We use cookies. You have options. Cookies help us keep the site running smoothly and inform some of our advertising, but if you’d like to make adjustments, you can visit our Cookie Notice page for more information.
We’d like to use cookies on your device. Cookies help us keep the site running smoothly and inform some of our advertising, but how we use them is entirely up to you. Accept our recommended settings or customise them to your wishes.

Interview: Michael Crandell, Celebros Site Search

Celebros provides site search, navigation and merchandising solutions for e-commerce websites. Recently, we had a chance to interview Celebros CEO Michael Crandell about site search in the world of online retail.

Larry Becker: Thanks for spending some time with us today, Michael. You lead an organization that’s been working with site search and merchandising since 2000. During that time, search’s importance to conversion has gained increased recognition and online search’s overall visibility has soared. How have these changes affected the way site owners look at their own search functions? How have these changes affected Celebros?

Michael Crandell: For starters, the conversation no longer starts with basic education of what improved site search is about -- almost everyone gets that, and they know they need it. The focus has changed to how search can help a site merchandise products in various ways beyond linguistically relevant search results -- and that's our major emphasis at Celebros today.

As retailers woke up to the significance of site search, it went from being a little box that was often internally ignored, to an acknowledged component of the UI. How far do you think we’ve come in understanding the significance of the UI to overall search performance, and what still needs to happen?

I think we're still just scratching the surface of UI improvements related to search. There's still too much reliance on shoppers "knowing" how to use a site, and many sites are crowded and confusing to the eye. It's a tough trade-off between simplicity and providing more information -- and that's our challenge.

Some things we've done are to offer both list and grid views in search results pages, AutoSuggest of similar searches from the search box (see for example www.karstadt.de), and pop-up product detail on search results pages similar to what Netflix does for movies. We'll see a lot more fast Ajax UI's in the next 24 months, with novel ways of accessing product information. We also need to become better on the back end at determining true relevancy -- almost to the point of reading the mind of the shopper.

For a typical retailer what are some key performance indicators that should be tracked to gauge and improve site search performance?

We look at % of site visitors using search over time, number of refinements accessed, product views and purchases. Purchases should be analyzed both in terms of overall search conversion, term-specific search conversion, and # searches per $ sold. Beyond standard analytics, we suggest monitoring both specific text searches (e.g. "men's diving watch") and searches within a broader concept (e.g. "watches"). Often a site will find that the #1 search concept is not the same as the #1 text string searched. For sites that use our navigation system for the left navigation and browsing links, we also track browsing clicks.

Beyond those basics, a site should seek to interpret the data in relation to the characteristics of their customers and shoppers. For example, if a store's shoppers have historically been devotees, they may have shown a lot of "persistence" in searching. The shoppers may have been willing to enter 6 or 7 search strings to find a product and buy it. But if that store is trying to expand its audience to more casual shoppers, reducing average searches to 2 or 3 per purchase will have a dramatic impact as the store expands its audience.

For that same “typical retailer” what kind of return can they reasonably expect from an investment in a third party search and merchandising tool like Celebros’s?

Results vary widely, but I would say 10-25% is not uncommon. We've had results ranging from 3% to 300%. A lot depends on how bad the existing search is, how much traffic the site gets, quality and organization of the product data, and the nature of the site visitors (casual shoppers, loyal followers, etc). We get asked this question all the time -- at the end of the day our pricing model guarantees an ROI to the online store or they don't have to keep our solution.

What are three indicators that site owners should look for to determine whether they’ve outgrown their current search solution, be it in-house or third party?

  1. Feedback. Customer emails, internal department comments, the boss's husband or wife -- anywhere you can gauge how easy it is for people to find what they're looking for.
  2. Conversion. Site search conversion is normally higher than general site conversion by a strong margin. A study in late 2005 saw site search converting at 3 times higher than browsing. Some analytics companies can off you data on your site search performance in relation to others in your retailing sector.
  3. Flexibility. Are you able to change site search behavior easily -- or can you tell your vendor to make changes for you on a service basis? Do you have the ability to merchandise products easily, change product rankings, highlight bestsellers, etc?

What makes the difference between a successful and suboptimal implementation of a third party search and merchandising solution? Who in the organization needs to be on board and what do they need to do make the solution perform to its potential?

Ideally, both the product merchandising staff and the IT department should be working together. That's not always possible in terms of IT resources, and so we often deliver our hosted solution, which reduces the IT involvement down to organizing a product feed and redirecting the search box. But the more the product merchandising staff is involved in specifying what they want to see from the search behavior, what past problems they want solved, what UI they want to test and verify -- this is extremely helpful in producing a successful implementation.

Our firm often describes site speed as the most overlooked and critical aspect of usability. What’s your take on the importance of speed in search? And what do think is often overlooked but always critical within site search?

Speed is paramount -- since day 1 of computing, fast response is always something that creates a better experience for users, and e-commerce shopping is no exception. Celebros designed its search solution to hold all information in memory, making it an order of magnitude faster than traditional disk-base database solutions.

Often overlooked (amazingly): searching with no text (empty search), zero results handling, spelling and phraseology variations and errors, business rules for ranking of results. This last one -- the ordering of results to match merchandising ranking rules -- is a big conversion driver in our experience.

There’s a very small set of tools on the market that compete for the same business as Celebros. Have you noticed any consistent reasons why one tool wins business over another in this competitive space?

I like to focus on our wins, Larry. We just added our 200th customer, and we're very proud of that. In our experience, a lot comes down to the customer-vendor relationship, and the business terms. Will your vendor not only be around, but be responsive to you? With whom can you talk at the vendor if there's a problem? What happens if the system doesn't perform to your expectations -- are you locked in for 3 years? Obviously, we think we answer those questions in a very good way for our customers.

Google’s entry into fields like site analytics and MVT has not made things any easier for other vendors in those categories. Do you have any thoughts you’d be willing to share about Google and site search?

Google has had a site search product ever since I've been in this business. It doesn't measure up to the state-of-the-art solutions like ours because it's 100% automatic. The experience it provides is like walking into a bricks and mortar store and being greeted by a robot instead of a human being. No purely automatic solution will do as well as a customized, hand-tailored solution in the foreseeable future.

That said, Google is an incredible company that keeps making life easier and cheaper for us with the tools they offer -- all except paid search that is :-) Yep, paid search is not free, but done right, it works. :- ) Ok, time for the requisite crystal ball question. Michael, What will site search look like in three years?

Site search is going to merge increasingly into a form of product recommendation engine. Behavior-based and other forms of analysis will make the products presented to visitors more and more relevant. The user interface is going to be simple, elegant and powerful. It's going to be a fun and exciting world to work in.

We think so too-- we're pretty darn happy to be working in this space. Big thanks for your time today, Michael—much appreciated!

Join the Discussion