Researchers showed strangers many photographs of the same people. Different pictures of one person produced about as much variation in what strangers thought of them as photographs of completely different people did.

The photograph is doing more of the work than your face is

This is the finding that justifies caring about any of this, and almost nobody writing about headshots seems to know it exists.

In 2014, Alexander Todorov and Jenny Porter published a study in Psychological Science with a title that gives it away: Misleading First Impressions: Different for Different Facial Images of the Same Person. They had people rate photographs on attractiveness, competence, trustworthiness, intelligence and several other traits.

The result is the useful part. Different photographs of the same individual produced about as much variation in how they were judged as photographs of entirely different individuals did. Lighting, angle, expression and the particular moment shifted the ratings that much.

Not three seconds. About a sixth of one.

You will read everywhere that casting directors look at a headshot for three seconds. We went looking for the source of that and there is not one. What replaces it is real, and it is worse news.

The three-second figure circulates entirely among photographers’ marketing pages and content sites, always stated as received wisdom and never traced to a casting director, a study or a publication. We could not falsify it because there is nothing there to falsify.

The commercial casting director Stuart Stone wrote something different and more defensible in Backstage in March 2016. “Within a fraction of a second, a casting director is consciously or unconsciously forming an opinion about you based on what they are seeing in your headshot.” That is a claim about how fast an impression forms, not about how long anybody looks. The two got conflated somewhere and the folklore grew out of it.

Ninety-six to a page

One sentence from one casting director organises this entire section of the Academy. It is real, it is on the record, and it has exactly one source.

Tree Petts founded and chairs the Casting Directors’ Association and has cast more than two thousand commercials. Speaking on video for Spotlight, she said this. “I think what people need to understand is that I’ll get 48 to 96 thumbnails a page. They need to have a strong, good headshot, and most importantly of all, they need to look like their headshot.”

Every rule downstream follows from that number. At that size a headshot is not a portrait. It is a very small signal in a dense field, competing with ninety-five others for a sixth of a second.

Does any of this actually work?

The same body of research that says impressions form in a tenth of a second also says those impressions are mostly wrong. Both halves are useful and most pages only publish the first.

Fast, confident and consistent between viewers is not the same as accurate. The 2006 study established the speed and stability of face-based impressions, not their truth — the researchers never checked whether the photographed people were actually trustworthy or competent.

Later work went looking for that accuracy. A meta-analysis published by Foo and colleagues found a face-level correlation of 0.14 between how trustworthy a face looks and how trustworthy the person is. A 2024 study by Jaeger and colleagues extended the scepticism to personality judgements generally, finding little accuracy and, more strikingly, little awareness among people of when their own impressions were wrong.

Put the two halves together and you get the only sane conclusion. Somebody will form a confident opinion of you from a photograph in less time than it takes to blink, and that opinion will mostly be wrong. The first half is why the photograph is worth getting right. The second half is why a headshot that is not working is not a verdict on you.

What casting actually says when you ask them

Spotlight put eleven casting directors on video and asked them about headshots. What is striking is what none of them mentions.

McWilliams says she does not know why outdoors works better. There is an answer, and it is just physics. Open shade and north-facing daylight are large, soft, directional light sources, which is exactly what an expensive studio softbox is built to imitate. The sky is bigger than any softbox and costs nothing.

The catchlight, and how to judge a photographer for nothing

A catchlight is the reflection of the light source in the eye. It is small, it is the most useful thing on this page, and it lets you assess a portfolio in about ten seconds.

The optics are not in dispute. A catchlight is a specular reflection off the wet surface of the eye, which means its size and position in the eye directly reveal the size and position of the light that made it. A big soft catchlight means a big soft source. A tiny hard dot means a small hard one.

Crop and background, and a correction to our own research

The advice here is right. The evidence we were given for it turned out to be three different studies fused into one, and one of them carried a claim we were about to publish about skin tone.

The practical rule comes from a casting professional: full heads, not cut off at the top, so the shape of the head is visible. The reasoning offered to us was that when image quality drops, viewers stop being able to read internal features and fall back on the outline — head shape, hair, silhouette.

There is one more limit worth naming. All of those studies measured whether people could correctly identify a face, which is a different task from forming an impression of one. So the full-head rule is sensible and well motivated, and it is not proven by that research. We would rather you had the reasoning with its weak joint visible than a confident sentence with a hidden one.

Background follows the same logic. If the outline of your head is doing identification work, the boundary between your head and what is behind it is load-bearing. Dark hair against a dark background is fine at full size and gone at ninety-six to a page.

The seven-point test

Everything above, as something you can actually do to a photograph in two minutes.

That last point deserves saying plainly. Your judgement about your own photograph is the least reliable instrument in this entire process. You are not asking whether you look good. You are asking whether you look like yourself, small.

Grading the sources

The science on this page is strong and our research had drifted on the details of it in four separate places. Two of those drifts would have been published as facts.

That last row is worth dwelling on, because it applies to this whole page. The research is about single faces, shown alone, for controlled durations. A casting director scanning a dense grid under time pressure is doing a different task, and the closest published work — on how people form averaged impressions of crowds of faces — is not the same thing either. The transfer is our argument rather than anybody’s finding, and nobody has ever put an eye-tracker on a casting director. Somebody should.

Where we differ from the standard account

Do this now and it takes two minutes. Open your headshot, shrink it until it is the size of a postage stamp, and look away for five seconds before looking back. If you cannot find your own eyes in the first instant, neither can anyone else — and they are not going to look for as long as you just did.