Improving Conversions: A Look at “Learning Search”
The dream of all e-commerce managers is to increase conversions rates on their websites, and in an increasingly competitive market the pressure is on them to perform. New Media Knowledge caught up with one New Zealand firm claiming it can improve conversions.
E-commerce has provided many firms with a vital revenue stream during tough economic conditions, although there are signs that even this growth could be slowing. On the plus side, a third of e-commerce managers expect to expand their teams this year, but the pressure is on those teams to perform by creating online stores that will convert visitors into sales and repeat visits. Web analytics firm Coremetrics provides some interesting data on UK conversion rates, which also implies that conversions rates are suffering.
Learning Search is a hosted tool built on technology from New Zealand firm SLI Systems and is being used in the UK by a number of retailers, including florists Interflora. The company says it continually “learns” from past site search activity by tracking visitors' aggregate search queries and click-throughs, and uses that data to deliver results based on popularity.
NMK spoke with the company’s vice president of marketing, Geoff Brash (pictured), to get the low down on how Learning Search works.
How did Learning Search come about?
At the heart of what we do has always been the idea of making search more relevant and usable for website visitors – whether they are looking for products or information, which is where the idea for “Learning Search” came from.
Studies show that people expect to find what they’re looking for on the first page of results, or they’ll abandon the site. Nearly three-quarters of online shoppers will also leave an e-commerce site within one to two minutes of not finding the products they are looking for, so it follows that improving search helps conversion.
How does it work, in a nutshell, and how is this different from other search systems?
Learning Search is a hosted site search tool that delivers the results people are looking for on the first page 95 per cent of the time, which means more satisfied, and potentially more loyal, customers.
Uniquely, it isn’t based on algorithms, unlike many other site search tools. Instead, our technology continually learns from site users’ behaviour to return the most relevant results, using an aggregation of past activity, including search queries and click-throughs.
The fact that our technology is continually aggregating and learning from past activity improves the quality of the results and reduces the time it takes users to find what they’re looking for.
Is there a semantic element to it?
Yes, there is a semantic element to what we do, and are we using the latest technologies, and available data, to help improve the search within sites.
What are the key five things e-commerce managers should demand from their search systems?
Distilling it down into five key things is difficult, because it really depends on the company, but, outside of the basics, like whether the solution’s hosted and whether there’s a free trial available, some areas e-commerce managers should be thinking about are:
· How easy it is to customise the results so that they fit with the look and feel of their website, so the page appears to be part of the site? Additionally, can the search results presentation include pictures, graphics and links (such as a “buy” button beside products) to make it easier for visitors to make a purchase
· How comprehensive are the analysis and reporting tools, and do they provide detailed information to help ecommerce managers improve their site, including; what visitors are looking for, what they are finding and not finding, and key metrics on search result quality?
· Does the system have merchandising capabilities and offer ecommerce managers the controls they need to promote products? This includes the ability to control product rankings and promote products on search results pages, specify landing pages, and control the order of results based on price, availability, margin or other metadata
· Can the system automatically generate navigational pages from visitor data, allowing users to browse and refine their browsing through the site via multiple paths, for example, by manufacturer, category, price range, or size?
· Does the system have faceted search capabilities, which present search results with the same navigational refinements as the rest of the site, providing a consistent user experience across search and navigation?
Finally, what’s the state of digital innovation like in New Zealand? Are many firms managing to gain traction overseas?
As a country, we have always been very successful with exports – originally with primary products, but now we are also successful in exporting technology, which is still a growth area for New Zealand.