The evidence

Why search speed matters

When a shop’s search or filters take seconds, people notice. This page collects the outside studies that look at speed and shop search, in plain words, each with who ran it, who paid for it and what it cannot tell you.

We list them so you can judge for yourself. None of them is about our product.

Eight studies

What was found, and the caveat next to it.

Findings are paraphrased. Every source link opens in a new tab and was checked on 4 October 2026.

Agencies 55 and Deloitte · 2020 (2019 data)

A tenth of a second on mobile

Across 37 brand sites, a 0.1 second improvement in four mobile speed measures went together with 8.4% higher conversion on retail sites.

Method
Over 30 million sessions on 37 European and American brand sites, watched for about a month.
Funded by
Commissioned by Google.
Caveat
Observational, mobile only, and each of the four speed measures improved on every page of the journey. It is about page loading, not search.

web.dev: Milliseconds make millions

Google (Jake Brutlag) · 2009

Slower search results, fewer searches

When Google added 100 to 400 milliseconds to its search results in a controlled test, people searched a little less (0.2% to 0.6% fewer searches per user), and some of the drop stayed after the delay was removed.

Method
Controlled experiment with a control group, delays added on the server side.
Funded by
Run by Google on its own search engine.
Caveat
Google web search in 2009, not a shop. It measures how many searches people made, not sales.

Google: Speed matters for web search (PDF)

Constructor, as reported by IT Brief · 2024 (October to December)

Shoppers who search carry a lot of revenue

In an analysis of 609 million searches on 113 retail sites, shoppers who used search were 24% of traffic and 44% of revenue.

Method
Search and revenue data from 113 retail sites, about $9.8 billion in revenue.
Funded by
Constructor sells e-commerce search.
Caveat
Correlation: people who search already know what they want. It does not show that fixing search raises sales. We link press coverage; the vendor report itself was not opened.

IT Brief coverage

HTTP Archive, Web Almanac · 2024

WooCommerce and mobile loading speed

In 2024, 34% of WooCommerce sites had a good mobile Largest Contentful Paint, below Shopify, OpenCart and Shopware.

Method
Field data from real Chrome users across the web, grouped by detected platform.
Funded by
Community project with no vendor stake.
Caveat
Covers all WooCommerce sites and is about how fast the first big element appears, which is mostly front-end weight. We do not fix that. It says nothing about search or filters.

Web Almanac 2024: E-commerce

Jakob Nielsen, published by Nielsen Norman Group · 1993

How long is too long, for a person

About 0.1 seconds feels instant, about 1 second keeps a person’s train of thought unbroken, and 10 seconds is about the limit of their attention.

Method
Long-standing human-factors guidance on response times.
Funded by
NN/G sells training; this is general guidance, not a sales study.
Caveat
A guideline about perception. It does not say what a delay costs a shop.

NN/G: Response times, the 3 important limits

Vodafone (case study published by Google) · 2021

One company, a real A/B test

In a 50/50 A/B test, Vodafone’s faster page variant (31% better Largest Contentful Paint) produced 8% more sales.

Method
Live A/B test comparing a page before and after several speed changes.
Funded by
Published by Google on web.dev.
Caveat
One telecom company, several changes bundled together (server rendering, images), so no single cause can be named. Page loading, not search.

web.dev: Vodafone case study

Baymard Institute · 2021

Search matters when the catalog is big

In Baymard’s tests on small direct-to-consumer sites, every tester began by browsing the menu and only 15% even looked for search. The study says search grows in importance as the catalog grows, from roughly 20 products up.

Method
Usability testing on niche direct-to-consumer sites and a benchmark of 26 sites.
Funded by
Baymard sells research and audits.
Caveat
Qualitative testing with small samples. Its point is that tiny shops rarely need search, which is why we write for shops with many products.

Baymard: DTC sites rarely need search

Read this part too

What these studies do not show.

  • That Searchmaker raises conversion or revenue.

    None of these studies tested Searchmaker or any local-index search plugin, and no Searchmaker store has a conversion measurement yet.

  • That page-speed results carry over to search latency.

    Most studies here measure how fast a page loads. Google’s 2009 test is about delays in Google’s own search results. Neither is the time a shop’s live search takes to answer.

  • That “56%” means lost sales in WooCommerce shops.

    Baymard’s 56% is a user-experience quality score of 170+ leading sites and apps. It is not a measure of sales, and it is not about WooCommerce.

  • That shoppers who search buy more because they search.

    People who use search already know what they want. Constructor’s numbers are a correlation, not proof that improving search lifts sales.

  • That we fix your loading speed.

    WooCommerce’s weak mobile Largest Contentful Paint is mostly front-end weight: themes, images, scripts. Searchmaker works on search, filters and feeds. It does not fix LCP.

What we measured ourselves

Our own numbers, with their scope.

These come from our benchmarks. They are different layers, so we never compare one with another.

One real store, about 6,500 products

The theme’s live search took about 1.4 s per search, round trip. Searchmaker’s engine answered in single-digit milliseconds. Different layers, so no ratio. One store, our test.

Engine level, synthetic 100,000 products

Filtered queries in well under a millisecond of engine time, with no web server and no WordPress. Not a page-load time.

Engine level, synthetic 100,000 products

Full index build in about 6 s, and a Google Merchant feed of 240,000 per-variation items in about 6 s. Not measured on a real WooCommerce database or a slow host.

See how it behaves on your own catalog.

Try the in-browser demo, then check the numbers yourself. The 30-day money-back guarantee covers the rest.