Issue 4 – Too many results and no possibilities to refine search results
- Data management and quality
- Digital services
- E-commerce and Omnichannel
The ultimate guide – how to solve 4 most common issues with site search
The site search seems to be an area in digital development which is hard to crack for the most companies. In this blog post series Tuija Riekkinen and Hans Ahlborg capture the most common issues and remedies to ensure a better performance of the most neglected sales booster of your website.
Like the collaborative writing process of this blog post series proves it, the key for tackling challenges with the site search, requires collaboration and dialogue between different digital competences – a multi-discipline team.
Issue 1 – Site search is not promoted – Design conventions and other reasons
Issue 2 – Missing content & data structures prevent engines to do their job
Issue 3 – Search results are not relevant – to anyone
Issue 4 – Too many results and no possibilities to refine search results
If the hits are relevant but there’s too many
Even if the site search returns correct results, the amount of hits can be high and it can be difficult for the user to separate which products are important for them right now.
There are a number of ways that can help the user to refine the search results.
Filtering and facets are very useful function to add to your search results page, however not all search engines are pre-configured for this and will need additional development to get them function properly.
And even if they were, the information architecture and data model might need to be worked on.
So, be prepared to continue looking under the hood before starting to implement filtering on the search results page.
How to solve this?
Large e-commerce sites often provide search scoping as an option to refine search results.
Scoping means that user can limit the search query to only one product category. Either before submitting the search query or when reviewing the long list of results.
While this might seem like a good and simple idea to implement, remember that it is actually only as good as your categorisation structure. And designing a category structure, which is intuitive and easy comprehend to majority of the users, is one of the most demanding tasks in e-commerce.
Categories are exclusive. When a user chooses a category, all the products outside of that particular category are ruled out.
Choosing a category might limit the selection too much or, since people perceive things differently, the results are not what user expected to see.
Especially, if the objective of improving the site search is to compensate categorisation structure, providing the same categories for search scoping, might not be a good idea.
However, if the categories are tested to be fine and you have good filters connected to them, scoping search with category selection is a great way to improve the search experience.
Do you have useful and well-functioning filters connected to the category structure? If you have, there shouldn’t be any reason for you not to start exploring how to use filters also on search results page.
If not, then it is very likely that your organisation has been neglecting the information architecture and information activities and instead of a mile, you have a couple of extra miles to go before you can start optimising the experience of search results -page.
But if you do have filters in your categories, start by defining how your organisation wants to use them on search results page. The simplest way is to provide filters applicable to all products as a way to refine the search results.
If the number of common filters across the range is so low that they are not really helping in the search result refining, then ability to present filters based on a relevance is needed.
One way of doing it could be simply defining a threshold percentage i.e. if xx% of products matching the results have shared filters, these filters are presented to the user in the context of search results.
In order to implement intelligence like this, the information architecture and data management processes need to be in a good shape.
For example, if there are filters (e.g. material) which are applicable to products in different categories, they need to be one entities and shared across all categories (and products).
In other words, if you have the data in place, some harmonising effort might be needed.
In information management sharing the filter attributes means that also the values are shared.
In practice, one subset is used for products in category A and another subset in category B. If products both from category A and B are included to the search result, the UI needs to know which values are relevant to which category and display the values for filtering with context.
For example: if the search result includes both tables and table cloth, filtering with “Material” and value “Linen” would rule out all tables. It would be less confusing to the users if the filtering user interface would clearly state that “Linen” is valid filtering option for table cloths and materials such as “Birch” for tables.
So, if you set your bar on this level, be prepared to work closely with your organisation’s information management people.
Faceted search brings a multiple filter set-up where you can use both the categories and the attribute driven filters and with a possibility to change results in real-time as you tune your search options (also called instant search).
However, don’t be fooled to think that quantity fixes the quality. Populating the site results page with badly performing filters, is only waste of important real estate for displaying the results.
Deploying language tools will help the user to refine and correct their search before they hit the button search.
Autocomplete for words or phrases is a tool that will help the users very much and is nowadays much smarter than before. Today autocomplete helps with search results in a real-time and can change search results while typing.
“Auto-correction” and “did-you-mean” together with spell-check of text in the search box is a very valuable tool. It helps both people with dyslexia or other language issue to search correctly. It also helps the search editor to spend less time cleaning out all search phrases with typos in the editor tool for your search engine. This will, over time, affect both speed and database size.
Recommendation engines are today almost indispensable for helping users to find associated items they are looking for. Additionally, it is a tremendous possibility for any site to cross-sell or upsell, either e-commerce goods or content, similar to what the user was looking for.
Most search engines today have at least a basic functionality for recommendations but you will most likely need to invest in development to ensure that it matches your specific needs.
All our experiences when using a recommendation engine that supports upsell or cross-sell shows that the sales conversion and content usage will rise with up to possibly 30%. That is an opportunity that should not be missed.
But! To implement a recommendation engine, you MUST have done your information management and architecture properly, you must have your categories right, you must understand how your users search and you must have invested in a good team that have the right tools to ensure a good user search experience.
Psst. Read more about designing an effective cross- and upsell model based on research about impulse purchase.
- If the site search returns too many results, offer ways to refine the search results and focus on them when designing the search results page
- If your product categorisation structure is not performing well, think other ways to refine search than by choosing a category
- If you want to offer more advanced ways to refine the search results or even use it as a clever recommendation engine, be prepared to work with tedious details related to information management