A website looking good and a website being built correctly are two different things. Design is visible at first glance; whether the address redirects to the secure version, or whether the page even declares a main heading, only shows up when you look at the source.
So instead of guessing, we measured. We examined the homepages of 31 institutional websites in İzmir with a single consistent method and we are publishing the results as they are. The goal is not to single anyone out, but to make the repeating mistakes visible and leave you a checklist you can run on your own site within an hour.
What we measured, and how
- Sample: 31 public-sector websites in İzmir (the metropolitan municipality and district municipalities). All publicly accessible addresses.
- Date: 18 September 2026.
- Method: Homepage only, a single request. Automated checks on the server response, the headers and the HTML source.
- Scope: HTTPS redirection, security headers, compression, mobile tag, heading structure, canonical, structured data, robots.txt and sitemap, image and script counts.
- Result: 30 of the 31 sites could be measured. One domain did not open at all: the address resolves to a network device rather than a website, and the browser raises a security warning.
A note on transparency: the percentages below are calculated over the 30 measurable sites. We do not name individual institutions; the point of the exercise is to show common mistakes, not to publish a scoreboard.
The results
| Check | Sites passing | Share |
|---|---|---|
| Compression (gzip/brotli) enabled | 29 / 30 | 97% |
| Page title present | 29 / 30 | 97% |
| Mobile viewport tag present | 27 / 30 | 90% |
| Cache-Control header present | 22 / 30 | 73% |
| Page language (html lang) declared | 22 / 30 | 73% |
| HTTP redirects to HTTPS | 21 / 30 | 70% |
| Meta description present | 21 / 30 | 70% |
| robots.txt exists | 20 / 30 | 67% |
| sitemap.xml exists | 17 / 30 | 57% |
| Open Graph sharing tags present | 14 / 30 | 47% |
| HSTS security header present | 10 / 30 | 33% |
| Canonical tag present | 10 / 30 | 33% |
| Sitemap line inside robots.txt | 9 / 30 | 30% |
| Exactly one h1 heading | 8 / 30 | 27% |
| Structured data (schema) present | 4 / 30 | 13% |
Speed was better than we expected: the median homepage response time was 0.44 seconds and only two sites answered in more than two seconds. The servers are not slow. What is missing are the basic signals a page uses to introduce itself to a search engine.
The eight most common technical mistakes
1. No main heading (h1) at all
57% of the sites had no h1 element on the homepage, and only 27% had exactly one. The h1 is the page saying “this page is about this”. A slider caption rendered as an image does not replace it. Every page should carry one meaningful h1.
2. No canonical tag
Only 33% declared a canonical URL. When the same page is reachable with and without www, with parameters or in different letter cases, the search engine picks the main version itself. That splits the strength of a single piece of content across duplicates.
3. HTTP is not redirected to the secure version
30% of the sites did not send an http:// visitor to https://. Having an SSL certificate is not enough on its own: without the redirect, the insecure version stays live and the same content is served from two different addresses.
4. Missing sitemap, or a sitemap nobody is told about
57% had a sitemap.xml, but only 30% referenced it inside robots.txt. A sitemap is the most direct way of telling a search engine which pages exist, and in most content management systems it is a single setting.
5. No structured data
Only 13% published structured data (schema.org). Providing the organisation name, address, phone number and opening hours in machine-readable form is the cheapest way to appear correctly in search and map results.
6. Missing sharing tags
53% had no Open Graph tags. Without them, a link shared on WhatsApp or LinkedIn shows no proper title or image — it looks like a bare URL, and click-through drops.
7. Homepages carrying too much weight
The median homepage contained 50 images, the heaviest one 317. Yet only 23% used lazy loading for images. On a mobile connection that means downloading data nobody has scrolled to yet.
8. Legacy technology left in place
80% still loaded jQuery, and on one homepage we counted 214 separate script tags. Legacy libraries are not a mistake in themselves, but carrying one just for a couple of small effects slows the page down and becomes a security risk once it stops being updated.
Check your own site in 15 minutes
- Address test: type your site with http://. If it does not switch to https:// automatically, the redirect is missing.
- Heading test: open “view page source” and search for <h1. If there is none — or more than one — fix it.
- Canonical test: search the same source for canonical.
- File test: open yourdomain.com/robots.txt and yourdomain.com/sitemap.xml. Both should load.
- Sharing test: send your own link to yourself on WhatsApp. No title or image means the OG tags are missing.
- Mobile test: open it on a phone. If you have to scroll sideways to read, the viewport setting or the layout needs work.
If you would rather not run these checks by hand, our free website technical analysis tool applies exactly the same checks automatically: enter your address, no sign-up required. Each item is shown next to the share of the 30 institutional sites we measured that passed it.
How our own site scores
We ran the same measurement on our own site, and we would rather publish that than keep it quiet. ikibinbes.com passes every check: HTTPS redirection and HSTS are active, compression and cache headers are set, the page has exactly one h1, canonical and sharing tags are in place, four structured-data blocks are published, and robots.txt and the sitemap both exist and reference each other. All 27 images on the homepage have alt text and are lazy-loaded.
Limits of the method
This measurement looked only at homepages, and only once. We did not measure perceived loading speed (Core Web Vitals); server response time is not the same thing. Inner pages, form flows, accessibility and content quality were out of scope. Because the sample consists of public institutions, the results do not represent the private-sector average — although we meet the same mistakes regularly on the commercial sites we review.
The method is deliberately repeatable, and we plan to run it again and publish the change. If you would like to review the technical state of your own site with us, see our web design in İzmir page or drop us a line.
Summary
The real problem in the sites we measured was not speed but missing fundamentals: heading structure, canonical tags, structured data and sitemaps. These are all free, one-off and technically small jobs — yet they are the most frequently skipped. Checking the eight items above on your own site takes about an hour. See our web design in İzmir page or get in touch to review your site together.