When the photographer field is filled in, it is almost never a placeholder

Image files have a field for the person who made them. Common sense says it is mostly junk — blank, or N/A, or unknown, or whatever the template put there. We went looking for that junk across 239,632 corporate images, and it is almost entirely absent.

The distribution

Fifty-three per cent of filled creator fields name an individual, across 121 companies. Placeholders account for less than one fifth of one per cent — and the placeholder count is the one number in this table that did not grow at all as the crawl went on.

This was the opposite of what we expected

The reason for counting was a disagreement about whether the field was worth reading at all. The argument for reading it was that a field full of N/A is a field you can safely ignore; the argument against was that if it turned out to be full of real names, we would have built a machine for collecting them.

It is full of real names. The placeholder theory was wrong by two orders of magnitude, and it was wrong in the direction that mattered.

So the design that was built for the optimistic case is the one doing the work in the pessimistic one: the name is never stored. What is stored is which of the six rows above a value fell into — a category, not a value — which is enough to publish this table and not enough to reconstruct a single line of it.

The 69 that are contact details

The smallest interesting row. Sixty-nine images carry an email address, a personal website or a social handle in the creator field — a freelance photographer's business card, embedded in a file, shipped to a corporate website, and served to anybody who asks.

That is a more direct exposure than a name on its own, and it is the one case here where the person affected has a plain commercial interest in the metadata surviving. A photographer who puts their email in the file is not making a mistake. Whether they expected it to be readable from a FTSE 100 press page is a different question.

What this says about the pictures

Read against the GPS finding, a picture forms. Two populations of images arrive at a corporate website. One comes from a professional workflow: it has been through Lightroom or Capture One, the location has been stripped, and the photographer's name is intact because the photographer put it there on purpose. The other arrives from somebody's phone, untouched, with a location and no name at all.

Corporate image handling is not careless or careful. It is two pipelines, and only one of them has anybody looking at it.

How this was measured

The first 64 KB of each image is requested, parsed and discarded. The creator fields read are EXIF Artist, IPTC By-line and Credit, and XMP dc:creator and equivalents.

Each value is sorted into one of the six categories above by shape alone, and only the category is written down. “A person's name” means it looks like one — two to four name-shaped words, no digits, no company suffix, not a known agency. It is not a claim that a person of that name exists.

• The classifier will be wrong at the edges. A one-word value is left unclassified rather than guessed at, which is most of the 1,415; a sole trader whose business is their own name will read as a person, correctly by our definition and arguably not by theirs.
• 4,816 of 239,632 images have the field filled at all — 2.0%. This article is about that 2.0%, not about corporate images in general.
• It covers the companies we could crawl. See coverage.

Collected 1–4 September 2026, over a completed first pass.