There are more measurements behind this page than any other category, because photographers were where this started. Twenty wedding photographers in Sweden, Norway, Germany, Belgium and Canada, measured in September 2026 on their own questions, each question asked three times to Perplexity, ChatGPT with web search, Gemini and Claude.
That comes to one thousand six hundred and seventy nine question and service pairs. Here is what came out of them.
14%
of question and service pairs where a photographer held a stable place
3 of 20
photographers came up on none of their own questions at all
424
answers that cited a measured photographer own website as a source
The four services do not behave alike
This is the single most useful thing in the data, and it is the reason most advice about AI visibility is unhelpful. The services are treated as one thing. They are not.
| Service | Stable places | Share of questions | Cited the photographer own site |
|---|---|---|---|
| Gemini | 81 of 419 | 19% | 1 answer |
| Perplexity | 73 of 420 | 17% | 230 answers |
| Claude | 44 of 420 | 10% | 129 answers |
| ChatGPT | 42 of 420 | 10% | 64 answers |
Stable means named in at least two runs out of three. Twenty photographers, September 2026.
Look at the two ends of that table. Gemini hands out names more often than anything else and almost never shows you where it got them. Perplexity reads photographer websites constantly and cites them two hundred and thirty times.
That split explains the most common complaint I hear. Somebody rebuilds the whole site, tests it in one assistant, sees nothing change, and concludes the whole thing is nonsense. Often they were testing against the service that was never reading websites in the first place.
Near misses are more common than wins
Across the twenty photographers there were eighty nine question and service pairs where the photographer appeared in exactly one run out of three. Not two. One.
Those are not failures. They are the closest thing to a win in the whole report, because the assistant clearly knows the business exists and simply does not reach for it reliably. They are also invisible to any tool that asks a question once, which will report half of them as a win and half as a loss depending on the day.
Three photographers came up on nothing
Not one question. Not one service. Not one run out of three. All three are working photographers with real weddings and real clients, and two of them have websites that are considerably prettier than mine.
What they had in common was that almost nothing on the open web described them in words. No city in the body text, no venue names, no prices, no third party listings that an assistant could quote. The work existed. The description of the work did not.
Where the answers come from instead
In six full photographer reports from Sweden, Germany and Belgium, the most cited sources were photographer directories rather than anything the photographers controlled: MyWed two hundred and seventy seven times, Wezoree two hundred and four, WPJA one hundred and fifty, TheWed one hundred and thirty, Reddit ninety three and Fearless Photographers sixty eight. Below those came other photographers own websites.
Note what is not on that list. In answers about the Swedish and German markets, the American platforms the assistants keep recommending did not appear at all. A free profile on the directory your market actually uses beats a paid profile on the one an American blog post told you about.
What to do about it
The long version, with seven specific actions and examples from a real report, is in the article how to show up when couples ask ChatGPT for a wedding photographer. The short version is three things.
Put the words in. City, region, venues by name, what a wedding costs with you. Not in an image caption. In text an assistant can quote.
Get described somewhere else. The directories your own market cites, venue supplier lists, real wedding features. This is the only lever that moves Gemini.
Measure the same questions monthly. One measurement tells you where you stand. Only a second one tells you whether anything you did worked.
Questions florists ask next
Why does ChatGPT recommend other photographers and not me?
Because it can read more about them than it can about you. In the measurements the photographers who held stable places were the ones being described somewhere other than their own website, in directories, venue pages and features. A portfolio site with no city, no venue names and no prices gives an assistant nothing to quote, however good the pictures are.
Which AI service should a wedding photographer care about most?
The one your couples use, which you cannot control, so the honest answer is that they behave so differently you have to know your own numbers. Across twenty measured photographers Gemini gave out names most often, on nineteen percent of questions, but returned almost no readable sources. Perplexity was close behind at seventeen percent and cited photographer websites two hundred and thirty times. ChatGPT and Claude both sat around ten percent.
Is it normal to appear on none of the questions?
It is more common than you would think. Three of the twenty photographers in the September measurement came up on not a single one of their own questions, on any service, in any run. All three were working photographers with real clients and good work. That is the point of measuring rather than guessing.
How long before a change shows up in the answers?
Plan for about a month before a new page can start appearing, and longer for the services that rely on other sites writing about you. Perplexity tends to react first because it reads your own site directly. Gemini is slowest, because it is waiting for somebody else to describe you.