter keywords in the site’s search field, and it does so by using a constellation of methods that consider the shopper’s relevant interests and thereby intuit his or her intent in doing the search.
The technology is based on @WalmartLabs "Social Genome," defined by the division as, "a giant knowledge base that captures interesting entities and relationships of the social world … people, events, topics, products, locations, and organizations." The Social Genome, the @WalmartLabs site goes on to say, is (cue the sci-fi music) "in a sense, the social world — all the millions and billions of tweets, Facebook messages, blog postings, YouTube videos, and more — a living organism itself, constantly pulsating and evolving."
According to information on the @WalmartLabs website, the Social Genome would allow, for example, the engine to determine based on a shopper’s social media interactions that they would be more apt to purchase a gourmet brand of coffee if just entering the keyword "coffee."
There is of course a lot more to it, but I thought I would try a simple test. Knowing that most online shoppers don’t think ahead (they just peck and go), I tried a vague search term — "glasses." On Walmart.com, the search returned a nice, neat vertical column of various drinking glasses on the first page. But, notably, above that was an offer to "Shop by Category" presenting a choice of "drinkware," "eyeglasses" and "dining & entertainment." Doing the same search on target.com, I got a seemingly haphazard, multi-column display, mixing together everything from sports sunglasses to margarita glass sets to eyeglass repair kits, with no apparent organization. Sears.com took me immediately to their Sears Optical sister site without offering any alternatives.
Somewhat impressed by Polaris, I then asked my colleague to try the exact same search on Walmart.com. His search results were identical to mine; same order of products lined up on the results page.
Highly unscientific, I warrant, but I didn’t see any evidence that Polaris was customizing the search based on my (fairly active) use of Twitter and Facebook. What it did do was organize the results much more efficiently, not leaving to chance that it may have been wrongly guessing my intent.
According to Computerworld, Walmart.com is getting a 10 to 15 percent boost in completed purchases via the new search engine. If accurate, we can assume they’re doing something right, in any case.
Discussion Questions
Will the resources of @WalmartLabs and its Social Genome give Walmart.com a significant advantage over competitors? Do you expect the divide between the big online enterprises — Walmart.com, Amazon, eBay, etc. — and smaller online retailers to widen as social media mining techniques improve?
I would also guess that the consumer’s history of site use on walmart.com plays a role in the ever evolving “data organism,” and that old users of walmart.com might begin to see more personalized results sooner than a new user of the site.
I’m a new user. Would love to hear from regular users out in the shopper universe on this one!
Leveraging this so-called “social genome” should not be a very high priority for Walmart — how about figuring out the value/nature of its customers first, and then using that information to drive search results? Walmart has a long way to go before they will deserve to be mentioned in the same paragraph as Amazon….
Any search refinement that delivers more relevant results to consumers is a big win. We’ve all spend valuable time sorting through items we don’t want on retailer websites. It’s a major frustration. If Walmart can make search results more relevant, they will get more sales, especially when combined with their reputation for low prices. I’m sure other retailers will be closely following Walmart’s progress.
Polaris is yet another example of how Walmart is benchmarking against the algorithm masters of e-commerce rather than traditional brick and mortar competitors. Polaris will give Walmart.com (and Walmart mobile) a huge advantage over traditional competitors and put them a step closer to Amazon. Polaris underscores that that returning encyclopedic search results isn’t the end game for e-commerce — driving conversion is. I would venture to say that Rick’s search experience might change on subsequent visits to the site since site usage behavior is a contributing factor for Polaris, just as it is on Amazon.com. In fact, were it not for my Prime membership and Amazon’s repeat-visit, search-honing capabilities, I wouldn’t venture to the “Amazonian jumble” nearly as often.
@WalmartLabs is a significant advantage for Walmart. They have acquired a number of interesting companies, and been able to recruit some very talented members of the e-commerce community; there work has become a real competitive advantage.
For example, they hired a number of top web performance engineers, and transitioned Walmart.com from middle-of-the-road pageload times, to being one of the fastest loading e-commerce sites on the web.
In the case of on-site search, they moved from Endeca (a well regarded commercial product now owned by Oracle) to SOLR (a popular open source product). While there is nothing wrong with Endeca, it didn’t allow Walmart to have a competitive advantage over any other Endeca customers. By building their own Search in SOLR they are able to add all sorts of competitive features (such as adding social signals and sales velocity to their relevancy calculations), that others retailers do not have.
To complete at the very top end of the Internet 500, you really need to be more agile than just offering the out of the box customers experience available to every other site on the web. Walmart labs does exactly that for Walmart.
Definitely a step in the right direction. Between all those that believe Amazon is untouchable (see “Point/Counterpoint: Can Anyone Beat Amazon At Its Own Game?”) and all those that complain about Amazon’s unfair advantage, this is the beginning of the competition getting a little smarter and more serious.
There are lots of things Walmart needs to do to improve its online selling power. Empowering search, which done well, saves shoppers much time and reduces site abandonment, is a good first move.
Smaller retailers will be at a disadvantage until powerful site tools fit their budget or until they get creative enough to use other means to outmaneuver the big guys.
Except that Amazon owns product search — go ask Google. Is Polaris enough to go to WM to begin with?
I was recently at lunch with a group of about 20 colleagues, most of whom were in their 20s and 30s. There was a good spirit of light social chatter going on all around when I informed the small group I was in conversation with that I read in the morning news paper that “Dead Mouse” was coming to the area soon to put on a couple of shows. The response was “who’s that?” So I said “you don’t know? Look it up!” Soon the whole table was silent except for the thumb thumping to find out who, what and where with no results. None, that is, until the correct spelling “DEADMOU5” was given. Faster hardware and powerful software must be properly tested, configured and maintained to get the results needed to compete successfully. Anything less will produce dead ends that the 21st century consumer hates and avoids.
No, I don’t think it will give Walmart a competitive advantage over its competitors. I still think Amazon is the one to beat online. Luckily for small retailers, they can sell through Amazon and other marketplaces to maintain a certain amount of visibility. I also think there is a “support the small guy” ethos that will support smaller retailers’ efforts as long as they are close on price.
Walmart’s image (from poor PR to the acknowledged “paycheck-to-paycheck” life of its customer base) is one of its biggest gating factors. Amazon does not suffer the same problem.
Making online product searches easier is great, but WalMart must maintain a pricing advantage to really succeed. I have been very satisfied with Walmart.com and also with Amazon. Anytime I have to make a purchase of any significance, I check several sources. If I wasn’t interested in saving money I would just go to the nearest store and pay what ever they asked! Google has all but put every other search engine out of business by making searches more relevant and quicker. If Walmart can make product searches more relevant and quicker, they will have a winner.
Smarter search results which Wal-Mart.com is providing will certainly have an impact on their business, but will it be a significant advantage over their competitors? Probably not. Only time will tell. It will not take long for their competitors to come up with their version of this “smart” search engine mining, and the true impact may never be realized. Again, only time will tell.
At this stage, no. The next version of their search engine — slated to be live in a few months — will supposedly do the kind of context searching Walmart said was in this initial version.
We tested the system for searching for apple. It reported back iPhones and iPads. We then searched for pears, peaches and grapes. Then we searched for apple again. Though our intent was clear, it still displayed iPhones. In the next version, Walmart said it should understand the search in the context of what else we have asked for. THAT will be crucial and it could, theoretically, make a competitive difference.
Bravo, Walmart! This is my second accolade in a week. Last week it was the “scan and go” feature being tested. I note that five years ago I was presenting on the “Amazonification of Walmart” around the world. Don’t know that that had anything whatsoever to do with this, but this is EXACTLY what I foresaw as the proper course for Walmart.
This also moves forward the proper “total knowledge” strategy I mentioned in the privacy/tracking post. The point is that TOTAL KNOWLEDGE will progressively grow the cloud of big data, and the winning retailer will be the one who uses it, not to hector people/shoppers (aka SPAM) but to SERVE THEM EFFICIENTLY, as Walmart is doing here with its Polaris (serving the shopper) from the Social Genome (total knowledge.)
I know that this is not the end of the game, but it is fascinating to see Walmart leapfrogging Amazon, more or less in one fell swoop.
I have long said that retailing is ALWAYS at the cutting edge of social evolution. Some commercial drudges should wake up to the fact that this is the most exciting game on earth!!!
So how exactly is this different from (for want of a better word) a “normal” search function? That the algorithm figured out “glasses” might mean not just drinking glasses but also eyeglasses, window glass, ice cream or maybe even a bio of Ron Glass doesn’t strike me as all that innovative. OTOH you can argue that Walmart always has a significant advantage over its competitors from the start…regardless.
The 2nd question is the more interesting of the two. Walmart’s advantage over its chief competitors — Amazon, eBay, etc. — will relatively short lived because they have the resources to play catch up. The big guys’ lead over their smaller competitors will grow because the smaller players do not have their resources.
We run the site search for over 500 online retail sites, big and small. From my experience there is very limited opportunity to personalize the search results. Once someone has given you a keyword they are clearly signalling their current intention. Past behavior and other information you know about them is normally not that relevant.
Many retailers that are smaller than Walmart can and have been getting great site search by using services of companies like ours. They get the benefits of a larger team of search experts, with a broader experience across many different types of retailers.