State of Agency Websites 2026
Agencies sell websites for a living, so their own sites should be the proof. We measured 61 of them the same way we measure ours: real tools, published methodology, and numbers anyone can re-run. Published 2026-09-06, from a 2026-09-04 snapshot.
The short version: the median agency website scores 49 of 100 on mobile performance, roughly half send one security header or none, and about one in five publishes no price of any kind. The industry that builds the modern web mostly does not hold its own work to the standard it sells.
Every figure below is an aggregate. We do not name agencies here: the point is the state of the industry, not a leaderboard.
The numbers
| What we measured | Result |
|---|---|
| Agencies measurable with PageSpeed Insights on mobile | 59 of 61 |
| Median mobile Performance score | 49 |
| Agencies below 50 Performance on mobile | 30 |
| Agencies at 90 or above on mobile | 4 |
| Median desktop Performance score | 65 |
| Median Accessibility score on mobile | 90 (only 5 of 61 score 100) |
| Agencies sending 0 or 1 security headers | 29 of 61 |
| Agencies sending any Content-Security-Policy | 21 of 61 |
| Agencies with quote-only pricing, no number on the site | 22 of 61 |
| Home pages with no JSON-LD at all | 12 of 61 |
| Home pages with images missing alt text | 20 of 61 |
| Home pages without exactly one H1 | 11 of 61 |
What it means if you are hiring
A slow, unlabeled, quote-only agency site is not automatically a bad agency. It is a missed proof. When only 4 of 61 agencies clear 90 on the phone in your pocket, "we build fast websites" is a claim you should test, not trust: run any agency’s own homepage through PageSpeed Insights before you sign, ask where their pricing is published, and view source to see whether the structured data they promise you exists on their own pages.
That is the test we invite on ourselves. Our own scores, headers, and markup are graded in public on the performance and security pages, and you can run the same checks on any site with our free tools .
Methodology
We measured 61 digital agencies with a public website, snapshot dated 2026-09-04. Performance and accessibility come from Google PageSpeed Insights (mobile and desktop, Lighthouse lab data); two sites could not be measured on mobile and are excluded from performance medians only. Security headers were read from each home page’s live response. Structured data, alt text, and H1 counts were read from each rendered home page. Medians are reported instead of averages so a handful of outliers cannot flatter the industry.
The aggregate table regenerates from the raw dataset with a committed script, so next year’s edition re-runs the same measurements instead of rewriting them. The raw per-agency data stays private by design: this report exists to describe the industry, not to name names.