Customers who use on-site search are far more likely to convert than those who simply browse for what they want. To ensure the best possible conversion rates and user experience retailers should analyse the performance and functionality of their on-site search. In this two-part article, we will examine the first 4 questions retailers should ask when assessing their own search performance. (Stand by for Part 2 coming up next week!)
1) Can my search handle common mis-spellings and synonyms?
Check that on-site search can handle things such as typos and misspellings, slang and language variants. Nothing is more infuriating for a customer than typing a product name into the on-site search that they know is sold on the site, only for the site to fail to find the product, or even to suggest that the customer has ‘made a mistake’!
Regular analysis of search term data (which can be done via Google Analytics or a reporting dashboard if you’re using a third party solution) should provide valuable insights into the most commonly misspelled search terms, as well as instances where a product is listed in the store catalog using one term, but customers regularly search for it using a completely different term.
This example from Oneills.com shows how their search solution (Klevu) helps them to automatically deal with errors. Customers who use on-site search are fare more likely to convert than those who simply browse for what they want.
Different words for the same item
A classic example might be a fashion retailer who lists all of their sweaters as sweaters, ignoring the fact that customers often search for jumpers or pullovers. A smart search system that uses natural language processing can understand that sweater, jumper and pullover are all synonyms, but a standard keyword-based solution is unlikely to match the queries.
A store selling calendars would need to ensure that things like ‘calender’ is recognised as a mis-spelling of calendar and return the correct results, without implying that the customer has made a mistake.
Another way to reduce the instances of mis-spellings is to provide suggestions around products and categories. This can avoid the customer having to type the entire search term and can therefore reduce errors, whilst also speeding up the process. It’s vital here that auto-suggest actually improves the user experience, so it must be regularly monitored for performance and accuracy. An auto-suggest tool that slows down the customer journey is a fast way to losing a potential sale.
The Redirect Option
Another option for resolving issues like this is to redirect some queries to a product list page in place of the search results. So for example, if someone types calandar, you could just apply a redirect for the destination URL to go to the main category landing page. Lots of larger merchants apply these redirects for the more generic queries, as top-level category pages are better merchandised than their search results page. Nothing is more infuriating than a site to failing to find a product sold on the site, or even suggesting the customer has ‘made a mistake’ :CLICK TO TWEET
2) Can my search handle all types of queries?
As ecommerce gets more sophisticated, so too do customer expectations. More and more customers use on-site search facilities to find answers to their questions, as well as searching for specific products. A review of site search data on many stores will reveal phrases like ‘opening hours’, ‘returns’, ‘delivery costs’ and ‘store locator’. It’s clear in these examples what the customer expects to see, but all too often, stores either present a zero results page, or, perhaps even worse, a page containing random products, which have somehow matched the search terms, via a keyword-driven search approach.
A great example of how to get this type of query right is Glitzy Secrets, as can be seen below. Type in ‘delivery’ and there are various pages suggested that provide delivery information.
Type ‘returns’ and a similar set of results is presented, giving all the information required. A similar search for ‘delivery’ on the animal.co.uk website, however, doesn’t produce quite such a helpful response. A ‘no results found’ response is generated, despite the fact that the site does have a dedicated content page for delivery information, that could have been shown.
Whilst UX issues like these might seem trivial, they do have a cumulative effect, which means it’s vital to do everything possible to tackle all identified weaknesses such as these.
3) Which and how many queries are returning zero results?
Probably the worst experience that on-site search can deliver is a ‘zero results’ page. Results pages that present irrelevant products offer a poor experience, but zero results is much worse.
The zero results page is usually interpreted by the customer as meaning ‘We have no relevant products’, and is closely connected to high bounce rates.
By analysing search query data, it’s likely that some common zero results queries could be fixed easily. Zero results pages are more likely to occur when the search platform is using a traditional keyword-driven approach to results, rather than using natural language processing.
As an example, if a user types ‘strapped dresses’ into a search box, the old-fashioned approach could easily result in a zero result page, as they’re purely matching products based on the name.
Search by Attribute
Another example where a zero result set could occur is if a customer on a fashion site typed in a colour, hoping to see all products in that colour.
Again, if the colour of the store’s products is held as a product attribute rather than being explicitly stated in the product name or description, then there is a risk of no results being shown, despite the store having many products available in that colour. A search system that can utilise product attributes, such as colour, size, manufacturer etc. will always deliver a better experience and more relevant results.
4) Are more generic queries better directed to category pages?
As on-site search becomes ever more sophisticated in terms of the technology and logic used to retrieve results, so too is the format of results developing. eCommerce is seeing a shift away from simple product results, to a more balanced set of results that includes popular related searches, blog posts / guides, CMS e-commerce pages and related category links.
This particularly helps customers who are at an early stage in their purchasing journey, and are using very generic search terms, such as ‘cameras’, for example. Many retailers are choosing to redirect the results page directly to the matching category page for search terms that are as broad as this. From a user experience perspective, this is better than seeing a huge list of search results, unrefined and presented in a seemingly random order.
By analysing search term data, retailers can ensure that their search function is tuned to direct such generic queries straight to the relevant category. On-site search is more sophisticated in terms of the technology & logic used to retrieve results, so is the format of results.
The Importance of Product Name
Also, a retailer that doesn’t adopt this technique risks losing sales, since, for example, a search for ‘mens shoes’ could easily produce inaccurate results if the men’s shoes that they stock do not contain the word ‘mens’ within their product name or description. If you don’t provide filters for users to refine results as well, this could be a more complex journey than using the category page or a product list page. Further reading:
- Definitive guide to ecommerce search - http://www.klevu.com/the-definitive-guide-to-ecommerce-site-search.html by Klevu
- Best practice for ecommerce search - https://econsultancy.com/blog/66658-24-best-practice-tips-for-ecommerce-site-search/ by Econsultancy
- The state of ecommerce site search - https://www.smashingmagazine.com/2014/08/the-current-state-of-e-commerce-search/ by Smashing Magazine
- 50 sites with excellent search - http://baymard.com/ecommerce-search/benchmark/site-reviews by Baymard
- Some really good general tips for more advanced search - https://www.pinpointdesigns.co.uk/optimising-magentos-search-function/ by Pinpoint
Paul Rogers is an ecommerce and digital marketing consultant based in London, UK. Paul has worked in a variety of ecommerce-focused roles, for agencies, merchants and as a consultant. Now concentrating on a smaller set of clients, mostly focusing on strategy and ecommerce technology, Paul also recently started Replatforming.com. Social Links : Twitter.com/paulnrogers Linkedin.com/in/paulnrogers