Like most e-commerce platforms on the market, Shopify Plus has a relatively basic native search capability, however, most of the better-known third-party search providers now have integrations available.
This article is written by Claudia Ditri from Klevu, which is a Finnish search technology startup with market-leading NLP and machine learning capabilities, amongst other things. Claudia will be talking through the key components to look for in a third-party search solution when working with Shopify Plus.
Over the last 18 months, Shopify Plus has undoubtedly grown, with more and more large, complex merchants moving over to the platform. As more enterprise-level retailers adopt the technology, the requirements and levels of customisation become more complex and one of the common areas is search, which we’re focusing on in this article.
Here are what I consider to be the key features to look for in a third party solution.
Strong native error tolerance
One feature that often drives retailers to look at third parties in the first place is error tolerance, as native search options will generally struggle in this area. An out-of-the-box search solution will likely look to match products based on a query exactly matching the title, which won’t allow for errors. The MVMT example below shows how the native search they use is unable to process results for a mis-spelling of “sunglasses”.
The O’Neills example below is a better example of how a more complex error still provides relevant, accurate results.
This may seem like an obvious requirement, but for smaller stores, it’s generally one of the more important feature requirements. Another consideration here is handling of 0-result search, which is also important.
Understanding shoppers naturally
A shopping experience happens through naturally-formed interactions either between humans or, in the case of online retail, between a human and a system. The search box on the shop is the most important element where shoppers actually express their desire on what they want to see and purchase. If these desires can be understood, the online shop will be able to serve relevant results, leading not only to more purchases but also a more positive customer experience.
Extracting relevant data from the shopper’s query in search requires an understanding of both the catalogue and the query itself. In technical terms, this is done through Natural Language Processing (NLP) techniques. Klevu processes the catalogue and search query to bring the most relevant results to users. The ability to process natural language is something that isn’t often needed on smaller stores, however, on one of our sample studies at Klevu, we found that over 34% queries require some form of natural language processing, resulting in a direct benefit of increasing the search to click conversion rate.
Ability to boost products based on tags and meta field values
The ability to merchandise results for individual keywords and groups of keywords is one of the core requirements for a third party search solution, particularly those with larger catalogues. However, with Shopify Plus, this often needs to span across categories too, as touched on below.
A third party search solution for Shopify Plus should be able to support boosting specific products and groups of products based on rules from both tags and meta fields. Most solutions on the market support tags, however meta fields is also important for a lot of merchants.
A search overlay can help speed up the search process considerably - serving different types of results such as products, blog content, guides, category listings etc without having to wait for the search results page to load. Here are a few examples of Klevu implementations that have customised our native overlay.
This example is from Made.com, which provides product and query suggestions.
The Baby Bunting quick search overlay features content results, category listings, query suggestions and products.
When working with a lot of platforms, a third party search solution can require a lot of integration work to get up and running, although this has improved considerably for most of the mainstream platforms. Integrations with Shopify Plus are generally fairly straightforward, as a result of the simple nature of the platform and the availability of the required data, which is what most merchants seem to be looking for.
In some instances, there will be a need for more complex integrations, but this is generally representative for a minority of merchants.
Klevu, like many of the other providers, has a plug-and-play integration available via the app store, which can extract the product catalogue and any other required information directly.
Advanced reporting/search insights
Understanding how customers are using search is really important, as it can provide insight into core issues with your merchandising, your product range and your navigation. There are two ways of capturing core search data; via your search provider and via your web analytics platform - we’d generally suggest using a mixture of both.
The above screenshot shows the generic search overview from Google Analytics, which is a good starting point for assessing the value provided by search. In this screenshot, you can see that search represents a good opportunity for optimisation, as the CR% is significantly higher via search-led sales, so encouraging more users to search could represent a good test. Beyond this report, you can also drill down into key areas such as device performance, pages where users are performing a search, the keywords performing well and badly etc.
For the more specific information around product performance and actual sales, we’d generally recommend that the merchant uses the search dashboard provided by the third party, which will generally provide more detail around the transactions taking place etc.
A layer of automated optimisation
Self-learning technology has become a really important aspect of e-commerce in recent years, adding value across several areas of e-commerce merchandising. Search is one of these areas, alongside things like product recommendations, category merchandising etc.
Klevu applies a layer of machine learning to optimise results based on the behaviour of users and how they interact with results - generally based on clicks, add-to-carts and completed purchases. This adds a second layer of input (alongside the underlying business logic provided by the e-commerce team) which helps to keep search results relevant and reduce the manual overhead.
There are various other examples of how machine learning is used across the different search and merchandising providers in the industry.
Beyond search: category navigation
Although this sits outside of search, a common requirement from Shopify Plus merchants is for a search solution to power the product grid and product filtering, due to limitations with Shopify’s native filtering capabilities.
Shopify’s out-of-the-box filtering is very basic, so merchants commonly look to a search provider to allow for things like multi-select and multi-faceted layered navigation, AJAX filtering, combined merchandising of categories and search etc.
I’d imagine over time Shopify will improve this side of things, but currently this a really common requirement for retailers.
Some of the other requirements here include:
- Ability to easily manage which filtering options are displayed
- Ability to serve some form of HTML snapshot for search engines (so the merchant isn’t relying on them crawling the JS)
- Ability to easily apply boosting rules for the different categories
- Control over the way the grid is displayed and the information being used for things like product labels
These are just a few very top-level requirements, there are hundreds of others as you get into more detail. If you would like to find out more about Klevu, you can do see here on their website.