Back in August of this year, while at Search Engine Strategies in San Jose, I sat in a session where one of the speakers talked about site search. He said something that I fundamentally disagree with but it got me thinking about why you should or should not implement a search feature on your own site.

I believe that implementing site search is smart for large sites, but only if you can be sure it works nearly perfectly. On the other hand, the speaker in this session (and I completely forget who it is) said that, for analytical purposes, every site should implement site search, even if it doesn't do a good job. This is what I fundamentally disagree with.

I'll agree that there is a wealth of information you can get from analyzing your site search usage, but I don't think it's wise to trade usability for data. Data is wonderful--and necessary--to help improve site usability, but data has to be more than just data, it has to be used to improve what's wrong with your site, not just get information.

According to this presenter, site search provides good keyword data, and it does. But if your site search blows you're doing yourself more harm than good, unless you're using your site search data specifically to help you improve your site search function.

When do you need site search and when should you forget about it?

The problem with site search on a lot of sites is that they often don't really work all that well. They'll produce good results for some searches but not for others. I've been on a number of sites testing their search only to find that there are "no matches" for exact searches on products they clearly sell. This is a problem.

When site search is available on large sites, people tend to use them. That's good. But if they search for something you have but the search doesn't find it, they'll assume you don't carry it. What do they do from there? They'll move on to the next site. That's bad.

If a site search can't produce the correct results, you're better off just removing it so your visitors will find it the old fashioned way--through navigation. At least this way you know they'll find what they want.

The first thing to look for in terms of making sure your site search is up to par, is making sure it returns results for products or services you carry. If the search can't find what is there, then either scrap it or fix it. Next, you need to make sure it produces relevant results for slightly incorrect searches. This is means that searching for "winter boots" will return results even if all your products are "snow boots".

This is where keyword research comes in handy (and, yes, site search analytic data, too!) You need to know all the different ways that people search for things that you offer. Then you need to test your site search to make sure it produces results for all the different ways people search for what you offer. Site search can't be so rigid that if something isn't described perfectly that it can't be found. Flexibility is key.

The minimum performance barrier

If you can get your site search to do those two things on a consistent basis then I'd say, go ahead and put it on your site. But, and this is crucial, always be working on improving it. This is where you can really use the analytical data you get from searches being performed. On a regular basis test as many searches as possible and see what the results are.

If you find more than 25% of searches produced incorrect results then you might need to pull your site search down and retool it some more. If you're getting less than 25% failed results then keep tweaking your search feature and eliminate the failures.

This should be a continuous process. Always test using your analytics data, test using keyword research, and test for any new products that are added.

A few other points about what your site search should be able to do. You want to be very accommodating to misspellings. If people are searching for specific product names it's easy to make a mistake and spell it incorrectly. Again, keep an eye on this with your analytics data. And finally, if you don't carry a certain product that was searched, your site search should produce results recommending similar products. Never leave your searchers empty handed.

October 1, 2008

Stoney deGeyter is the President of Pole Position Marketing, a leading search engine optimization and marketing firm helping businesses grow since 1998. Stoney is a frequent speaker at website marketing conferences and has published hundreds of helpful SEO, SEM and small business articles.

If you'd like Stoney deGeyter to speak at your conference, seminar, workshop or provide in-house training to your team, contact him via his site or by phone at 866-685-3374.

Stoney pioneered the concept of Destination Search Engine Marketing which is the driving philosophy of how Pole Position Marketing helps clients expand their online presence and grow their businesses. Stoney is Associate Editor at Search Engine Guide and has written several SEO and SEM e-books including E-Marketing Performance; The Best Damn Web Marketing Checklist, Period!; Keyword Research and Selection, Destination Search Engine Marketing, and more.

Stoney has five wonderful children and spends his free time reviewing restaurants and other things to do in Canton, Ohio.


Very good points overall. One thing to note: site search can also serve as a crutch for poor experience and site design.

If your site isn't intuitive and/or products aren't easy to find, then many visitors will just rely on site search to find what they want.

And, if as Stoney points out, they type in the wrong keyword phrase "winter boots" vs "snow boots" and can't find them, then you have just lost a customer.

Thus, consider how many visitors are using site search and determine is it because it's a crutch?

@ Jennifer - I agree, site search can be a crutch and it's a good idea to make sure that the site navigation works as best as possible as well. The problem is if the nav sucks and the site search returns poor results. This is a severe usability issue. Work on both and make sure they both function properly.

Totally agree with your assessment of the internal site search. Data is great but it always comes back to one fundamental rule.

Rule #1 It's always about the user experience.

Forget how great your SEO is, if you have a billion backlinks, the greatest free tool ever, or you are getting millions of unique visitors a day. If your site isn't built with the user experience as your main purpose then you will piss them off and they won't come back.


I run a site in which we sell steak branding irons and I use the data from searches to determine new designs that we can roll out. I think our search works well but when I review the data, generally every couple of months, I'll see trends for designs that people are interested in.

Because of this some of our best sellers were developed because of our site search. I agree that the data needs to be usable to the customer but I think it's still very important to offer a search because it can allow you to keep up with trends.

Any thoughts?

@ Domain - that's a great way to use the data. My point was only that you still need to make sure that the site search functions otherwise you're getting data that ultimately does little good. If people are using search to find things that you've added but it won't lead them there, they why bother?

Nice points about site search while implementing in our website. But in some of our websites the site search is not completed and when we try using the same to find some information they are found to be incomplete and does not show complete results. In this case what shall we do?
We can leave as it is till the full implementation is completed.
We have to remove completely from the website and once the the site search works fine in our demo server can we implement.

@ CAP - My suggestion would be to remove it until it works right. No sense getting people to your site only to have them leave because your site search tells them that you don't have the information they want.

Stoney, I really enjoy your style of writing and look forward to reading your articles, even if they cover basics I am very familiar with, but especially when you clarify a difficult to explain topic. I've had a pent up peeve about your articles' consistent lack of basic proofreading and editing, however. This article's headline finally pushed me over the fence to comment to you about it (just re-read the headline out load to get my meaning). I can overlook the regular typos, but since I like to pass them along to others (Clients, Programmers, Managers, Marketers, etc.) for their benefit and to teach concepts to them, your credibility, and mine, suffer due to such simple oversights. Please consider having another proofread your articles before going live with them. Looking forward to more of your insights.

@ Eric - wow, that was a big one. I do try to proof all my articles and generally have another set of eyeballs on them as well. Some things still get through. I always appreciate these bing pointed out so i can fix them.

Comments closed after 30 days to combat spam.

Search Engine Guide > Stoney deGeyter > Making Site Search Work for You (and Your Visitors)