The Quest for the Holy Grail: “eCommerce Search Happiness”

Is it possible to achieve Search Happiness?

As with everything in life, well ok almost everything, when you have something big to solve, the best solution is to follow the path of the partitioning algorithm. Don’t worry, I’m not going to deep dive into the world of mathematics ;). This is the path we’ve taken to achieve success in our quest for search happiness, and over the course of three articles we’re going to help you understand this path and how to achieve your own search happiness.

So let’s begin with looking at the search journey…..

Part 1 Conversion is a false friend and not the ultimate goal

Measuring the success of site search solely on sales performance is a common error in the industry. The search box is increasingly being used by online shoppers to not only find an already known product, demonstrating a high purchase intent, but more and more frequently as a different way to browse and discover new products. That’s why measuring search performance purely on conversion indicators is a misleading approach.

First of all, let’s look at a single search visit. It’s fascinating to see how search journeys evolve from a single term to go off in so many different directions.

The following diagram represents a real, and very simple case, where a user starts typing dress, then moves to stripped dress and finally ends up buying a bikini!

Can we conclude this search visit was successful because it generated a sale? Was the user perhaps disappointed because she couldn’t find the dress she was initially looking for? Is a search visit with 100% CTR an indication of success? Should conversion be the unique metric to measure the success of a search visit?

In order to understand and answer these questions, let’s think about the role of search across an entire customer journey. Hopefully it will help us establish some of the different use cases for which the search feature can be used within an ecommerce.

Let’s meet Eva Thomas, who’s 36 years old and from London. Eva wants to find an outfit for a gallery opening next week, here’s an outline of her search activity using her favourite fashion brand site:

Meet Eva Thomas: Browsing her favourite site while going to work

She makes a first query, “dresses”, then scrolls and refines the results.

This is a Browse/Discovery type usage.

Eva tries a second query, ¨silk dresses”and performs a series of actions with the interface, including a few clicks, scrolls and applying a colour facet filter (pink) followed by a colour selection on black.

Looking up and glancing around, Eva spots a woman in front of her wearing a very flattering denim jumpsuit and it inspires her to search for something similar.

While she keeps searching, new clothing matching her criteria appear and she keeps interacting and refining her search results using filters and facets until she finds something that she really likes.

After a few additional queries (jumpsuits, playsuits) she finds the item she was looking for.

As a last step she checks her size and heads to the shop suggested in the last query where she can buy the product.

Eva Thomas: Heading to the shop where she located the item

Eva arrives at the shop to buy the item she found online.

Unfortunately when she arrives they’ve just sold the last jumpsuit in her size.

The shop assistant gives her the SKU number so she goes back onto the site and searches using the item code, using part of the number and then in it’s entirety.

As no results are displayed for the product number, Eva searches again for the product and finally finds it again.

While the store didn’t have the size of the jumpsuit she wanted, they did have it in a different colour so Eva was able to try it on for size. She then decided to go online to purchase it but finding the same product again was so difficult!

She couldn’t get it using the reference number and it took her ages to find it by typing in text. Finally, after browsing through SO many products she got hold of it and bought it. The search feature this time was lousy!

So what can we conclude from these examples?

We can clearly see that a good search experience doesn’t necessarily lead to a purchase, it might just not be the right time. In Eva’s case she wanted to go into the shop to try on the item. Yet, we can also see that a sale doesn’t necessarily stem from a good search; not finding the item in store Eva needed to make the purchase online despite her frustration at the experience.

This shows us that people must be treated as subjects rather than objects and that means we need to understand and present search as a human experience.

Search success depends on the user intent.

Search intent changes not only for each search visit, but also within a single search visit (i.e. from the discovery of products, to finding a size guide or a store).

When a user is in the Inspiration Phase of the customer journey, they value the experience and serendipity, that is, “discoverability”; the ability of search to surprise, to uncover and display relevant products while having fun searching.

When a user is in the Purchase Phase of the customer journey, they value speed and precision, that is, “findability”; the ability of search to quickly find an already known product with the minimum effort.

Search performance must be measured from a findability and discoverability perspective, that means addressing precision and conversion, and also serendipity and experience.

Search relevancy is therefore clearly not correlated to conversion.

In the next post we’ll further elaborate on the user intent in search, as a prelude for the introduction of a framework for measuring search happiness.