How Warely Proprietary E-Commerce Search Engine drives conversions at Smoke Cartel — A Case Study

In the competitive e-commerce space, more and more savvy merchants are realizing the power of optimizing in an environment where very fraction of a percentage change in metrics like conversion rate and bounce rate are yielding the difference between thousands to hundreds of thousands of dollars towards the bottom line at the end of the year.

We’ve helped our merchants realize the compounding effect of optimizing for each and every interaction & microinteraction within an e-commerce site can make a big difference in your revenue. One crucial part of every e-commerce site — search.

There are a number of search solutions out there. In fact, most e-commerce platforms like Shopify, Woocommerce, Magento, Big Commerce, etc all come prepackaged with a basic search engine. Why is this so important? And what real difference does it make?

Here’s some numbers to think about:

• An average of 30% of users will use search on an e-commerce site
• Users whom use search can often convert 50% higher than average

Of course, if you think about it, this makes perfect sense. A customer that is looking for something specific will often search for it, and if they find what they are looking for in the results, will often checkout more frequently than the folks that are just browsing.

So if search makes up almost a third of your potential customers, it makes a lot of sense to try and optimize for them knowing every improvement yields a greater chance at converting an already high intent customer.

Our client, Smoke Cartel, is the industry leader in cannabis accessories such as bongs, vaporizers, and more. They are hosted on Shopify (our personal favorite platform) and have tried numerous search options available on the app store.

They needed a solution that could do the following:

• Accommodate products in an industry where the lingo is still developing and changing
• Optimize for many new users whom are previous unfamiliar with anything in this product demographic
• Provide accurate and smart product results
• Be lightweight, lightning quick, and not impact their site load times
• Be SEO friendly and have a linkable structure
• Display a complicated and customized product structure that relies on metadata and product tags
• Rank results based not only based on accuracy, but optimized for business logic and e-commerce goals

Pretty tall order.

We discovered that most out of the box Shopify search app solutions lacked the ability to do all of these requirements, or more commonly — lacked the flexibility and customization needed.

Enter, Warely Search.

We deployed a product that fit all of their needs, and more important — was built specifically for e-commerce search. Many common search apps are adaptations or extensions of existing search providers, who’s roots and core functionality was built around text content based sites like news and information. We built a search engine with only one thing in mind — e-commerce.

What’s the magic?

• Lightning quick (30–50 milliseconds response times) queries across an optimized index for products
• Machine learning interaction tracking that continuously optimizes the results both across the network as well as individually for the customer
• Intelligent business logic that moves out of stock products to the end of the results, and boosts visibility of products with high inventory counts and are trending

At this point, we felt like we had already delivered a solution that was cutting edge and awesome for our e-commerce clients.

We didn’t stop there.

We did some more analysis. The ah-ha moment — content is e-commerce is visual.

First, we taught our machine learning about visual correlations between products.

Visual correlations between “fab egg” style products

Visually similar product group

So let’s say you search for a product on the client’s site. Pure synonym and intelligent contextual search would yield 2 results. Boring.

Then, we use a secret sauce of machine learning to make recommendations based on the original two results — products that customers also viewed, products that are visually similar, etc.

Horray! A much richer result set.
For example, click here to see a search that contains related products.

Then we asked again, how can this be improved even more?

We were absolutely dead set on producing the best e-commerce search ever. So we had to keep going. Some more analysis, and some brainstorming, we added:

• Collections are a better subset of products hand crafted and carefully tuned by the merchant, why not defer to that for generalized product category queries? Great — auto redirect to collections on specific queries, and suggest collections whenever appropriate and relevant.
• Users may search for a query often found down to the variant or SKU level, so we reconfigured the results to display the most relevant product variant image to the search. For example, a generalized search for “pipe” might show the default image, but a search for “green pipe” would yield the more relevant variant image within the product set.
• If no results are found, don’t give up — show the user recommendations based off their behavior on the site so far.
• Finally, search within other pieces of content as well, but display extremely relevant results at the bottom in case the user is interested — like articles, pages, etc.

And so, Warely Search was born.

We’re pretty happy with version one, and Smoke Cartel is too. In fact, it had a significant boost on their improving their conversion rates within search. At the end of the day — that means we delivered more dollars to that client’s bottom line.

Warely Search In Action at Smoke Cartel — The #1 Online Headshop

To find out how Warely can help your business succeed, and to hear more about our full service offerings from cutting edge development to marketing; get a quote at Try.Warely.io