Information architecture of search facilities on financial websites

Posted on by jasonbrown

Previously I analysed the information architecture of supplementary navigation on financial websites, now it’s time to look at searching.

Brief overview of searching on financial websites

I have looked in relative depth at searching recently and the different ways of refining results and how the different methods can be used. It appears that the vast majority of financial sites I have examined do not offer a search facility and if they do then it tends to be a very basic one with little-to-no options for filtering and often returns a lot of results. It appears that on financial websites the expectation is that the user uses the various navigation methods to find what they are looking for. Perhaps the assumption is that a user only visits a financial website if they know exactly what they are looking for and so will navigate to what they’re looking for without searching.

Search button position

Typically, on financial websites the button tends to be placed in the upper right of the site (or on that side of the page) and is typically just a search icon (the generic magnifying glass), or it may have a rather understated ‘Search’ label with the magnifying glass icon next to it. The two exceptions I found on my sample sites were Nationwide and Admiral, whom both present the user with a large search box located in the centre of the header in Admiral’s case and in the centre-right of the header in Nationwide’s case. This makes the search more of a call-to-action, so perhaps these companies feel that for some things, a user would prefer to search.

Admiral’s search facility is one of the main calls-to-action in the header of their site, being placed in the centre of the header and directly to the right of the logo. It is even placed above their other menu buttons!

 

Nationwide’s search facility is not as over-stated as Admiral’s mainly due to its positioning, but not as under-stated as most. It looks tidy whilst inviting the user to search by looking obvious.

Search suggestions

Not all of the financial websites I examined even had search suggestions, but the ones that did showed relevant results and presented in a visually-appealing and easy-to-use way. Search suggestions are helpful because not only do they speed up searching, they also allow the user to check if what thy are looking for exists on the website. Search suggestions are so common these days that it feels weird using a search facility that does not offer this – it almost feels like ‘taking a shot in the dark’ when searching without one.

Lloyds give a few relevant search suggestions for ‘credit’, not overwhelming the user with a lot of options to look at and keeping the search interface clean – probably good because the site already looks a little ‘busy’!

 

Barclays give a few more search suggestions for the same search.

 

An example from the world of insurance, Admiral’s search suggestions panel covers the majority of the webpage, reenforcing the action of searching to the user. The use of their blue branding is also extremely prevalent in their search suggestions panel.

Incredibly, Lloyds and Barclays were the only major UK banks that I could find that employed search suggestions on their site and Admiral the only insurance site in my small sample that did.

Search results and refining

The search results pages generally look very similar between the companies with not a lot differentiating them. Most offer little-to-no filtering options and provide a lot of results, signalling a poor search algorithm, but results tend to be contextualised with some text underneath the returned page. Perhaps search facilities on financial websites are not the most important feature of the site and developers would rather spend time focusing on improving online banking portals and security – which are both probably more important! Perhaps the financial companies hold analytics data that prove that searching is not a commonly-performed task, therefore investment in research, time and money is not appropriate for search facilities or algorithms.

Searching for ‘car insurance’ on Aviva renders 836 results – and the user can only filter between ‘All’ and ‘Documents’. Items are contextualised, but it seems the algorithm is picking up results which also include the word ‘insurance’ without ‘car’.

 

Admiral renders 586 results for ‘car insurance;, suffering from the same problems as Aviva’s site. There is also no way to refine results.

 

Of the three insurance sites I tested for this post, Direct Line was arguably the most accurate when it came to searching for car insurance, with 387 results found. There are also no filtering or sorting facilities.

 

CitiGroup’s search facility only allows for results sorting, no refining options are available. 987 results are found for ‘credit’.

 

NatWest offers a very small number of filtering options, shown in the upper left pane of the search which is a conventional location. Only three things can be used to filter results though: FAQs, Support Centre and Travel. These cover the basis of NatWest’s services, however filtering by FAQs will render 380 results whereas filtering by Travel will only render 2. It seems unbalanced.

 

Nationwide offers more filtering options than NatWest – the user is able to filter results by ‘All’, ‘About’, ‘Guides’, ‘Products’, ‘Support’ and ‘News’ by using the buttons at the top of the search results which is a slightly unusual location for the buttons.

 

Barclays offer a massive range of filters that the user can choose from, so many so that they look unsightly, require a separate scrollable panel and it takes a while to find the filter to refine by – ironic! But at least they offer something to help refine results.

 

Of all of the examples I found, Lloyds probably offered the best mix of results and filters. Filters are arranged in order of results on the left side and the list of filters doesn’t require its own scrolling section, Searching for ‘credit’ on Lloyds’ site did however produce a whopping 5,370 results, hence for the filters!

It seems that generally the insurance websites have the weakest results pages with no filtering and a lot of results being shown and that the banks have better search facilities, though sometimes filtering options are a little sparse (NatWest) or a little overwhelming (Barclays) and hundreds or sometimes thousands of results are found per term.

Search algorithms

These are definitely using a ‘recall ratio’ algorithm, where the number of search results is defined by the number of relevant documents retrieved divided by the total number of relevant documents in the system. This makes for a lot of search results since search terms are checked against content on every page on the site.

Search query builders

Most of the search facilities demonstrated in this post are capable of picking up on things like spelling errors and some even allow the user to ask a question, but I feel that it only finds answers where the pages it is returning have questions in the page name. The search algorithms are nowhere near as complex as that of Google’s, so basic pattern matching is used to construct queries with an element of spell checking, phonetic identification and natural language processing tools being used too. Stemming tools and controlled vocabularies and thesauri are often not used.

Barclays automatically searched for ‘credit’ instead of ‘cerdit’, which had spelt incorrectly.

 

As you type a question into Barclays’ site, it comes up with question search suggestions.

 

When you search for a question on Barclays’ site without using the suggestions, the site will try to answer it for you – in this case it got it wrong by offering information about a recent data breach, but the idea is good and shows the thought processes behind searching using natural language terms.

 

Lloyds also shows possible question suggestions when you start typing a question into their search facility.

 

However, Lloyds did not automatically search for ‘credit’ instead of ‘cerdit’, despite picking up the error.

 

Nationwide was able to identify spelling errors.

 

They were also able to identify questions and find relevant results, but I feel that it might have been looking for pages with questions in the title.

 

Aviva couldn’t detect the misspelling of the word ‘car’, but did find some results to do with something called ‘CRA’ and ‘insurance’.

 

Their search facility did fare better when I put an additional E onto the end of the word ‘insurance’, finding the page about car insurance which is what most people would want.

 

Their search facility did allow the user to type a question and provide some relevant results, however, though I feel that actually typing insurance with two Es rendered better results than typing this question.

Conclusion: why don’t financial sites have the best search facilities?

In addition to the other reasons mentioned in this post, it must be considered that these sites have tens of thousands of pages sometimes, so the user could search for anything and it could be on thousands of pages. How do you design an effective filtering system that covers all bases? NatWest have limited filtering options, but it covers all searches. Barclays has more comprehensive filtering options, but the interface looks cluttered and there too many filters.

It’s interesting to note that on their business site, Lloyds do not have a search facility at all. Business users are perhaps encouraged to contact the bank directly about FAQs and issues rather than find information for themselves in the interest of security.

It is likely that other things on financial websites take preference over creating amazing search algorithms and search experiences. I believe that customers tend to go to banking websites to find out about accounts and do online banking and they go to insurance websites to get quotes – neither of these really require searching for anything. Though that being said, The Financial Bird puts a search engine strategy at number 6 of 10 great features of banking websites, so perhaps there is a place for it. I guess anything to make finding information easier is important, as Usability Geek points out.

References

Arrango, J., Morville, P. and Rosenfield, L. (2015) Information Architecture, 4th Edition. Sebastopol, CA: O’Reilly Media

Callendar, J. & Morville, P. (2010) Search Patterns. Sebastopol, CA: O’Reilly Media

Creighton.edu. [online] Available at: https://www.creighton.edu/fileadmin/user/HSL/docs/ref/Searching_-_Recall_Precision.pdf [Accessed 22 Oct. 2018].

Brown, M. (n.d.). 10 Features of Great Banking Websites. [online] The Financial Brand. Available at: https://thefinancialbrand.com/42072/bank-credit-union-website-design-tips/ [Accessed 22 Oct. 2018].

Arad, D. (2012). 3 Ways Your User Experience Team Can Improve A Banking Website. [online] Usabilitygeek.com. Available at: https://usabilitygeek.com/user-experience-improve-banking-website/ [Accessed 22 Oct. 2018].