By the numbers: the ROI of keyword and AI site search for digital commerce

Did you know that the tiny search bar at the top of many ecommerce sites can offer an outsized return on investment (ROI)? According to Econsultancy, about 30% of ecommerce consumers use on-site search. It can be much higher for marketplace search. Website visitors who perform site searches have been found to convert about twice as much as non-site-search visitors. 

If you’re in the market for new ways to drive higher online revenue, reduce support costs, and boost brand loyalty, this is the blog for you. I’ll share some of the stats for how on-site search can help, and show how AI-powered search does even more. Let’s dig in.

Well-optimized ecommerce site search can have a powerful effect on a company’s success. For example:

  • Algolia customers have seen in excess of 350% ROI with optimized keyword search alone!
  • Just increasing your site search box’s visibility can boost conversions. When one company made their on-site search more visible, goal completions increased by 242%.
  • Even if visitors don’t convert the first time they visit your online store, Moz noted that “People who perform an on-site search are more likely to return to the site with an intent to purchase.”
  • Speed matters, too. You need your search engine to be fast. Page load time affects purchasing decisions for almost 70% of shoppers. Amazon found that a 100-millisecond delay can cost millions of dollars as bounce rates go up.


With the average ecommerce conversion rate hovering around 2–3%, even a small lift in conversion from better on-site search can drive massive revenue improvements. These charts, based on two hypothetical companies with annual revenues of roughly $750 million and €50 million respectively, show how a change of as little as one tenth of one percent in conversion can have a huge effect:

Above and beyond revenue

Site search can also help to improve more than revenue. Better on-site search improves customer support and brand loyalty, and it can improve your ability to build better first-party data for personalization. 

    • A good search engine should deliver more relevant results that are ranked intelligently. Accenture showed that “40 percent of consumers have left a business’s website and made a purchase elsewhere because they were overwhelmed by too many options”.
  • According to another study, people who are fed up with your search are likely to abandon ship and swim over to one of your competitors.
  • Site search is at the heart of providing better customer self-service. Studies have shown that it can lead to a massive reduction in the number of support tickets. This can save your business money and improve your customer satisfaction at the same time.
  • According to Harvard Business Review, a whopping 81% of customers attempt to resolve an issue on their own before contacting customer support. High-quality site search allows your users the option of resolving their problems on their own.

If all the numbers above aren’t convincing enough, consider this: site search also gives digital retailers incredible data from which to make great decisions. Search queries are intent data — they tell you exactly what your customers are looking for and what’s most important to them. With this data, you can make better decisions about what content to create, how to improve the customer experience, and more.

Search data can also be combined with customer data to provide personalized experiences. Ninety-one percent of consumers say they are more likely to shop with brands that include relevant offers and recommendations. Companies using advanced personalization report a $20 return for every $1 spent.

Site search is even better with AI 

Those amazing stats above? They’re only for keyword search. With AI search, the numbers go even higher. Accenture indicated that AI search may someday account for 30% of all revenue generated. Moreover, sites with a semantic-based search engine have a low 2% rate in shopping cart abandonment, compared with as much as 40% on sites with only full- text search.

Just like generative AI such as ChatGPT, AI search uses large language models (LLMs) to better understand user inputs — in this case, searches. Its job is to deliver the best results from your site. Whereas ChatGPT is generating results based on large public datasets, AI search is delivering results based on your site’s data.

In doing so, AI search improves your ecommerce store’s results. It:

  • Boosts conversion rates even more!
  • Improves conversion rates across your catalog
  • Eliminates null results and enriches search results

I’ll touch briefly on each of these points. 

Boosts conversion rates even more

By better understanding user intent, AI search engines can improve both relevance and ranking to deliver higher-converting products. Algolia NeuralSearch, our own AI platform, has shown tremendous results in a very short period of time. Early beta customers have seen numbers such as a 70% drop in null results and 17% uplift in search-driven conversion rates in just the first few weeks of testing. Results vary from site to site, and the product is only newly launched and many new features are coming, but we expect to see it continue to rise as more data is received for self-learning optimization.

This brings up an interesting point. Data that’s generated on ecommerce websites from queries, clicks, conversions, sign-ups, etc., is used by the AI engine to improve results automatically. AI-powered ranking and re-ranking means you can set it and forget it, and in time, things will self-optimize. However, retailers can also get more immediate outcomes, influencing results by tweaking the search algorithm to match their business priorities and improve their numbers even faster. 

Improves conversion rates across your catalog

While most retailers have only enough time and resources to optimize search for the top 20% of their product catalogs, AI search can optimize all 100% of your catalog with less effort.

On most sites, the top 10–20% of search queries use “head” terms — common phrases that most keyword search engines can easily handle. The bottom 50–80% of queries use long-tail keywords; these are less common queries but they make up a huge percentage of all search queries. Improving these queries requires writing rules and synonyms or doing other hacks and workarounds like keyword stuffing.

AI search offers a solution to optimizing everything. It has human-like understanding and doesn’t rely on keywords for retrieving results. It can understand the intent of someone searching for “something to keep my tea warm” or “best pedometer” — even if these long-tail keywords aren’t in your search index — to deliver better results to improve the end-user search experience. It can improve revenue everywhere — from the head to the long tail — without adding any additional work or overhead.

Eliminates null results and enriches search results

Not only does AI search drastically reduce null searches, it enriches queries that would have produced only a handful of results. We can measure the improvement using precision and recall. 

  • Precision is the percentage of retrieved documents that are relevant. Ideally, search results contain only relevant items, but this is not always the case. See the example below. 
  • Recall is the percentage of all relevant documents that are retrieved. Ideally, every record relevant for the query appears in the results, but some relevant documents (for example, a record containing “NYC” for the query “New York”) may not be found.

Precision and recall are useful metrics for helping determine whether the results are any good. Ideally, precision and recall would both score 100%, and while in practice with a traditional keyword engine that’s very difficult, with AI search, it’s much closer to the ideal! 

Here’s an example. We entered the same query, “chocoholic gift”, in both a traditional keyword search engine and an AI search engine.

The keyword engine results (shown in the image above) were ok, they didn’t return very many results. “Chocoholic” is a colloquialism for “chocolate enthusiast”, but a keyword engine wouldn’t understand that unless the retailer had created search-engine synonyms or rules indicating that the phrase is related to chocolate.

The AI engine (see image below) not only understood the query and was able to return good results (precision), it delivered much richer results (recall). Plus, the fuller results also improved the list of dynamic filters to choose from!

The bottom line

The better your AI solution for site or mobile search, the more revenue you can deliver from website traffic. Whether you’re using traditional keyword search or AI search, better search functionality can have a big impact on your company’s bottom line. AI search not only increases conversions, it gives you an immediate competitive advantage because it reduces consumer frustration. Algolia NeuralSearch is our own AI engine that combines the power of AI and keyword search through a single API for any ecommerce platform. Contact us to see how search can improve user experience and ecommerce ROI to new heights.

About the authorJon Silvers

Jon Silvers

Director, Digital Marketing

Recommended Articles

Powered by Algolia AI Recommendations

How to optimize your ecommerce site search

How to optimize your ecommerce site search

Louise Vollaire

Louise Vollaire

Product Marketing Manager
What is AI-powered site search?

What is AI-powered site search?

John Stewart

John Stewart

VP, Corporate Communications and Brand
How AI search enables ecommerce companies to boost revenue and cut costs

How AI search enables ecommerce companies to boost revenue and cut costs

Michelle Adams

Michelle Adams

Chief Revenue Officer at Algolia