Site search has been one of the most criticised aspects of Magento in recent years, across both the Community Edition and Enterprise Edition (which does have a Solr-based search). Although Magento recently announced that they’re moving to ElasticSearch as part of Magento2 Enterprise, search remains an area that can be drastically improved with the platform.
That said, Magento are certainly not the only platform with shortcomings in this area, as most of their competitors (both EE and CE) have fairly average search functions, paving the way for specialist third party solutions. Lots of enterprise-level merchants tend to rely on services like Klevu, SLI Systems and Algolia, regardless of the platform they’re using. If you look at merchants using IBM Websphere, Demandware, SAP Hybris etc, you’ll find that lots of them are using third party search providers, even though the platforms don’t necessarily want them to (because they consider their search function ample).
We’ve used lots of different third party search solutions over the last few years, but we’ve now settled on Klevu, which has a very strong integration with Magento and provides all of the features you could possibly want around search. We previously used Algolia, which is a great solution, but found that Klevu can deliver better results and is easier to integrate with Magento.
Here are some of the features we wanted from a third party search solution when we were trying to optimise our clients’ Magento stores.
Category merchandising is considered a fundamental part of trading online, however merchandising search results is often overlooked or not even considered. This is something that should really be part of the merchandising strategy and it’s a fairly key feature of a search solution. One of the reasons we actually chose Klevu was because it allows for product, category and attribute boosting, meaning that our clients are able to create rules around these, resulting in less of an overhead.
Key things for merchandising search results is the same as product list pages, ensuring that you give the most popular products more exposure, providing a variety of products (across price, features etc) and you’re promoting products that are favourable from a business perspective (stock and margin).
For us, this is a key consideration in optimising Magento’s OOTB search function, which provides limited merchandising capabilities.
Ability to handle synonyms and common errors
This is another key requirement for search, as users often input queries with errors or use different language to the naming convention, which can often result in irrelevant or ineffective results. The more enterprise-level solutions use natural language processing to understand more about the query and reduce the reliance on matching keywords. A good example of this is from the Zimmermann store, which can be seen below.
This is based on Klevu and you can see from searching for “ankle shoes”, it returns a number of results for sandals, which are classified as shoes via the NLP technology. This will also help to solve issues with spelling mistakes and other types of mistakes within queries.
Ability to apply redirects
Although there are often arguments around whether search results pages are better than product list pages for specific queries, lots of merchants want the ability to redirect specific queries to a static category page. All of the solutions that we listed above were capable of this (as is Magento’s OOTB solution).
With the self-learning solutions (which merchandise automatically based on how users interact with results), as well as solutions with NLP capabilities, you may find that the results pages do convert users better than the actual category page (which is probably something to test / review).
Reporting is another key requirement around search and, really, you want to be able to get data from outside of just Google Analytics. If you’re using Enhanced Ecommerce with Google Analytics, then you’ll be able to get more data around what and where users are clicking etc, but ideally you’d be able to see which products are getting the most clicks, the best click-through rates, the best conversion rates from search etc.
Again, Klevu’s reporting dashboard provides lots of transactional and behavioural data, which is very useful.
One thing we would recommend doing is ensuring that you use Klevu’s query parameter in Google Analytics, to ensure that you get the data from the auto suggestions being used. As you’re not actually going through to the results page if you’re using the auto suggestions or pop out box, it’s more complex to track. All you need to do is add klevusearchterm,q as the parameter in the the admin section of Google Analytics as below.
Auto suggestions is another feature that most merchants use nowadays, but Magento doesn’t allow for with the out of the box function. Most third party solutions are able to provide this and there are also plenty of free and paid modules availabe via Magento Connect that can help to achieve this.
Klevu allows for auto suggestions for content pages, blog posts, categories and products, as well as serving images if required. This is something that our clients have really liked about the solution.
Remember, if you’re using this – you need to make sure this is tracked in Google Analytics (will require you to change the parameter).
Content search was a key requirement for us, as we’ve found that lots of users tend to search for things like “delivery information” or “opening hours”. We use Klevu for this on a number of stores, including some WordPress hybrid stores where CMS pages aren’t handled by Magento, which is great. Again, this is something that we think merchants need and is something to think about if you’re looking at third party options.
Other features that we’ve benefited from
In addition to the features that we wanted to be able to provide to our clients, we’ve also really benefitted from the following:
- Pop our search box – this is really good for getting users through to where they want to be faster, which can deliver really strong improvements to conversion.
- Self-learning search – this is a great feature that can help to improve the quality of results massively, particularly for non-primary queries. This wasn’t something that we needed initially, but is probably the reason we wouldn’t move away from Klevu now.
- Faster results – both Klevu and Algolia are capable of delivering results in miliseconds, which is great from a UX and conversion perspective.
- Faceted search – we’ve found that adding filters to search results has been good for users. Although they’re not used on all queries, for clients with bigger catalogs they can help to make it easier to find something that’s very specific from a more generic query.
Solutions we’d recommend
Although we tend to use Klevu for all Magento projects now, we’ve had really good results from Algolia as well. The solutions that we’ve used that we’d say provide a good level of service and have the features our clients have needed are:
- SLI Systems
We’ve found that Klevu delivers the best conversion rates and search results – but something like Algolia can provide faster results. Algolia is an excellent search engine, but it’s not designed specifically for ecommerce, which is probably where Klevu adds value.
We’ve seen some really positive results from using third party, premium search solutions, which have easily paid for themselves. It’s a no brainer for merchants to invest in this kind of technology, especially with the number of users searching going up and the conversion rate from users who perform a search generally being a lot higher than a store’s average.
We would definitely recommend Klevu, which we consider to the best available, but there are lots of other solutions that can help to achieve your objectives.