TechnicalReplatforming ManagementPlatform SelectionProject DeliveryRequirements GatheringRFP Management
Online retailers often underestimate the power of internal search, but tapping into the data it reveals can really help with optimisation, personalisation and even buying and sourcing decisions.
One of the most under-appreciated aspects of an e-commerce website is the search function. It’s not something that attracts the pages and pages written about how Google works, or even how to optimise landing pages or the checkout. However, between 30 and 60% of site visitors will use the website search box rather than navigation (a number that is higher on mobile), and those that do use the on-site search show higher purchase intent and convert at a higher rate. Once you separate out visitors who access the site through Google Shopping and other lower funnel channels and just focus on entrances through the home page, you have a function that a lot of customers interact with and that creates a lot of value for your business.
So, if you don’t know the share of customers using search, the per visit value differential between search and non-search visits, then the good news is that these are all easily accessible through Google analytics. From there, optimisation follows a fairly conventional set of steps that you can use to improve your customers’ experience and increase the value you get from your website. If you don’t have tracking set up in Google Analytics, it’s straightforward and requires no extra development - just follow the instructions set out here. If you have predictive search enabled - a function that suggests products and categories as you type - then be sure to track this by implementing event tracking that captures both the search and the choice. Because this allows customers to make a choice about where they are going, this type of functionality does not require much in the way of optimisation, providing it’s been set up correctly (so remember to track performance and test it periodically to ensure proper function).
Internal search occupies a similar step in the website customer journey to product listings pages - indeed, most e-commerce platforms will generate them in similar ways, using product data to create matches to user or fixed queries. As such, you can use a similar set of metrics.
Just as with listings pages, the customer journey step is for customers to move to a product description page. The inverse of this is exit rate - customers view the search returns and leave the website. You need to look at this not just in the aggregate, but have a view of what the best and worst-performing searches are.
A search that returns zero products is called a Null Return. This may or may not be important - if a search is completely irrelevant to what you sell, then it doesn’t really matter. However, this is a relatively unlikely occurrence - generally, customers search for something that they expect to find from you, so tracking and reducing these Null Returns is an important part of the optimisation process .
Search depth tracks how many searches are made. As with any part of the customer journey, you want to enable customers to find what they want as quickly as possible. If customers have to refine or change their searches a number of times, this is an indication that the product listings you are returning are not relevant enough.
Per visit value for search users
Providing you have enhanced e-commerce set up in Google Analytics, one of the metrics offered is per visit value. Tracking this value against those visitors who don’t use search gives you an idea of how effective your optimisation work is being.
The tools that you have to optimise site search differ from platform to platform. However, the basics are pretty much shared and form some actions that will move your metrics in the right direction. In order to create a priority list for your actions, creating a list of search terms ordered by worst to last by your chosen metric allows you to categorise searches by issue type. This takes a little while and is quite labour-intensive, but will give you a really good insight into the issues that your customers are facing. Layer onto that the customer testing that you are doing and you should have a pretty good idea of what you need to tackle.
Most platforms will allow you to put in redirects from a search page to a category page. This can sometimes be useful when your categories have different names that customers use - you can only use one, the most common, but your website may not be able to deal with the nomenclature. In cases like this, you can quickly reduce exit rates.
Remerchandising and re-ordering
As with listings pages, most websites will allow you to change the way that products are listed away from the default. It’s worth experimenting with this on high exit rate pages to see if you can find a presentation that is more effective for customers. Failing this, you can create entire new merchandising for high volume pages, in a similar way that you might create landing pages for PPC. This can be particularly useful for searches for brands that you do not carry; customers are searching for these brands because they want to find products that these brands create, and you have an opportunity to showcase alternatives that you do.
Introducing context into search returns
Different searches can mean different things in different verticals of your website. For example, if you run a sports retail website, a customer searching for ‘trainers’ is trying to find shoes in the running section and turbo trainers in the cycling area. If your search function is context-aware, you can modify the results to provide better product listings to your customers.
As with listings pages, site search results are ideal for integrating personalisation and machine learning tools. Increasingly, e-commerce platforms are shipping with these already integrated, but there are some very good ones on the market such as Klevu, Instant Search and many others - your choice will depend on budget, platform and capability. In any case, this is something that you can draw a fairly direct line between investment and uplift, so making a ROI-based business case should be fairly straightforward.
These tools take the points about context and ordering and apply live customer data to the returns that they generate. These tools are not just plug and play - you will still need to take time to seed data into the tool and to help it to optimise along the way, but used correctly, your chosen tool can drive value, and quickly. The key to either approach, however, is to have someone who is responsible for doing this, otherwise it can drift and be left as something that works just well enough not to be regarded as a noticeable pain point.
The great thing about the data you collect from site search is that it is your customers telling you what they are looking for. This is hugely valuable, not just for you as leader of the e-commerce team, but for other parts of the business too.
The first tenet of marketing is to listen to what your customers are saying. Producing a weekly site search report gives useful data on what language customers use for the creative teams and those responsible for your search media spend. These search terms are a great resource for generating additional keywords for bidding or for organic optimisation.
Buying and Merchandising
Knowing what customers are searching for - especially sizing, color and other variants - is tremendously useful for people who have to order and reorder products. Brands generally produce a limited amount of stock, so knowing where the demand lies gives your buying and merchandising team a competitive advantage over other companies that are not using search as business intelligence. This, of course, can be combined with data from Google, especially from Google Shopping and the range report that is now offered as part of the merchant center interface.
In conclusion - site search may not be sexy, but you will benefit tremendously by applying some focus and investment to it, all the more so because many of your competitors will not. It’s long been Amazon’s most underrated customer feature and the one that is more easily replicable, so take a look at your numbers today.