Knowing what people search for on your site is very, very interesting.

After all, these people are already on your website. And they’re probably using your search feature because they can’t immediately find what they’re looking for. At least, that’s what we usually notice during user tests.

What do you have to do?

1. Make sure you can analyse the search queries on your website
   Earlier, we talked about how to hook up your own site search to Google Analytics.
   Of course there are other tools out there, but they’re often expensive and quite frankly not as good.
2. Analyse the list of most frequently used search words a couple of times per year
   Take into account spelling and wording variations and group these together. People looking for a ‘gun license’, ‘handgun license’ and ‘gun permit’ are all looking for the same thing. The filters in Google Analytics come in quite handy here.

Typical discoveries when analysing a search feature

• People look for things that appear to be hard to find through the navigation structure
• They look for things that aren’t on your website
• They type in old product names and even your competitors’ product names
• They don’t use the same words you do
• People can’t spell very well… at all

How can you turn that knowledge into a profit?

• Improve the structure of your website and your overview pages
  Bring the top tasks to the fore in your structure or draw attention to them by giving them a prominent place on the homepage and overview pages (also called landing pages, index pages or category pages).
  Rewrite your content where necessary. Make sure you create pages that are easily scannable.
  That way you improve your information architecture and overall website usability.
• Expand your content
  Don’t get me wrong. I’m not saying you should start churning out random content. Far from it. But, after you’ve deleted all the content people aren’t interested in (and there will be loads of it, trust me), you should think about creating content people áre looking for but that you’re not offering yet.
• Connect discontinued products to similar new products
  Don’t disappoint people who type in names of discontinued products or old product names. Tell them which current products correspond to those older products they already know and apparently still want.
  Do the same if people type in your competitors’ product names or product codes. People who do that are most likely not looking for that exact product but just something similar.
• Use the same words as your users or make your search feature smarter
  If your visitors don’t use the same words your website does, you’re the one who needs to change. Or at least expand your vocabulary. You can also make your search feature smarter by hooking it up to a thesaurus with synonyms.
• Take common spelling errors into account
  Depending on your search software (and your budget) you can automatically correct spelling errors or suggest alternatives.
  If you can’t afford to do that, and it turns out 1 out of 4 people use the same miss-spelling for a particular word, put that spelling error in the meta-data of the most relevant page, or use the wrong spelling on the page somewhere.

Examples

• Gun permit on government site
  On a government site, we noticed that ‘gun permit’ (and all its variants) was continually in the top 5 searches. Because guns aren’t exactly a popular topic, it wasn’t deemed very important and was put on the 3rd level of the structure somewhere. But it’s apparently one of the site’s top tasks.

• E-commerce site with unclear delivery information
  Almost the entire top 5 searches on this e-commerce site consisted of things like delivery options, shipping costs, payment methods etc. A clear indication these things needed to be clearer and not just tucked away under a link ‘Terms of sale’.
• Old product names and competitors’ product codes at a cable manufacturer’s site
  On the site of a leading cable manufacturer we noticed large volumes of searches on product codes and product names. Makes sense, right. But when we probed a little deeper, we noticed that a lot of the searches weren’t producing any results. Delving deeper still, it appeared that the product names were old names the cable cmpany didn’t use anymore but apparently the customers still did.
  User research taught us those people weren’t looking for support for the old products, they were simply looking for a replacement. They wanted to re-order the product.
  The solution: tell customers the old product name or code doesn’t exist anymore and show them the new products with similar characteristics.
  The result: an increase in sales.

You might also like:

• How to hook up your own site search to Google Analytics
• Navigation versus search
• Search results: layout tips
• Search = type-in field+ button

10 years ago I thought so. More specifically: I thought I knew everything.

Surely my usability expertise, my deep knowledge of information architecture would be enough to find and solve all the usability issues on every possible website?

Without research you can never truly know your users

10 years of user testing have made me a bit more modest. Okay, a lot more modest. Expert knowledge alone is not enough.

Does that mean I’m not a good usability expert? Hell no. I think I know more about web usability and information architecture than anyone else in Belgium. (Non-Belgian readers, please suppress that giggle.) Actually, that’s a lie: Els knows more about it than I do. See, I told you I was modest.

And still, despite all of our combined expertise, we often say things like ‘That depends’, ‘We’ll have to ask your customers that’ or ‘We’ll have to test that’.

Usability rules are not stone tablets

A lot of web designers and programmers resent those answers. They like it when everything is fixed. That makes it nice and easy.

Admittedly, Jakob Nielsen’s rules, the interface pattern libraries you see all around and even the usability tips we write on this blog, sometimes suggest it’s all black and white, clear-cut.

Wake up call: it isn’t. Usability isn’t that thing you read about in books. To fully understand usability, you have to see real users in action.

There are lots of rules and often those rules are right. But sometimes they’re also flexible, open for interpretation. Quite often the things that we as experts thought might cause problems end up not troubling users at all, while the things we thought wouldn’t be too much of a problem cause half of the test users to trip up.

Is that because we’re stupid? Because we’re not really experts after all?

I don’t think so. And our customers, who gave us a client satisfaction score of 100% in our latest audit in 2009, obviously don’t think so either.

Content and structure are determined by your visitors

When it comes to your site’s information structure and your content, the rules don’t really help you much. They get more than a little vague.

A few examples:

• Use words that your visitors understand for your menu labels
• Put the 3 to 5 things that matter most to your visitors above the page-fold on a detail page
• Start every page and every paragraph with the most important information you’ve got

While these rules are all true, they’re also very theoretical. Which words do your visitors understand? What exactly are those 3 to 5 most important things according to your visitors?

Pretentious

In all modesty, we think web builders or information architects who claim they don’t need to involve real users because ‘they know what’s best’ are pretty pretentious.

Sure, those web builders’ and information architects’ common sense will probably keep their customers safe from all too big disasters. But is that really good enough?

User research gets you facts

Are you really so all-knowing that you know what people exactly want to know about a hotel, an airconditioing unit and an oil-free compressor? And a forklift, a handbag and a car loan?

Really? Wow, you must be making around 10.000 euro/hour. Congratulations!

In all fairness, nobody knows those things. The only way to find out what your visitors really want is by doing user research. That doesn’t have to be ridiculously expensive. You can do some of it yourself. A good online survey and interviews with potential or existing customers will already point you in the right direction.

Facts trump opinions

One of the biggest advantages of user research is that it gets you facts. The case for the information structure you present to your client is a whole lot stronger if you can base your decisions on facts rather than just your own opninion. Data from logfile analysis, survey results, user test videos – those are things that make compelling arguments, not your opinion.

Gathering facts is the main reason why we push clients to include user research in their project. Agreed, it costs a bit more when you’re doing it, but it sure saves time on meetings full of endless discussions based on nothing more than opinions.

Related articles

• Webdesign process: is the customer king?
• Information architecture: the basics

What do you think?

Are we right or are we right? Feel free to disagree.

You’re probably thinking that should be a statement instead of a question. Well… yes, the customer is king. And no, he isn’t really.

One of the main reasons a lot of websites fail to deliver is because they’re made according to the specifications of the company that ordered it. Don’t get me wrong. It’s good for a web builder or usability expert to find out what the customer wants. But it’s not enough.

We tell most of our customers at the start of a usability project that it doesn’t really matter all that much what they think or want.

What we think doesn’t matter that much either.

It’s not your customer’s opinion that counts, it’s his customers’ opinion

You don’t make a company’s website for that company. You make it for that company’s customers.

Their needs and expectations are what matters. A website can only be truly successful if it caters to the needs of its visitors, not to the needs of its owners or makers.

Finding out what visitors want is crucial if you want to base your functional analysis and information architecture on facts rather than feelings.

Companies know their customers

Yeah, right.

That’s what I thought. But 10 years of experience have taught me that a lot of companies know surprisingly little about their customers.

Try it yourself if you don’t believe me. Ask your customer (or yourself if you’re a CEO, communication or marketing manager) questions like:

• Who visits your website?
• Why do people visit your website?
• Why do people buy your product and not your competitor’s?
• Why do they buy product x and not product y of the same series?
• What are your potential customers’ 10 most frequently asked questions?

It’s amazing how often companies don’t know the answers to these questions. Or how often the answers start with “I think” or “According to us”.

Which basically means “We don’t know”. Time for user research to make sure you can start your answers with “We know”.

User research

How do you find out what people want from a website?

That’s easy: you ask them. And you check what they’re already doing on your site.

A few user research methods we use:

• Logfile analysis
  Logfiles can tell you what people are doing on your website. They can’t tell you what they’d like to do or what they can’t do.
• Keyword analysis
  Analysing the keywords people type into your search feature tells you what people are looking for on your site and which words they use.
• Customer contacts analysis
  Ask everybody in your company who comes into contact with customers to keep track of the questions they ask so you know what the 10 most frequently asked questions are.
• Online survey
  An online survey is a great way to find out more about your visitors’ profile and what they’re looking for on your website. A short survey with the right questions still gets loads of responses.
• Interviews
  Interviews with potential and existing customers are vital if you want to get details about what they really want from your website.
• User tests
  User tests will show you what people like and dislike about your website and your competitors’.

Is user research always interesting?

Absolutely. You always learn something new.

Of course you don’t always have to use all the methods we described. Our advice is to choose at least 1 analytical method (logfile analysis, keyword analysis or customer contacts analysis) and 1 other method.

Don’t get carried away with the number of people you interview or do user tests with either. Around 10 people is plenty, unless of course you have a huge website with loads of different target audiences.

Examples

• Environmental government agency
  The agency thought their website was mostly visited by 2 types of visitors:
  - Farmers and environmental professionals looking for scientific data.
  - Companies looking for information on environmental permits.
  Research showed these assumptions were wrong. Sure, 20% of visitors were companies and around 3% were farmers and environmental professionals. But about 75% of visitors were regular people looking for information on air and water quality in their area and tips on how to be environmentally friendly.

• Air conditioning manufacturer
  Our customer thought the website was very important for engineering companies, architects, etc.
  Research showed these people rarely visited the website. The real visitors were end customers looking for information on air conditioning. And they weren’t too impressed with sales arguments like ‘titanium air filter’ and ‘only 69db’. Their main reasons to buy a particular air conditioning unit were things like ‘it’s really quiet’ and ‘it fits with our decorating scheme’.

• Hospital
  The hospital we worked for was convinced their website had to inform people about medical conditions and treatments. User research told a different story. People were mostly looking for contact information, phone numbers of doctors and visiting hours. As a matter of fact, more people used the website to look for information on the cafeteria than information about a medical condition or treatment.

• Scientific research institute
  User research showed that the visitor profile was very different depending on the language version of the website. The Dutch version of the site was mostly visited by teachers, students and non-scientists, looking for general information or didactic material. The English version of the site was mostly visited by staff and scientists from other research institutes and industry, looking for information on specific research projects or scientists’ contact details.

So don’t talk to the customer?

Of course you should. Talking to your customer is always important.

But don’t just talk to the CEO or the web manager. Talk to people from all divisions of the company so you can find out what their expectations are about the website.

And always, always check the company’s wishes and expectations against those of the people it’s all about: the website’s visitors.

Conclusion: talk to your customer and his customers

What does all that talking get you? An excellent view on what the website should and shouldn’t do and what the visitors’ top tasks are.

The end result: happy website visitors. Which in turn leads to a happy customer for you.

Care to share your experiences?

We’d love to hear from you.