9 Mar 2016
Usability tests that will improve your SEO
Reading time: 19 mins.

Usability tests that will improve your SEO

Usability tests that will improve your SEO

At first the SEO discipline was a question of reading a lot of articles and carrying out various tests, although there were disciplines such as programming (especially html) which could give you a significant leap in quality.

Over time we web positioning professionals have become “little one-man bands” that have inevitably acquired knowledge of various disciplines very closely related with our world, such as

  • Web analytics
  • Writing
  • Online advertising
  • Servers
  • Usability…

Personally I am very passionate about the last one, usability. It is a discipline to which I have dedicated a large part of my day for quite some time. In fact I believe it defines the route we SEOs have been taking throughout these last years: “from traffic to conversion”.

For a director of web projects mastering both disciplines is fundamental and ensures to a large extent that online business objectives are achieved. It is also important to understand all positions when the usual war breaks out between designers and technicians during the development process.

Usability test

To optimise the usability (and therefore the conversion) of a website different techniques are used, some based on the previous knowledge and experience of the development team and others, the more interesting in my opinion, based on the actual user experience.

To make decisions based on user behaviour we generally use two sources of information: web analytics or the user himself.

While the analytics offers us large volumes of information to interpret, the user test (or usability test) offers us real conclusions first hand. Although with much smaller user samples of course, not everything was going to be perfect!

Through a usability test one or various pages of a website can be audited in order to help detect flaws that a team of various people often overlook. Things are always seen more clearly from outside, without the flaws of the development process itself.

These tests can be carried out in test environments or production environments and in different phases of the online project:

  • Initial development
  • Redesign process
  • Web audits…

What does SEO have to do with all this?

Firstly it should be noted that the previously mentioned phases of a web project are usually the same ones in which we put in the most effort for SEO. Therefore it would be logical to make the most of these tests to obtain relevant information from a SEO point of view.

In fact the usability itself and SEO share some basic objectives. Two very clear examples:

  • Terminology = Speak in the same language as the user
    Using the correct lexicon will allow being present in more quality searches in Google and also allow the user to consume more efficiently the content of the page
  • Web architecture = Clarity and simplicity
    A well optimised hierarchical page structure helps not only the user to find what he is looking for but also the crawlers to correctly index and catalogue the information on the site.

And even though the usability guru Jakob Nielsen said that “SEO happens before the first click, and usability takes over from there”, Google insists on evading that theory through an algorithm where the user experience is increasingly important.

For all these reasons the following tricks and advice will be given in order to obtain information of a SEO nature from a usability web test.

Orienting the usability test to SEO

To start it should be remembered that a usability test is just that, a usability test, and the results cannot be compromised only to obtain “clues” with a view to enriching the site’s SEO.

Anyone not previously familiarised with user tests is better to focus on testing that which the test is being carried out for, and maintain a neutral stance so as not to condition the results, or the test will in itself be a failure.

Therefore the following tips should be applied in moderation, by professionals with some experience and only when it is not in conflict with the objectives of the actual test.

It goes without saying that the best way to organise subsequently all this information is by recording videos of the navigation sessions, even when they are not going to be presented to the client, so that they can later be reviewed internally and notes taken on everything.

Start the test with a search

If the test is going to be carried out on a published website (and indexed by Google of course) the user can be asked to start the session by performing a search.

It is usual to start usability tests on a “neutral” page so as not to condition or mislead the user, and often (at least that´s how we do it in Aukera) the webpage used is the Google homepage.

Google search

Taking advantage of this situation we can, after explaining the first task to be performed, suggest that they enter in the website not directly through the URL but by through an online search (in fact the tendency of the user is to use the search box even when they know the URL).

In this way we see how the user searches for the content in Google, obtaining generally good hints and search patterns that can help us to improve the SEO of the page.

Example 1:

A test to check the online sales section of home appliances of a large department store, can be started by suggesting to the user:

“As your fridge has broken you are going to buy a new one in “Dabanhams”, so try to enter directly in the section that you think the fridges will be found”

In this way we can see how the user searches in Google with terms such as:

  • fridges dabanhams
  • home appliances dabanhams
  • home appliances section dabanhams
  • white goods dabanhams


  • dabanhams fridge
  • dabanhams online fridge
  • dabanhams buy fridges
  • dabanhams home appliances

Example 2:

To start the test with even more information in terms of SEO, we can involve the user to a greater degree, using his actual case. For example he can be asked:

“If you were to buy a new fridge and you didn’t have any budget restraints, what characteristics would you like it to have that your current fridge doesn’t include?”

Note that on this occasion the term “refrigerator” is used as well as fridge. It is not too much to give the user a range of terms to use to ascertain which he decides to use or how each is used, if we only state one continuously we will again be conditioning.

With this question we manage that the user orients the search on one characteristic of the product (instead of product type) and in this way detect new search patterns.

We ask the user to look for the type of fridge on the website that we want to test. Imagine that what the user wants is a bigger fridge, he will look for things such as:

  • big fridges dabanhams
  • wide fridges dabanhams
  • spacious fridges dabanhams
  • double door fridges dabanhams
  • XL fridges dabanhams
  • large capacity fridges dabanhams
  • 500 litre fridges dabanhams

You can be a little bad and take this practise further. If we don’t tell the user which site is being tested (we don’t specify that he has to reach the web of Dabanhams) he will carry out a generic search (not brand) in Google and will find results from the competition.

In this way we can see how he behaves with the different snippets and even ask, before entering any of the sites, in which result would you click and why.

This can give us additional information to know which works better in the case of:

  • Meta tags
  • Rich snippets
  • Adverts vs. Organic

Lastly there will be times when the user can be asked how would you look for something very specific?, this is especially useful for products that follow a wide variety of search engine patterns.

An example would be technological products that regularly respond to searches by technical specification and also are susceptible to brand and/or model searches, amongst others.

This practice will be limited to test for products that the user is familiar with first hand, in order that he has enough information to decide on one specific type of search and therefore make the task minimally “personal” and add variety to the test.

Example 3:

Lets asume that a test is carried out on the section of mobile terminals for the fictitious telephone operator “Narange”. A subtle task to not influence the user could be:

“Imagine that your current mobile is no longer working well, however you are so happy with it that you are going to buy one exactly the same. Access from Google a page in order to do this”.

In this way we can understand how they classify mobile terminals in their mind and how they organise this information when formulating a search:

  • apple mobiles
  • android telephones
  • iphone 6s
  • smartphone htc

And if the sample is sufficiently large we will find a range of search patterns:

  • mobiles + [operating system]
  • mobiles + [operating system]+cheap


  • [brand] + [model]
  • [brand] + [model] + [characteristic]…

Arriving at this point some will think that, for a large sample, it would be better for the Keyword Planner and other tools that allow a complete Keyword Research to lend a hand. True, however the difference is in the human factor: people in place of statistics.

In usability improvement projects the definition of “people” (and “settings”) is standard in the process in order to establish demographic, cultural and – in the case of the internet – technical patterns in order to “give a face” to the user and define targets that are more specific.

In this way not only do we detect search patterns but we can also identify “people” typologies or, where we already have this segmentation, try to distribute the users participating in the test within each group of “people”.

Definition of people to improve web usability

Putting this into practice, a usability test that includes enough users and diverse socio-demographic conditions can provide us with information like:

Identifiable users like the person “Peter”

  • Searches with a lot of words that describe the product or need.
  • Read carefully the snippets before clicking on one.
  • A lot of patience in finding the information once on the site.

Identifiable users like the person “John”

  • Succinct searches avoiding stop words.
  • Quickly click on the snippets.
  • If they don’t find information immediately they go back to Google.

Identifiable users like the person “Katie”

  • Very diverse searches.
  • Often click on adverts, do not discriminate.
  • If they don’t find the information they return to Google and change search.

To sum up, starting the test with a search will only take the user a few seconds and also will not interfere with the remainder of the test, so carrying it out will almost always be of interest.

The crucial thing is that the user ends up reaching the page in question (in this case the fridge or mobiles section) and once there starts the rest of the tasks that actually make up the test.

Ask what is expected of each section/category

Throughout the test it is typical to talk, talk a lot, with the user. It´s all about understanding what is going on inside his head as he uses the website and that is the main difference between cold analytical data and data obtained directly, with feedback from the user himself.

One of the practices that we find most interesting is, once the user has arrived at the site in question, to tell him not to click yet on any element and ask him an open question:

“What type of content do you expect to find when you click on this section of the web?”

This can be more than interesting when testing the effectiveness of the main navigation menu or secondary menu of an e-commerce. But additionally the question can be suitable for other clickable elements such as an anchor text:

  • “What do you expect to find when clicking ‘Choose a size’?”
  • “What do you think is behind the button ‘World foods’?”
  • “Where would you reach if you click on ‘Exclusive products for you’?”

With this we obtain valuable feedback on the link texts (anchor texts) that we use, knowing if they are sufficiently descriptive and intuitive, in such a way that the SEO reward will be double.

  • Improve the relevance of internal linkage
  • Optimise semantically both the content and snippets

Use the internal search engine

We often think that “SEO” is only optimising for Google (ok, and for Bing too) but in reality that is not totally true.

If the initials are a reference to “Optimisation in Search Engines”, why are we going to neglect the only search engine over which we have total control?

Internal search engine

If this kind of practice is especially interesting in an e-commerce, they can also be carried out in corporate or content websites and the system can even be extrapolated to test usability of Intranets and other applications with large quantities of information.

A special case would be the usability tests that study the internal search engine of the site itself, as more time can be dedicated to these tests (and therefore more “clues” that allow results to be extrapolated to Google searches).

In such a case, a recommendable practice is to offer the users different search results, that is to say, carry out an A/B test of the snippets shown after the search. This is very interesting to draw conclusions that can be applied to the meta tags that we suggest to Google for its snippets.

Returning to the internal search engine of the web site and electronic commerce, these are a rich source of information. The internal website searches that the users carry out will help us to know firsthand several things of interest:

  • What products / categories are hardest to find for the user
    More interesting from a usability and conversion point of view
  • How a user searches for products / categories
    Again key when orienting the content to SEO level
  • What non-existent content is being requested
    Exceedingly valuable to generate new content and opting for more searches

In this way a series of tests can be suggested to the user throughout the test that will oblige him to use the internal search engine to continue to collect valuable information to improve positioning.

It is generally better not to suggest this type of task at the beginning of the test but to carry out other tasks prior to this, otherwise we will be influencing the user by telling him that an internal search engine exists. This could give unwanted results:

  • Cover up a usability problem of the search engine itself
    If we force the user to find the search engine, we will never know if he would have tried to use it on his own to carry out certain tasks.
  • Corrupt the behaviour of the user during the rest of the test
    There are those that once having tried the internal search engine will start using it for whatever task, thus changing their behaviour.

Example 4

In this example the user is proactively advised to carry out an internal search on the site, without waiting for him to find navigation problems

Between two different tasks the user is taken to a task that involves an internal search, let’s say the site is for small electrical appliances:

“Use the site’s search engine to find a blender that can be used to crush ice to be used in making different cocktails”

Some possible search terms are:

  • glass blender


  • ice blender
  • blender to crush ice
  • cocktail blender


  • steel blade blender
  • high powered blender
  • 900w blender


  • ice crusher

It is interesting to suggest searches that help to overcome certain SEO challenges that have been posed when organising the content of a section. If all the users of the test are asked this same question, we can orient the content far better for this type of product.

Example 5

In this further example we imagine that the user finds a problem in finishing a proposed task, therefore it is suggested that he helps himself with the internal search engine to progress in his task.

Let’s say that a site selling second hand cars is being tested and the task proposed is to compare the different purchase options and contact the seller. The user has children and therefore has decided to check the price and characteristics of some MPVs.

However he struggles to segment this type of vehicle using the characterisations that the menus are divided into, he doesn’t find this type of vehicle in “SUVs” or in “family cars” nor in “vans”. Seeing that he is struggling to progress it is proposed to him:

“Why don’t you try to carry out a search using the site’s own search engine?

On reaching this point and following the example, it would be most likely to find queries focussed on resolving the segmentation problem, something like:

  • MPV
  • MPVs
  • MPV [brand]
  • MPV [place]
  • MPV [characteristic]


  • Cheap MPV
  • Diesel MPV
  • 7-seater MPV

Note that these searches fit into the context of an internal search engine, the user already knows that he is on a website selling vehicles. Therefore it is logical that he economises terms in the search string, for example in Google he would include “buy” prior to the word “MPV”.

This would be a good moment to ask the user to repeat the search suggesting other contexts to see how he handles specific needs in the form of the search term:

  • “Now imagine that you have a maximum budget of 10,000€”.
  • “Let´s say it has to be a car you are going to go and pick up in person”.
  • “You have to make sure that the vehicle will allow you to benefit from certain government assistance”.

In all of these cases the sample of users will always be diminished but the exercise allows us to get inside the head of the user and understand why he is searching in this way. The additional explanations which the user gives us first hand often help us to optimise keyword research based on including or removing certain terms that a priori prove to be ambiguous.


If social signals (supposedly) are not a direct factor of SEO positioning for Google, it is true that there exists a strong positive correlation between their existence and the rankings.

Generally a site that is heavily shared is a quality site, but also the fact of growing the reach of a publication or piece of content increases the chances of generating external links (direct SEO factor) and improves the brand recognition on the part of the user (which will in the long run positively affect other direct SEO factors such as CTR or Pogo sticking from Google).

In the same way that starting a test with a search hardly takes any time and provides interesting information, finalising certain tasks testing the shareability of the content will also be quick and valuable for the general optimisation of the site.

In a usability test it would be surprising that the user, without having it suggested to him, would try to carry out an action to share in social networks. Therefore the appropriate moment has to be found to suggest these types of actions. A good idea to obtain more information is to propose 3 prior questions:

  1. “Do you think the content is good enough for you to want to share it using your profile in social networks?”
    Although the user tests are highly dynamic, this question should serve to detect aspects that negatively affect the trust and credibility conveyed by the site.
  2. “In which social network or networks do you think it would be pertinent to share it?”
    In this way useful information is obtained on the point of configuring social buttons, such as defining which social icons to include and in which order to show them.
  3. “Would you have picked up on the fact that there is an option to share?”
    Very often a user will not recognise that he has struggled to find an element if he is not explicitly asked, so the location of these buttons can be improved.

After these questions the user will be asked to proceed to simulate the sharing of the content through the social network he has chosen.

The idea is to detect usage problems and other flaws of these social buttons to optimise their implementation in the pages and boost in this way social dissemination of the content, as a way of collaborating indirectly with the site´s positioning.

Usability in itself

It is well known that Google rewards quality sites in its rankings, including a good user experience, something as right as it is intangible. What is clear is that a usable site is a better site. And a good site always has a head start when it comes to SEO.

One of the direct SEO factors that will be seen to benefit from a good usability in the landing pages will be “Pogo sticking”, as the user will find it easier to find the information he is looking for in our page without having to return to the Google results.

Additionally it has been shown that those pages where the user perceives a good experience have more possibilities of being shared and linked, leading the usability of the site again in direct and indirect SEO factors.

Pogo sticking and web usability


You have to know to extract useful information for SEO in all phases of a web development project, from the preliminary conversations with the client right up to the feedback from the users that convert, going through the data from the web analytics and other tools such as usability tests.

Defining an intelligent script for the usability test will allow the results to reach another level, generating highly useful information for the usability and conversion managers, as well as for other profiles within marketing online such as is the case for search engine optimisation.

If in addition a meticulous selection of “target” users is carried out for the test (actual potential clients of the brand) this will always contribute something extra compared to the cold and sterile data of the keyword research based on online tools.

The keyword research or web analytics give data based on a much larger volume of users but, in the case of usability tests, the information is 100% valid. There is no place for interpretation; if we still have any doubts… we have the user there to ask him.

Therefore you can exploit the value of an actual user and compile the user test from now on thinking also of SEO.