Walmart's figures just don't add up
By Jonathan Ross, head of development in the UK for FACT-Finder
Walmart has received a lot of coverage recently with the unveiling of its new search engine Polaris, which it has developed in-house through WalmartLabs. While this move does demonstrate the retail giant’s commitment to e-commerce, a closer look at what has been done is likely to leave you questioning why the retailer doesn’t have more to show from its investment.
Despite claiming that it developed its own search engine because third-party products had limited capabilities, it soon becomes clear that while the WalmartLabs creation certainly has some interesting functionality, it highlights a number of reasons why not all brands should “go it alone” like “Amazon and eBay” but are instead better served investing in best of breed search engines.
Undeniably it is a great experience for customers entering the site who know what they are looking for, but the true long tail (free-form search requests not directly reflecting the product catalogue) is still largely disregarded and data quality is still a huge issue. Furthermore, obvious misspellings such as “body sope” continue to produce incoherent results, even though “Body Soap” is available in Walmart stores.
Sri Subramaniam, vice president of WalmartLabs is quoted as saying that “the search engine is the crown jewel of any e-commerce platform”. While this may be true, any search engine’s results are only as good as the data that they are drawing on. Indeed, 30% of all conversion loss remains data related in nature. Although Walmart has gone a long way towards improving these numbers, I still don’t see the data quality issue being addressed in any significant way.
Subramaniam goes on to add: “Typically, if you are a business you want to own [the] crown jewel and you want to actually get under the hood.” Again this may well be true, but the big question for Walmart is: at what cost does this ‘tinkering’ come?
It would seem that a vast amount of resources have been invested for an unrealistic breakeven date somewhere in the distant future. How much would Walmart have to increase conversion, in order to break-even on the resources pushed into their think tank WalmartLabs? A relatively modest investment in a solution like FACT-Finder with limited in-house resources would allow for an even greater conversion rate increase compared to the 10-15% Polaris is being so heavily praised for.
So, does the cost really justify the means? Personally, I would say, “no”. Walmart is ploughing huge amounts of money into a project that could have been more efficiently and cost-effectively done by a third-party agency working closely with the retailer. Walmart and WalmartLabs will need at least another five to 10 years before their search algorithms have reached the level of other third-party search engines. And this alone begs the question: was it really a good business decision to try and develop something in house?
Even the “Social Genome” which the Polaris system is apparently based on – a system that allows the engine to determine search weightings based on a shopper's social media interactions – is not much more than data feeds being aggregated to build up a picture of keyword usage and could be relatively easily integrated into a third-party product.
For me Walmart’s comments regarding the limitations of third-party search functions doesn’t wash. The harsh reality is that while Walmart is talking big in terms of its commitment to online retail, it could have saved itself a lot of time, energy and money, as well as achieving even better results, by working closely with a third-party search specialist on a larger research project. Conversion increase would have been higher, more immediate, flexible and sustainable, while at the same time lowering in-house maintenance costs. All things the current solution is unable to meet despite a massive outlay.