How well are you doing at giving your online customers what they want? If your web site is typical, the answer is not too well. And that’s not good for your customers.
Leading retailer web sites get mixed grades when rated on the ability to offer pinpoint search results for online consumers, even though research shows that more experienced shoppers favor sites that provide more relevant search results.
As a marketing executive and CEO with 20 years in the industry, I can tell you that, until your retail web site returns relevant products in a natural-language search, you’re not going to see significant sales.
According to research from my company, which I’ll admit has an active interest in the findings, even the touted Google search engine does far from a perfect job of returning search results within the right context in response to a question phrased in natural language.
Here’s an example of what I mean. Try typing the phrase, “Looking for a white painted bureau with glass handles” in Google. You will get about 60,000 possibilities. The list will include everything from white bureaus at an inn in Vermont to doorknobs made out of glass handles to white glass and white paint.
That means that 99.9999 percent of the content that is being delivered is not relevant to helping a person get information to make a purchase decision. That’s a key disconnect to sellers, because making a decision translates into buying behavior.
A commercial site that is not answering a query in a way that helps customers make decisions by finding what they are looking for will not make money. While this might be OK for “informational search” sites, to those engaged in commerce, it can mean lost sales. And you know what lost sales mean.
From search to sales conversion
Kitchen furniture giant retailer MFI conducted research on its customers and their searches, uncovering an 11 percent sales conversion rate of online customers who were able to successfully search and find the cabinets they wanted—and who then made an appointment to complete the sale at a retail store location.
On commercial sites, content and the ability to search within the context of a query are very important. Also important is how that content is organized on a site. And finally, how that content can be searched and used within the context that of a person’s query is key.
Google today has an index of some 10 billion pages. Finding a precise answer to a question posed to that large a stack of information is akin to the “needle in a haystack” notion. If you do find exactly what you want, it’s just plain dumb luck.
Google is often the tool of students and, perhaps, businesspeople with extra time on their hands. Ignoring the ads that appear on the right side of the Google search page and may not be relevant to the person’s exact search, a customer—someone ready to purchase if the right product appears—has to sift through the thousands of pages that are served up for a query.
Taking a closer look at retail search
Recent research sponsored by ActivePoint illustrates the need for more intuitive search and discovery capabilities with familiar retailer web sites.
Our research randomly selected five online retailers recognized by the e-tailing group, inc., as “merchants that excelled at online customer service” in Q4 2004. We tested the five sites—Ann Taylor, Coldwater Creek, Land’s End, Neiman Marcus and Nordstrom—on the basis of how well their search windows were able to recognize natural language search terms.
The results of the research, which took place May 31, 2005, show that merchants such as Coldwater Creek and Ann Taylor may provide top-notch email response times and shipping status updates, but their natural language search capability is not on the same level.
The research methodology employed three natural language phrases:
- Yellow raincoat with a hood
- V-neck cashmere sweater
- Boot-cut khaki pants
Except for Land’s End, which found 15 relevant products—all yellow with a hood—results for “yellow raincoat with a hood” were disappointing. Both Ann Taylor and Coldwater Creek searches failed to turn up the item, listing “no results” found. Even given the possibility that the merchants did not have a yellow hooded raincoat in stock, online searches failed to offer similar products, such as a “waterproof jacket.”
On the other hand, Neiman Marcus delivered 113 results, offering everything from sunglasses to makeup. But sometimes less is more; the first “raincoat” didn’t show up until the third results page.
Land’s End fared well with the “V-neck cashmere sweater” search, locating 13 cashmere and cashmere-blend sweaters. Unfortunately, Ann Taylor and Coldwater Creek’s search engines were unable to decipher what we were looking for and, instead, came back with the disappointing message: “No results were found as an exact match.”
A consumer looking for “boot-cut khaki pants” would be out of luck with Coldwater Creek, which gave researchers the message: “We’re sorry, at this time we are unable to locate the item you’re searching for on our site.” Although Nordstrom did not deliver exact matches, it did offer the consumer a choice of similar products: jeans and maternity pants, for example.
The results clearly illustrate that, even at the best e-tailers, much needs to be done to respond to online shoppers’ demands, which, by the way, are increasing. Research from Forrester Research,
Winning Over New Web Buyers, by Carrie Johnson, May 19, 2005) shows that more experienced online shoppers are more concerned about getting good results from site search. Some 74 percent of those who have been active online purchasers (one purchase within past three months) for four or more years say that relevant site search results are important.
Merchants must not forget that online consumers will patronize sites where the shopping process is easy and intuitive, whether that means delivering pinpointed results or similar suggestions based on the natural language search. Commercial sites need to allow visitors to use their own natural tongue to make a query on the site.
For example, on a very large electronics web site, it can be very complex to find cables that connect one hardware system to another. Cables have different end fittings, lengths and quality, based on what it is they need to do.
With a natural language processing discovery engine, coupled with the ability to recognize full sentences and in the context that is being asked, you can pinpoint the exact cable you need.
Using this type of system, an electronics firm has found that it has been able to increase its conversion rate of queries to sales to 11 percent. This is a very high conversion rate, given that the standard is somewhere under 1 percent.
It is this type of discovery technology that pinpoints answers that people are seeking, less so the large aggregators of pages, without real context, that Google and others offer at a price—based on ads served up, not conversions.
By providing valuable textual content, a site can not only increase its credibility but also improve its conversion rate.