E-commerce Search Evaluation: 5 Tests to tell if you are geared for driving conversion

E-commerce Search Evaluation: 5 Tests to tell if you are geared for driving conversion

We speak a lot about how your store's Search should be your top salesman. Everyone can give a product as a result if the user types the product's name perfectly. If only the users were that considerate!

We present a simple 5 point test to evaluate how good of a job your top salesman is doing.

Test 1: Keyword Dependency

Try non keyword, descriptive search queries. Do they work well?

If you are a fashion store, try "dress for party". If you are a toy store, try "educational toys for toddler". You get the gist.

Most Search systems rely on users typing the right keywords. In the world of GPTs, users expect you to understand what they want. They want to tell you about their need, not the exact product to solve the need. Only if you can give users results good enough to satisfy the need, can you expect to make a sale.

Relevant results will thus drive transactions, and keep the users coming back, because they'll feel that you understand what they need.

Test 2: Handling No Results

Often users will type in something that you don't have in your product catalog. Do you simply show a No Results page all the time?

There are times when you must show no results, such as when a user searches for Honda Civic Brake Pads at a grocery store. However, in most cases, there is a better way to handle this. Most users will search for something at least related to what you sell. For example, they might search for sweatshirts at a store that specializes in suits, or for sofas at a recliner store. In these cases, you can show products that meet the same or a similar need in a different way.

It will not always translate into a sale, but every now and then, it will. And the user will keep that in mind the next time they need the product.

Test 3: Cross Sell

A user searches for Coke. What do you show them?

Good answer: Coke

Better Answer: Coke, and things they are likely to buy with Coke

Yes, that's it.

Test 4: Analytics, Insights and Actions

Do you track how well each search query is performing? Obsess over it?

Your top queries by volume? Top converting queries? Least converting queries? Your missed opportunities - large volume with low conversions?

Unless you are looking at this data on a regular basis and making sure everything on the dashboard is always improving, your are missing out on potential revenue and users.

This is hard, takes time and effort, and is expensive. It takes people from product and business to monitor the dashboards, and the solution may lie with data-ops, engineering or search teams. A cross team coordinated effort for everything that can be improved.

Test 5: Optimizing based on user behavior

You showed the results. The user may or may not click on them. They may or may not buy the product.

This data tells you more about the quality of results than any other metric your team may have.

Using this data to improve the results is the holy grail of search optimization. This primarily works at two levels:

  1. Using user behavior to improve results across the system. Eg: Most users searching for Juice, go for Orange juice, so it should rank higher for the query
  2. Using user behavior to personalize the results for that user. Eg: A user may always prefer dark colored shirts, or books for toddlers. Using this to rank the results better makes for a delightful user experience, helping you improve retention and conversion.

Conclusion

Getting Search up and running is no big deal today. There all kinds of plugins, apps and SaaS products out there.

Making sure it does its job is a whole other query! (Sorry!).

How much did you score?

We would love to know at hello@neuralens.ai, and work with you on getting a perfect 5/5.