One in five pages is not a page

The page list on a statistics page tells you which of your pages the visits landed on. On 26 August 2026 that table held 101,485 rows, and 20,055 of them — 19.8 % — have the page recorded as N/A.

That is not a bug and it is not a missing page. It is the service writing down, honestly, that it did not know.

Where the page address comes from

There are two ways a counter can tell the server which page it sits on.

• The script version reads the address of the page it is running in and sends it along.
• The image version cannot. An <img> tag runs no code. All the server gets is the browser's referrer header — and that is missing whenever the page is served over a policy that withholds it, or the image is fetched in a context that sends none.

When neither is available, the row is written as N/A rather than guessed at. The queries that build the page list then exclude it explicitly, because a page called "N/A" would sort into the list as though it were a real page and win.

Why the share is this high here

Because of where the counters actually live. The largest single platform using this service puts counters in a profile description field, which accepts no scripts, so the image version is the only option there. That is not a small corner of the traffic — it is the majority of the counters whose page we know.

So the fifth of the page list that says nothing is mostly one large, entirely reasonable use of the service, doing the only thing it can.

What you can do about it

If you use the image version, the embed step has a field under the code for the address of your page. Filling it in is the whole fix: the address travels with the request, and your visits get attributed to a page instead of to N/A. It takes ten seconds and it is easy to skip, because nothing is obviously broken if you do.

If you can use the script version, it handles this by itself, along with page changes on a site that swaps content without reloading.

The part worth generalising

A missing value has to be visibly missing. The tempting alternative — fall back to the counter's registered site address, or to the last page seen, or to the domain — would have filled that fifth of the table with plausible, unmarked, wrong rows. Nobody would ever have found them, because they would look exactly like the right ones.

N/A is ugly in a list. That is its entire job.