7 functionality requirements for an enterprise-level site search solution

Every year, eCommerce technologies and platforms become more and more sophisticated, and delivery speeds increase exponentially. This increased sophistication and speed continue to drive up customer expectations, with today’s consumers looking for an effortless shopping experience that really connects with them on an individual level to deliver a personalized, relevant purchasing journey. One area where technology and innovation are making a significant impact in ecommerce UX and conversion improvements is in on-site search. This article takes a look at some of the key requirements for retailers looking for an enterprise-level search solution for their ecommerce stores.

Optimal language processing & self-learning results

One of the most exciting developments for eCommerce on-site search is the number of natural language processing (NLP) search solutions coming to the market. The days of a rigid keyword-driven approach to on-site search look distinctly limited for all but the smallest of online retailers, as the NLP approach gains ground amongst the larger online retailers. By taking a meaning-based approach to search, results become far more accurate.

As a brief example, an NLP search tool could connect the words flower, flowery and floral, to ensure results containing either of those terms were returned during searches for ‘flower print dresses’, for example.

Similarly, a customer looking for a new computer may search for an ‘8GB PC’ or an ‘8 gigabyte PC’. In an ideal world, both of these searches would return the same result set, with NLP algorithms recognizing that ‘8GB’ and ‘8 gigabytes’ are identical in meaning. A quick search at PCWorld however, shows the first search returned 165 results, and the second returned 210 results. Interestingly, both result sets contain quite a number of laptops, memory cards, and other non-relevant items, indicating that this particular platform is perhaps not using NLP to drive results.

With an NLP-based search, results sets should be smaller, but far more accurate at delivering what the customer actually hoped to find. This, in turn, should lead to dramatic improvements in sales conversions.

With a self-learning search system, results will become more effective over time, as the system learns about common mis-spellings and also which results in users responding to them. Klevu automatically optimizes results based on products that users have clicked and products that users have gone on to purchase, which generally makes a big difference to conversion rates. This is a really important feature for larger merchants, particularly those with complex product catalogs.

We’ve found that these are two of the biggest requirements of our larger clients, as they look to improve how certain queries are being handled. A good example came from a furniture retailer we were working with, who had issues with similar queries returning the wrong results. One of the examples was “table lamps” and “lamp tables” returning inaccurate results because the technology was matching products containing “table” and “lamp” in the name, whereas by using Klevu, we were able to match the full queries and serve the right products, which lead to a significant improvement in performance for this kind of terms.

Ability to Merchandise Results

Whether search results are generated via an enterprise-level solution or a standard out-of-the-box option, ecommerce stores can benefit greatly by improving how those search results are merchandised. If search results are presented in a jumbled, incoherent way, conversion rates can suffer, as customers are not able to quickly find what they are looking for. With a search system that allows store owners to weigh individual products, categories, or attributes, the search experience becomes much more positive. It may be as simple as weighing your products based on relevance, your stock levels, or the ones that have the best conversion rates or margin, but this can help to improve profitability and user experience.

Other merchandising techniques that work well for search results include suggesting and promoting links to categories, popular queries, and specific products.

Further Refining of Search Results

Leading on from merchandised results, offering customers the ability to refine their results using faceted navigation is a fundamental requirement for an enterprise-level search solution. Customers may start their search with a very generic term, such as ‘satnav’. Being able to then filter results by brand, screen size, map coverage, and price band all help the customer narrow their selection and make a purchasing decision.

These examples from Reds Gear show how they allow users to filter their results within their pop-out search box and on the results pages.

A number of Klevu customers have benefitted from huge performance improvements with layered navigation within search results – as this is a core issue with Magento.

Ability to Serve Content as well as Products

As consumers get more and more used to sites offering intelligent search, their expectations of what on-site search can offer on an ecommerce store are increasing. Analysis of search terms will reveal that shoppers often enter search queries like ‘delivery costs’, ‘returns’, ‘contact’, ‘store locations’, and ‘promotional code’. A search system that can respond to these queries by directing users straight to the relevant CMS page gives the customer exactly what they want, and therefore increases the chances of a successful purchase. Likewise, a smart search may be able to recognize queries that indicate the customer is still at an early stage in the purchasing journey and may choose to direct a query to a detailed blog post or feature article, in order for the customer to obtain enough information to progress to a ‘ready to buy’ stage.

This is another key feature of Klevu, which can be seen below on the Baby Bunting website.

Lewis from Pinpoint has used this feature on a number of stores and described it as something that has had a really positive impact on his client’s stores.

“We’ve been using Klevu for quite a while now and only realized how important the content search was when one of our clients raised it as an issue. They’d noticed that a lot of customers were searching for delivery costs and delivery information and were just getting product suggestions. When we started serving the CMS pages in results, we saw a big improvement in conversion, as users were able to see that the retailer delivered to their region and what the costs were. This also flagged an issue with how we served delivery information on that site.

We’ve now implemented this feature on most of the stores we work with.”

Auto Suggest / Complete

Auto-suggest has been a feature of many ecommerce platforms for some time now, but there are still improvements coming through that are helping to drive conversions and improve the user experience.

Auto-suggest search tools start searching as soon as you enter the first few characters of the search term, and display progressively more accurate results as you continue to type. This functionality can help focus the customer’s attention on a few key results, whilst they are still actively engaged with the search process. Including thumbnails of the auto-suggest results also helps customers identify well-matched items as quickly as possible.

When using auto-suggest, it is vital that results are highly accurate, as otherwise, it could cause confusion or frustration. Using auto-suggest alongside an NLP-based search tool should lessen the potential for this type of poor experience since results should be both relevant and optimized for the individual.

We generally suggest using our pop-out search box alongside auto-complete, which should speed up the overall searching process, as can be seen below.

We’ve seen great results with this and it’s one of the biggest features for clients.

Detailed Search Reporting

Analytics should be at the heart of every decision on eCommerce technology and data management. Search is no exception to this and on-site search should be configured to produce reports on lots of key areas, such as high and low-performing search terms, search bounce rates, zero results pages, search-to-cart ratios, and more. A solution like Klevu gives you access to a comprehensive reporting dashboard that provides lots of insight into query-level and transaction-level data.

By studying how customers are using your ecommerce store, you can fine-tune your search system, spot gaps in product coverage, correct product data and design experiments to improve conversion rates. You can also use this data to weight products, categories, and attributes to ensure that key products are being served to users.

Lightning Fast Results

All of the innovation appearing within on-site search typically comes at a price – performance. Auto-suggest, in particular, can often slow down a site, due to the resource-intensive nature of the constantly refined results. The same principle applies to a number of the other enterprise-level features since the site is having to do ‘more work’ to produce its results than one using a traditional keyword-based search function.

Clearly, a balance needs to be made between creating a better search experience and maintaining acceptable response times. Fortunately, at the enterprise-level end of the market, scalability has been central to the development of more sophisticated search solutions and solutions like Klevu can easily handle high-traffic sites with extensive product catalogs, which serve lightning-fast results. This applies to filtering as well, which can represent a real performance issue with lots of platforms, especially Magento.

Conclusion

It’s clear that demand for intelligent on-site search is definitely growing in the online retail world, as firms of all sizes wake up to the potential of more customer experience-led search technology. In an increasingly competitive ecommerce world, search represents one of the few remaining low-hanging fruits, where there are major gains to be had, in terms of conversion rates, user experience, and competitor advantage.

By considering the key requirements identified in this article, larger retail businesses should be able to assess potential solutions clearly and effectively.

—-
Paul Rogers runs the UK side of Klevu and has been working in and around ecommerce for around 9 years. Paul has worked for systems integrators, in-house and as a consultant, and has been mostly focused on Magento for the last 5 years. You can find out more about Paul on his personal website.