Between two Chrome UX Report snapshots four months apart, .ir went from 6,250 origins to 102,363. That is a 1,538% rise in four months — comfortably the largest move of any extension, and by a wide margin the most interesting number we found.
We are not publishing it as growth, because it is not growth. This post is the check that told us so, because the check generalises and the number does not.
The candidate list
Biggest four-month movers
Extensions with at least 10,000 origins in either snapshot. Earlier snapshot to later snapshot.
Source: two Chrome UX Report top-origins lists four months apart, compared band by band.
.ir is not slightly ahead of the field. It is forty times the next largest move. A gap that size is almost never a fact about the subject; it is a fact about the measurement.
The check: does the change have a shape?
A ranked list gives you a free diagnostic. Split the change by rank tier and ask whether the tiers agree. Real growth has a shape. A coverage change does not.
When an extension genuinely grows, the growth lands somewhere specific. New sites appear at the bottom, so the tail tiers move first and the head barely moves at all. When the instrument starts seeing a population it was not seeing before, every tier fills at once — because the sites were always there.
Growth multiple by rank tier
Later origins ÷ earlier origins, computed separately in each tier. A flat-ish line across every tier means the whole population appeared at once.
- .ir
- .app
Source: two Chrome UX Report top-origins lists four months apart, compared band by band.
.app is the control, and it grew properly: +38.5% overall, and a multiple between 1.1× and 1.4× in every tier, tilted slightly toward the tail. That is what a real four months of new registrations looks like — mostly new small sites, a few of them climbing.
.ir multiplies by between 14× and 51× in every single tier, including tiers where the earlier list had almost nothing. Iran did not build 96,000 popular websites in four months, and it certainly did not distribute them evenly across four orders of magnitude of popularity. What changed is that CrUX started reporting them.
The rule we now apply
- Never publish a change from two totals. A single percentage from two numbers carries no evidence about its own cause. Split it by any dimension the data already has — rank, region, source — and look at whether the parts agree.
- A uniform multiple is a red flag, not a strong result. Real change is lumpy. When every slice moves by roughly the same factor, the most likely explanation is that the denominator changed.
- Keep a control. The .app line is what makes the .ir line readable. A shape means nothing until you have seen the shape of something ordinary.
- Say what you dropped. .ir is excluded from every growth ranking on this site, and this post is why. A number removed silently is indistinguishable from a number nobody found.
The cost of the check was one extra GROUP BY. The cost of skipping it would have been a headline claiming a country’s web grew sixteenfold in a single quarter, which is the kind of thing that gets cited for years.