On-site search is dead

This morning, we were reviewing comps for the new kulapartners.com site that our design team whipped up (side note—they’re gorgeous, can’t wait to show you). Anyway, in reviewing these mockups we noticed that the on site search box was taking up some valuable real estate that could be used for other things.

Before removing it though, we went and looked at our Google Analytics to see just how many were using the on-site search tool. Less than 0.5% of the pageviews on our site were generated by the search tool. Of those, 2/3ds were for “search kula”, and the other third were for searches I personally had performed looking for old blog posts to reference.

So, we pulled the site search from the next mockup. Why? Well, if less than 0.5% of people are using the search by mistake when they click the Go button with the default text in it and no other actual visitor is using the search, it’s not required.

We used to think that due to our reliance on web search, like Google, that users were either searchers or browsers. Meaning that some of the rudimentary analytics we had access to showed that on-site search was getting about the same amount of traffic as pages that people accessed via the navigation links. For the Kula site, that’s just not happening.

So I got interested in this and decided to look into the analytics of other sites we’ve created to see how valuable on-site search has been for many of them. For some, there was a significant volume of traffic, but it still wasn’t a huge percentage of their overall traffic. For example, on a university site there were over a hundred searches for employment related terms. I went back and reviewed the homepage and found that there was no mention of student jobs directly on the homepage. There were still more than ten times the visits to the employment page than there were searches for it. In this instance, I’d seek to provide some direction to this popular page from text on the homepage.

For a site like the one mentioned above, would I completely abandon on-site search? Probably not, but I would look to learn from it and reduce dependency on it by providing better interface design that adapts to the changing needs of site users. If your site is for a media property, such as a well-read blog or news media site, obviously site search is going to be extra important.

For most business to business sites (which makes up a large portion of Kula’s portfolio), it’s likely not necessary. If the site is properly architected, it shouldn’t be required. This of course, doesn’t preclude using custom searches within a specific context, such as business directories or product database searches. The fact is, if someone’s searching for your content, they’re likely doing most of that searching on Google or Bing, finding your site and then browsing through it. This makes the interface design of your pages even more important.

The alternative, of course is to make on-site search more useful and essential. To do this requires making most CMS search engines better than they are now. Google is constantly improving its algorithms, but the same can’t be said for most built in search engines. Even if your on-site search is excellent, it’s not going to be better or faster than Google or a well designed navigation system.

If you think you could gain back some valuable real estate from the search area of your site, it might be worth it to check out your analytics to see how it’s being used. If the searches are largely for the default text, then you can ditch it. If they’re for key content buried within your site, make access to that content easier through better highlights, links and navigation. And, if through these subtle and steady improvements you find people relying less and less on your on-site search usage dwindles, you’ll know you’ve made the right choice by yanking it.

Do you use on-site searches, or do you use a Google and then browse behaviour?

Join us in conversation…