Measuring Ecommerce Site Search Relevance: Precision and Recall

Measuring Ecommerce Site Search Relevance: Precision and Recall

There are hundreds of factors that impact a customer’s journey before they purchase your product. From advertising to previous shopping experiences to UX, any one of these things could be the deciding factor in a purchase decision. 

One of the foundations of many successful ecommerce transactions is relevant site search results. But even the concept of relevance can seem daunting and vague. How do you measure the effectiveness of your site’s search engine?

Ecommerce Site Search Performance Metrics

While there are many metrics that ecommerce companies need to measure and track consistently, very few provide detailed feedback about search relevance. 

Enter precision and recall.

Site Search Precision

Precision is how many useful search returns are delivered versus the number that are not relevant to the customer. In other words, how many of the results are true positives vs. false positives? 

For example, looking at a search for “black boots” on an e-commerce site, if the results show ten different products, six of which are black boots, two of which are brown boots, one of which is black shoes, and one of which is brown socks, then the precision is 6 out of 10 or 60%.

One pair of boots that doesn't belong is an issue of search precision

Site Search Recall

Recall is how many of the total number of relevant options on your website are returned by a search. This is equivalent to the idea of false negatives—how many items might be relevant but aren’t surfaced in the search results?

For example, If there are ten products relevant to the “black boots” search query, but the system only returns eight, then the recall is 80%.

If there are black boots missing, that's an issue of search recall

Why Site Search Precision and Recall Matter

At a small scale, having some precision and recall “noise” in the search results may not seem so bad. However, when you scale this up to hundreds of results, it could mean dozens or even hundreds of irrelevant or unlisted results that your customers have to wade through to find what they are actually looking for. This can cost you sales by burying the items your customers are actually looking for under layers of noise.

To get a general metric for precision and recall, consider manually running 50-100 searches on your site and measuring the values based on the results you get.

Improving Your Ecommerce Site Search Relevance

There are a few ways to improve precision. The easiest is to remove fields that contain a lot of “noise” from being indexed by your search engine. For instance, in the example above, brown socks might be showing up in a search for “black boots” because their description might include something like, “These pair well with many kinds of shoes, from white flats to black boots.” Removing the “description” field from your search engine will prevent this problem from occurring.

Often, however, description fields contain many useful and relevant keywords for an item. In that case, it can be useful to create a separate field that includes the relevant keywords from the description field without the irrelevant terms. This process usually has to be done manually, however, which is quite time-consuming. 

To improve recall, you can take the opposite approach: add more keywords to your search fields. One simple way to do this is to find synonym lists for common keywords and add those to your search engine so that, for instance, the word “shoe” is added to any item containing the word “sneaker.”

As you can see, improving precision often hurts recall, and vice versa. For this reason, it’s important to try to make improvements incrementally, letting you review the impact of each decision and giving you the insights to make better decisions in the future.

How Machine Learning Improves Site Search Automatically

Rather than manually optimizing hundreds or thousands of products to ensure your site delivers the right search results, machine learning offers a way for your ecommerce search engine to optimize itself automatically over time based on the actions your site visitors take. 

A machine learning-based ecommerce search engine can automatically optimize and rerank search results to provide results that are more relevant and attractive to each individual visitor, and more likely to convert into sales.

Bringing Natural Language Processing to Ecommerce Search

One of the benefits of using a modern ecommerce search and product discovery platform is natural language processing (NLP). NLP is a powerful tool that lets your search engine understand language like you and I do: dynamically and with nuance. NLP-powered search examines queries and uses AI to deliver relevant results by breaking down and understanding the meaning and context behind words.

Going Beyond Relevance in Ecommerce Search

Providing relevant search results is important! But it is foundational. To create a truly dynamic and impactful shopping experience, you need to deliver results that are both relevant and attractive to your audience as whole, while making personalized adjustments for each unique visitor. 

Attractive results are ones that are most likely to lead to a sale, rather than just the results that match the words in a search query. To learn more about how the best search engines optimize for relevance and attractiveness, check out this video.

Constructor is an AI-based product search and discovery platform that provides personalized results to each visitor that are relevant and attractive, leading to higher RPV. See how it works.