CWV pass rate for stores with and without a product recommendation engine
| Group | Pass with | Pass without | Median JS with | Median JS without | Stores with (CrUX) |
|---|---|---|---|---|---|
| Shopify | 59.4% | 57.7% | 4.22 MB | 2.93 MB | 2,784 |
| Other store platforms | 37.5% | 41.3% | 2.00 MB | 1.41 MB | 432 |
| Top 100K | 35.9% | 31.8% | n/a | n/a | 184 |
| Rest of top 1M | 57.4% | 49.1% | 3.95 MB | 1.91 MB | 3,082 |
| All stores | 56.2% | 48.4% | 3.94 MB | 1.93 MB | 3,266 |
Source: PageSpeed Matters, Sept 2026. Other store platforms excludes Shopify and WooCommerce. Median JavaScript was not broken out for the Top 100K.
Summary
Product recommendation engines fill product, cart and home pages with "you may also like" and "frequently bought together" blocks. They usually arrive as a third-party script, so store owners often ask whether the extra code costs them speed.
At first glance it does not. Top 1M stores with a recommendation engine pass Core Web Vitals 56.2% of the time, against 48.4% of stores without one, even though they carry about twice the JavaScript. Most of that gap comes from which platform the stores run on, and comparing like with like tells a more careful story.
What we measured
We used the HTTP Archive crawl of the top 1M sites from September 2026, on mobile, and kept homepages that run on a named store platform. A store counts as having a recommendation engine when the HTTP Archive crawl detects one of the common recommendation apps or platforms on the homepage.
That gave 3,413 stores with a recommendation engine (3,266 with Chrome UX Report field data) and 90,598 stores without one (83,021 with field data). We joined each store to CrUX data at the 75th percentile. A store passes Core Web Vitals when loading, responsiveness and visual stability are all rated good. JavaScript weight is the median total JavaScript on the homepage.
These are correlations. Larger brands and Shopify stores adopt recommendation tools more often, so the groups differ in more ways than the tool itself. WooCommerce stores with a recommendation engine were too few to report on their own.
Stores with a recommendation engine pass more often overall
Across the top 1M, 56.2% of stores with a recommendation engine pass Core Web Vitals, against 48.4% of stores without one. That is the headline result, and it holds despite a clear weight difference: stores with a recommendation engine load a median 3.94 MB of JavaScript, against 1.93 MB for stores without one.
On its own, that looks like heavy scripts going with faster stores. It is more likely a sign that the two groups are built on different platforms, which is where the next finding comes in.
56.2%
Share of top 1M stores with a product recommendation engine that pass Core Web Vitals, against 48.4% of stores without one.
Most of the gap is platform mix
83.3% of the 3,413 stores with a recommendation engine run on Shopify. Shopify stores pass Core Web Vitals more often than stores on most other platforms, so a group that is mostly Shopify will look faster whatever apps it carries.
Within Shopify, the gap mostly disappears. Shopify stores with a recommendation engine pass 59.4% of the time (2,784 with field data), against 57.7% without (42,473). They still carry more code: a median 4.22 MB of JavaScript against 2.93 MB. In other words, comparing Shopify with Shopify leaves the two groups close on pass rate, while the recommendation group remains noticeably heavier.
Recommendation engines do not make a store faster. The overall gap mainly reflects the platform those stores are built on.
Off Shopify, the pattern flips
On store platforms other than Shopify and WooCommerce, stores with a recommendation tool pass less often: 37.5% (432 with field data) against 41.3% without (20,056). They also load more JavaScript, a median 2.00 MB against 1.41 MB.
On these platforms a recommendation engine is often added as a separate third-party script, and extra script work while the page loads usually hurts responsiveness and loading. That is a likely reason the with-tool group falls behind here, though the data shows the link, not the cause.
Top 100K stores pass less often, with or without one
Among the most visited stores, both groups pass far less often than the top 1M as a whole. In the Top 100K, stores with a recommendation tool pass 35.9% of the time (184 with field data) against 31.8% without (3,542).
The bulk of the overall result comes from the rest of the top 1M, where stores with a recommendation engine pass 57.4% against 49.1% without. In the Top 100K the with-tool group is slightly ahead, but it is only 184 stores with field data, so treat that comparison with care. The low pass rates in this tier apply to stores with and without a recommendation engine alike.
What this means for your store
A recommendation engine can earn its place by helping shoppers find more products, but the speed cost depends on how it is loaded. The data does not say the tool is free on any platform: in every group with JavaScript figures, stores with a recommendation engine carry more code.
- Load recommendation widgets after the main product content and price have rendered, not in the head of the page.
- Reserve space for each widget block so it does not push content down when it loads.
- Only load the script on templates that show recommendations, such as product and cart pages.
- Remove duplicate tools, such as an old recommendation app left installed after switching to a new one.
- Test your product pages in PageSpeed Insights with the widget on and off, and check your field data in Search Console, so you know what the feature costs against what it earns.
Sources
- HTTP Archive crawl of the top 1M sites, September 2026, mobile
- Chrome UX Report (CrUX) field data, 75th percentile, September 2026
- Core Web Vitals overview, web.dev: https://web.dev/articles/vitals
Embed this research
Paste this on your site to embed the chart. It links back to the source automatically.
<iframe id="psm-product-recommendation-speed" src="https://www.pagespeedmatters.com/embed/data-studies/product-recommendation-speed" width="100%" height="600" style="width:100%;border:1px solid #E2E3E1;border-radius:12px" loading="lazy" title="Product Recommendation Speed: 56.2% vs 48.4% Pass CWV - PageSpeed Matters"></iframe>
<script>window.addEventListener("message",function(e){if(e&&e.data&&e.data.psmEmbed==="product-recommendation-speed"&&e.data.height){var f=document.getElementById("psm-product-recommendation-speed");if(f){f.style.height=e.data.height+"px";}}});</script>
<p style="font:14px/1.5 system-ui,sans-serif">Source: <a href="https://www.pagespeedmatters.com/resources/data-studies/product-recommendation-speed">Product Recommendation Speed: 56.2% vs 48.4% Pass CWV - PageSpeed Matters</a></p>
Frequently asked questions
Do product recommendation apps slow down a store?
They add code: stores with a recommendation engine load a median 3.94 MB of JavaScript against 1.93 MB without. Pass rates depend on platform. On Shopify they pass 59.4% against 57.7% without, while on other store platforms they pass 37.5% against 41.3%.
Why do stores with recommendation engines pass Core Web Vitals more often?
Mostly platform mix. 83.3% of stores with a recommendation engine run on Shopify, which passes more often than most platforms. Within Shopify the gap mostly disappears, at 59.4% against 57.7%.
What percentage of stores with a recommendation engine pass Core Web Vitals?
56.2% of top 1M stores with a detected recommendation engine pass Core Web Vitals, against 48.4% of stores without one. In the Top 100K, the figures are 35.9% against 31.8%.
Recommendation widgets can lift order value, and loading them after the main content keeps that gain from costing product page speed. Need help with this? See our ecommerce CRO agency.