Page builder traces and speed on Turkish manufacturer websites.

What did we find?

• For 43 of the 77 sites we could reach, Chrome’s real user data (CrUX) was available (55.8%); 15 of them were at the good threshold on mobile for all three metrics.
• Sites with LCP at the good threshold on mobile: 22; with good INP: 39; with good CLS: 35: the metric that pulls passing down most is LCP. The median LCP is 2,351 ms; the good threshold is 2,500 ms.
• Sites where we found a WordPress trace on the home page: 24; 18 of them also carried a page builder trace (Elementor or WPBakery).
• Of the 19 sites carrying a page builder trace, 1 of the 14 with CrUX data passed.
• Sites where we found no trace of a known builder or content management system on the home page: 51; 12 of the 23 of them with CrUX data passed.

What did we measure, and how?

The sample is that of the same measurement in which we measured certificate pages: 81 sites of exporting manufacturers in four provinces that have their own website; measurement day November 10, 2026. We fetched the HTML of each site’s home page once and looked in it for traces of known builders and content management systems: Elementor, Divi, WPBakery, WordPress, Wix, Squarespace, Webflow, Duda, Shopify, Joomla, Drupal. We recorded a site where we found no trace as “not detected”; we do not read this as “custom built”, because a system may hide its trace, or it may be a system not on our list.

We did not measure speed ourselves: we took the CrUX data Google collects from Chrome users, for mobile and at domain level. The values are the 75th percentile of the 28 days before the measurement day; that is, what three quarters of users experienced, or better. A site counts as passing if it is at the good threshold for all three metrics: LCP 2.5 seconds or less, INP 200 milliseconds or less, CLS 0.1 or less. CrUX does not collect data for sites without enough traffic; that is why the base of the shares is the sites with data.

Which builder traces are seen?

On most sites, we saw none: 51 of the 77 sites we could reach had no known trace (66.2%). Almost all of those carrying a trace are from the WordPress family: WordPress 24, Elementor 12, WPBakery 10, Shopify 1. We found no Wix, Squarespace, Webflow, Duda or Divi traces in this sample (0, 0, 0, 0 and 0 respectively).

The WordPress and page builder counts are not independent of each other: Elementor and WPBakery are plugins built on top of WordPress. Of the 24 sites carrying a WordPress trace, 18 also had a builder and 6 did not.

How fast are they on mobile?

Of the 43 sites with data, 15 passed all three metrics together (34.9%). Looking metric by metric, the problem is concentrated in one place: while 39 sites were at the good threshold for INP, which measures response time to interaction, and 35 sites for CLS, which measures how much the page shifts, only 22 sites were at the good threshold for LCP, which measures how long the main content takes to appear. The median LCP is 2,351 ms, the median INP 105 ms.

LCP is often determined by a large image, a slow-responding server or scripts that delay the main content. This measurement does not show the cause; which one matters on which site can only be seen from that site’s own measurement.

Is there a relationship between builder traces and speed?

In this sample there is a direction, but the numbers are small. Of the sites carrying a page builder trace, 1 of the 14 with CrUX data passed; of the 18 sites carrying a WordPress trace, 2; of the 23 sites where we found no trace, 12. Because the groups are small, we give no shares, and this is not cause and effect: other shared characteristics of sites using builders (hosting, use of images, number of plugins) can produce the same result.

Global data points in the same direction. In HTTP Archive’s September 2026 data, the share of sites passing Core Web Vitals on mobile is 52.8% for all sites, 48.3% for WordPress sites, 38.7% for WPBakery and 36.3% for Elementor (we calculated the shares by dividing the number of passing sites by the number of sites tested). This is a global data set; our sample is exporters in four provinces, so the comparison gives a direction, not a match.

What is the situation for machinery manufacturers?

Of the 8 machinery manufacturer websites we measured separately, 6 had CrUX data and 5 passed. The group is not large enough to give a share; we give the number only for the record.

What are the limits of the measurement?

• Builder traces come only from the home page’s HTML; inner pages may have been built with a different system.
• Not finding a trace does not mean the site is custom built.
• CrUX data is at domain level and covers the last 28 days, not a single page or a single day.
• We do not know the speed of sites without CrUX data; the base of the shares is the sites with data.
• The relationship between builders and speed comes from small groups; it is not cause and effect.
• The sample is exporters in four provinces, not all of Turkey’s manufacturers.

Where are the data and method?

Anonymous per-site data is published as JSON and CSV under the CC BY 4.0 license: rasathane/uretici-2026-11/veri.csv and veri.json; each row has the builder trace and CrUX values, but no site name or province. The measurement was done with tools/uretici-olcum.mjs and the summary with tools/uretici-ozet.mjs.

What does it mean for your own site?

Look at your own site’s field data: if it has data in CrUX, you will see which of the three metrics falls below the good threshold; on most manufacturer websites, it is LCP. What field data is is explained in the field data entry, and how to measure in the how to measure site speed guide. If you use WordPress and a page builder, where to look is covered in the slow WordPress site guide; the three-year cost of a template and a custom build is in the true cost of a cheap website guide.