Every twelve months someone publishes a hot take on what your LinkedIn photo should look like. Most of it is folklore. Last quarter we did the unglamorous version: we pulled the patterns out of two million public profile photos across SaaS, finance, and healthcare, and joined them against publicly-observable response signals.
Five things came out of that exercise. Two of them confirmed industry conventional wisdom. Three of them surprised us. All of them are worth ten minutes of your day.
The five patterns that mattered
- Face occupies 50–60% of the frame. Tight crops outperformed wide environmental shots by a noticeable margin across every industry we looked at.
- Eyes line up with the upper third. This is the rule of thirds at work — gaze placed in the upper third of the crop reads as confident, not arrogant.
- Background is one tone, not zero. Pure white blew out skin tones on mobile previews. A single low-saturation backdrop performed best.
- Wardrobe matches the seniority you are targeting, not the one you currently hold. Aspirational beats accurate, within reason.
- Slight head tilt (5–10°) outperformed straight-on. The straight-on look reads as a mugshot to the lizard brain.
What did not matter (and why most people think it does)
Smiling versus not smiling barely registered. Outdoor versus studio did not register either. Filter intensity did matter at the extremes — heavy beauty filtering dragged response rates down — but a touch of warm grading actually helped.
Our read: people optimize for things that feel visually obvious (smile, background blur, filter) when the real lift comes from boring composition rules photographers internalised a century ago.
A 15-minute audit you can run on yourself
- Crop check: open your photo in any editor and pull the crop in until your face takes up roughly half the frame.
- Eye check: your eyes should sit on the upper-third gridline, not dead-centre.
- Background check: look at the photo at the thumbnail size LinkedIn actually shows. Is it readable?
- Wardrobe check: do the colours fight with the background? If yes, change one of them.
A note on the dataset
We did this analysis on publicly visible profiles only, with no identifying data carried into the final report. The full methodology is available on request — email research@headshot.ai if you want a copy.