We pulled 95,384 real URLs out of the sitemaps of 410 Persian sites and counted their shape. A third of them contain Persian letters once decoded, and 232 of the 410 sites use at least one Persian URL.
Contrary to the usual advice, this does not hurt rankings. Google has read Unicode URLs for years. The problem lies somewhere else entirely, and it is the same reason we write ASCII slugs on this site.
What Persian URLs actually look like
| Across 95,384 URLs from 410 sites | Value |
|---|---|
| Median path depth | 2 segments |
| Deepest path | 8 segments |
| Percent encoded paths | 25,729 (27%) |
| Paths with Persian letters once decoded | 31,912 (33%) |
| URLs carrying a query string | 447 (under 1%) |
| Paths with an uppercase letter | 9,320 |
| Paths using an underscore | 719 |
| Paths carrying a date | 1,340 |
| Paths ending in html, php or aspx | 2,320 |
In the same data only 447 URLs carried a query string, under one percent; the era of parameter URLs is effectively over on the Persian web. Median path length is 35 characters. The ninetieth percentile is 158. And the longest path we found ran to 614 characters: the entire opening paragraph of a story, used as the page address.
The real problem with a Persian URL
It is not ranking. It is daily life. Copy a Persian URL and it becomes percent encoding, and that string cannot go into a WhatsApp message, an invoice, a support ticket or a phone call without tripling in length and becoming unreadable.
We write short ASCII slugs on this site for exactly that reason, not for SEO, which is the same rule that sits in our pre-publish checklist. If somebody tells you a Persian URL damages your position, ask for the evidence. There is none.
And the part that does affect ranking
Length. A ninetieth percentile of 158 characters means one URL in ten is long enough to be truncated in the search result and unreadable in the address bar.
That 614 character path is not an accident either. It was generated from the full post title, and the title was itself a sentence. The system builds the slug from the heading and nobody set a ceiling.
Rules that fall out of this data
- Two segments is enough. That is the actual median, and most sites work at one or two. Depth of four and beyond usually comes from nested categories nobody shows in the menu.
- Take the date out of the path. 1,340 URLs in our sample carry a date. A piece you update three years later still has an address announcing it as three years old, and changing that structure afterwards is expensive.
- Drop uppercase and underscores. 9,320 paths contain an uppercase letter. Servers are generally case sensitive, which is how two different addresses end up serving one page.
- Do not keep the file extension. 2,320 URLs still end in html, php or aspx. No harm today, and on the day the platform changes every one has to be redirected.
If your URLs are bad today
Our position is simple: do not change them without a strong reason. Changing the address of a ranking page throws away its history unless a 301 is in place at the same instant. Even with the redirect done correctly you will see weeks of movement.
The working rule we apply to ourselves: URL structure changes only when it solves another problem at the same time, and never in the same deployment as a title and meta rewrite. If rankings drop you need to know which change caused it, and diagnosing a drop is impossible when three things moved together.
What this count cannot tell you
The URLs come from sitemaps, so we only see what each site declares. Pages absent from the sitemap are absent from this sample. For large sites we read at most four hundred URLs from two child sitemaps, so large sites are under represented relative to their real size.
Questions people actually ask
Are Persian URLs bad for SEO?
No. A third of our sample is Persian and those sites rank. What is bad is a long, uncopyable URL, and a Persian one incidentally tends to be both at once.
Should the keyword be in the URL?
If it fits naturally, yes, but the effect is small. The URL is a weak signal and the rest of the on page checklist moves more.
How many words in a slug?
Three to five is usually enough. Our sample's median of 35 characters lands in about the same place. If your slug passes fifty characters, you have probably copied the title verbatim.
Keep the category in the path?
If categories are stable, yes. If they might be reorganised, no. A category in the path means every tidy up of your taxonomy moves URLs. That is one reason the search work we take on treats a category change as a planned migration rather than a quick clean up.
If you want to know what your own URLs look like and how many pass 158 characters, send us the sitemap and we will run this same count.
Measured 3 September 2026: 95,384 URLs from the sitemaps of 410 Persian domains, at most 400 per domain.
Comments & Questions
Have a question about this article? Ask, we'll answer.
No comments yet; be the first.