Enhancing filter-based product search, step by step

Share

There are many documented issues with product search filters, as we discussed in an earlier post. The outdated and impersonal way product search operates hasn’t changed from its original design many years ago, leading to frustrated consumers, abandoned carts and a lack of loyalty from buyers. Two years ago we started the Traverz project with a mission to modernise the existing search paradigm – to help business make a more human connection with consumers and to solve the frustration of unsuccessful product search. The resulting ‘Search-by-Traverz’ concept is a preference & agent-based approach, that boldly ditches the outdated concept of ‘product filters’ – after 20 years of trying to mask the issues it is time to move on.

As part of our journey to seek out more modern and personalised product search mechanisms, we have gradually transitioned away from the filter approach in a series of logical steps. Each of these steps (for example personalisation, in-place feedback, the agent – as we will explain below) introduces a material improvement in the user experience, whilst initially building on top of existing UI/UX and search structures. There is an understandable reluctance to introduce new purchasing experiences to customers, so we’ve created a transition process that allows e-commerce platforms and marketplaces to introduce the Traverz concepts to consumers in a fully phased way – one benefit at a time. This also allows you to test and benchmark different approaches to provide insightful data at every stage of the product search journey.

We’re excited to be able to launch our staged approach to revolutionising the world of product search, creating a stepped introduction to a more human search experience.

A win-win user experience for e-commerce and marketplaces

Let’s start with the traditional “filter bar” product search model, which retains the advantages of familiarity for consumers and allows them to maintain their current shopping behaviours. The search result list (based on user input) continues to be driven by your existing platform’s search engine processes, ensuring that you are able to prioritise specific products and take into account real-time pricing and availability adjustments. On top of this existing product search process we can then begin to layer some key improvements.

Personalisation

The first step to an improved customer experience is to offer a degree of personalisation. Rather than taking a one-size-fits-all approach to the display of product lists, we provide users with product information that is specifically relevant to their search journey. A combination of smart logic and Machine Learning is used to determine which product information helps the user with both the option to confirm that the product meets their needs and is worth exploring further, and highlight differentiating features which helps to trigger recognition of needs.

Personalisation

Feedback everywhere

The next step is to add ‘in-place’ feedback , so customers can add additional context-specific filters from anywhere in the platform without having to go back to the original search bar. The traditional approach of requiring them to go back to the results overview and find where the filter setting is located is a key ‘pain point’ and one of the main reasons why users are frequently seen to give up on trying to use the filter mechanism at all. Releasing them from the shackles of the filter bar enables users to focus on their search journey, and provides significant data on their actual needs.

Feedback everywhere

Agent dialogue

The final step of enhancing the filter approach is to add a pro-active personal assistant / agent. This pivots the search from being a single-shot, one time request, to a multi-step journey, and brings in the power of Natural Language interactions. It also allows the agent to deliver personalised insights at appropriate moments, to add a clear element of intelligence into the customer experience.

Running in the background and driving all this is the Traverz Core, which calculates and injects personalisation, feedback and the agent into the platform via a straightforward front-end Javascript component library. An additional benefit of Traverz Core is that it applies Natural Language Processing and image analysis, smoothing over any data quality limitations in the product catalogue.

Delivering a best in class user experience – ‘Search by Traverz’

These user experience benefits are the result of joining forces between your search & filter system, and Traverz – all delivered via the original search interface.

However the continued use of that original search interface means that during their search journey consumers won’t be able to find products with filter terms that don’t exist in your existing filter bar functionality – for example they will still only be able to filter for a cottage with a garden and not be able to indicate that it should be a ‘rear’ or ‘south facing’ garden.  Basic usability means that existing filter bars are inherently limited in the search terms they can use – too many search options are almost more crippling than too few – but data and quality issues in the underlying product database generally would not allow such broadening of the filter bar scope in any case.

The answer to that limitation is of course to take one more step – of introducing a ‘preference’ concept, which allows users to interact with a much wider set of product features. The earlier mentioned in-place feedback and agent are then able to fulfill an even more important role, by unlocking that wider feature set. Not unimportantly, a side-effect of focusing on preferences rather than filters is that we can pivot to a much fuzzier search concept, involving multiple preference levels and the notion of an importance ranking amongst competing (and potentially conflicting) preferences. More on that in a later post!

In summary, we have outlined a staggered introduction of product search user-experience enhancements, enhancing your existing platform with the additional functionality that Traverz offers step by step. This allows you to move customers through to a more personal and enjoyable product search experience, removing barriers to purchase and allowing them to more easily find the products that match their requirements – reducing cart abandonment and increasing conversion. And this all happens without you having to spend considerable amounts of time refining the product data catalogue.  It also means you can keep complete control of the process of delivering search results, all whilst carrying out A/B testing to understand how the additional functionality impacts your bottom line.  And once you’ve fully moved to ‘Search-by-Traverz’, you gain access to a vastly better level of insight into your customer choices, as well as facilitating a far better shopping experience for consumers.

Want to find out more about how adding personalisation, feedback everywhere and an ‘in search’ agent could drive better search performance, improving consumer loyalty and reducing abandonment?

Get in touch to see Traverz in action in our ‘demo marketplace’.

 

Next: Humanising product search and making it truly personal

Share