The way we use the internet, including on ecommerce stores, is changing rapidly. More and more of us are choosing to shop online using our tablets and smartphones, rather than desktop computers. This, coupled with a gradual moving away from category-based menu systems, is bringing search into the spotlight, as consumers demand a quick and easy way to find exactly what they are looking for when shopping online. This is even more applicable on mobile devices.
As a result, growing numbers of retailers are starting to realise the potential that a strong, feature-rich search solution has for their business, and are exploring ways in which their own search offering can be overhauled to provide a better customer experience. In this article, we look at some of the ways that ecommerce site search can be improved, in order to bring it up to date with the latest developments in search technology and best practice.
1. Implement an NLP-based search tool
Natural language processing (NLP) and machine learning are taking the ecommerce world by storm, shaking up various functions of an online store, including search, product recommendations and merchandising. More advanced, enterprise-level search solutions, like Klevu, use NLP to understand more about the query, in order to match results more accurately.
In search, natural language processing is used to understand more about the query, allowing the technology to answer what are essentially more complex asks. An example of a query that NLP would help with could be “salmon coloured backpack with a front pocket” – in this instance, Klevu would extract the data and use NLP to understand the key variables in the query and match to the terms that are used in catalog.
This context-driven, meaning-based approach of NLP means that search results are finally relevant to the customer’s search phrase. Clearly, the more accurate that search results become, the more likely the customer is to find what they want and actually make a purchase. The benefits go way beyond that initial purchase though, as a happy customer quickly becomes a loyal customer, returning again and again to a site that they feel really understands them as an individual.
Promote the use of your search function
From what we’ve seen with our clients, the use of on-site search has risen in recent years (generally around 10% – 25% of all users, depending on the prominence of the search box and the nature of the store), due in part to the growth in mobile internet usage. Despite this and the reports available in web analytics platforms (which generally show an uplift in search-led user journeys), it’s surprising to see that many online retailers are not positioning their search box more prominently – especially given that many of the market-leading merchants position search as a primary navigation option (eBay, AO.com, John Lewis, Amazon).
A prominent, bold search box that is clearly defined and easy to find could make a considerable difference for many retailers, helping users to find their desired product(s) quicker. Using language that encourages users to search, such as “search by product name, code, category or type” rather than a tiny magnifying glass icon, could also make a big impact. This is important on desktop, but far more so on mobile, as finding products via categories can be laboursome and increase the time to purchase considerably.
Include content search in results
When a visitor uses the search function on an ecommerce site, they could be at any stage in their purchasing journey. Some will be ready to commit to a purchase, others will be at the start of their journey, and could be looking for information about the product or about the store they are visiting. Including content pages in site search results can improve the customer experience for these early-stage customers, by giving them the information they are asking for. A search for ‘delivery’ or ‘returns’ should show the store’s delivery and returns pages, rather than some random products that somehow happen to have a keyword match, or no results at all.
Similarly, showing size guides, detailed specifications, product reviews, blog content and even buying guides could really help convert that information-hungry potential customer. Content search is not common on ecommerce stores currently, but it’s something that is gaining traction, as search tools become richer and more customer-focussed.
Use a good auto-suggest / predictive search
When a customer searches on an ecommerce store, they are generally trying to find something quickly. By adding ‘as-you-type’ product and category suggestions into the store’s search function, you are able to speed up that search dramatically. If the search is powered by an NLP-driven solution, product and category suggestions are likely to be accurate and highly relevant and can serve results that aren’t purely based on the keywords being used.
People inevitably make typing mistakes, or are unfamiliar with the spelling of brand names or products. Auto-suggest can kick in to present likely results after just three or four characters are typed into the search box. This reduces the potential for errors and speeds up access to results, with the end result being that the customer moves closer to a successful purchase transaction.
Implement a rich search interface
Using auto-suggest is just one part of a trend towards speeding up the search experience. Introducing a richer ‘quick results’ interface for search is another way that results can be presented more efficiently and faster to the customer. These panels will typically show thumbnails of the first few results, along with a link to view all results.
However, progressive retailers are also including links to relevant categories, content links, and even faceted search options in their dropdowns. This approach in a lot of cases takes the entire search process into the drop-down panel, removing or reducing the need for the traditional search results page. Redsgear.com, an outdoor gear specialist, has a great example of a rich search dropdown that also features infinite scroll to show all results.
Merchandise your results
Assuming an NLP-driven engine has been adopted to power search results for a site, the next step is to merchandise those results, to drive the maximum volume of sales. Search merchandising is made up of a number of component parts, but the key one for the more advanced merchants is around weighting the results.
A key requirement, especially for merchants with larger product catalogs, is the ability to weight key products, attributes and categories to ensure that the best products for the user and the business are being served. An example of this could be a fashion retailer weighting their top-selling products and also boosting a ‘summer’ attribute when they’re going into the new season, meaning their summer products will be promoted for their chosen queries.
One of the key features of Klevu is its self-learning technology, which adds a layer of boosting based on how users interact with results. As an example, if lots of users are clicking through and purchasing a specific product, this will be displayed higher for the relevant queries. The key drivers for this are purchases, ‘add to carts’ and clicks, which can make a big difference to the relevance and quality of results, particularly for longer-tail queries.
Improve zero results page
For stores using traditional keyword-driven search tools, the zero results page is an all-too-familiar occurrence and, be it far less, it still exists when using the most advanced technologies. Rather than simply stating ‘No results found’ or even suggesting that the customer has somehow made a mistake, a better approach is to try to salvage something from the situation and encourage the user to continue their journey.
We generally recommend that merchants display links to the most popular results and even a product recommendations block.
Analyse search data to improve product listings
We’ve focused so far on design and functional changes that can improve the search experience for online shoppers. One other key opportunity is in the area of search reporting and analytics. By examining site search statistics on a regular basis, it should be possible to make significant, material improvements to a store’s product catalog.
Identifying repeat searches that have a low conversion rate, despite there being an obvious set of products that should be converting for those phrases, could allow retailers to address issues in the product listings for those items. Products may have weak listings that could be improved, links to size guides might be added, or the product in question may have inventory errors that need to be corrected, which are preventing customers from buying those items. Analysing the poor performers in this way should provide trading opportunities for the store, and should also improve the customer experience over the long term, as they find it easier to locate the items they are looking for.
We’ll be doing a follow-up post around understanding the value of search in the coming months.
For ecommerce stores, it can be hard to reach decisions on how and where to invest in third party systems, for maximum ROI. Looking at on-site search, however, could actually prove to be one of the most beneficial strategic decisions that a retailer could make, and could potentially generate substantial long-term improvements, by way of increased conversion rates, order values and repeat transactions, as well as optimised user journeys.
For more information on Klevu, visit www.klevu.com.
For more posts by Paul Rogers, visit his blog www.paulnrogers.com.