Photo authority for Person entities (2026): how to make Google pick the right image

Most “photo fixes” fail because people treat the image as decoration.

In Google’s entity world, a person photo is a resolution outcome: the system chooses one image that best matches the identity cluster it trusts. If your panel photo is small, wrong, or keeps changing, the root cause is almost always lack of visual consensus across your strongest identity surfaces.

This page is a practical playbook for making that consensus obvious without hacks.

If you want the canonical identity anchor first: start at the person profile page:

/person/mikhail-drozdov

Direct answer (what to do)

Create one canonical portrait set and reuse it everywhere:

Square (1:1) — (1200 \times 1200)
Vertical (4:5) — (1200 \times 1500)
Horizontal (16:9) — (1600 \times 900)

Then pin the square portrait as the canonical identity image:

on your canonical person page
in Person JSON-LD (image)
and on your strongest external profiles (LinkedIn, Instagram, X, ORCID, etc.)

Do not “optimize images” by adding more variants. More variants = more candidates = less certainty.

Why panels pick “weird” photos

When a panel photo looks wrong, Google is usually not “confused about your face”.

It’s doing one of these:

Choosing the most confident cluster, not the newest one
Preferring a profile hub that has stronger corroboration than your site (even if you dislike the image)
Fallbacking to a small thumbnail because it can’t confidently promote any candidate as the canonical portrait

The fix is not “upload a better photo”. The fix is to make one photo the obvious winner across your Tier‑1 identity surfaces.

The model: visual consensus beats visual quality

Think in three layers:

Candidate pool: every headshot, avatar, and thumbnail tied to your name / alias.
Authority surfaces: which pages/profiles the system trusts as identity references.
Consensus: whether the same image repeats across the strongest references.

You win when the same image shows up on the few references that matter most.

What counts as an “authority surface”

For a person entity, the usual Tier‑1 set is:

a canonical person profile page on your own domain
LinkedIn
Instagram (if it ranks for your brand query)
X (if it ranks and is stable)
ORCID (for researcher positioning)
Crunchbase / Wellfound (if complete and consistent)

Everything else can exist, but it should not compete as “the canonical identity hub”.

Make your person page the image source of truth

On the canonical person page:

Use one portrait image consistently (same file path, same alt text)
Make sure it is present in HTML and not injected late
In JSON-LD Person schema, set:
- image to the canonical portrait URL (or an ImageObject @id that resolves to it)
- mainEntityOfPage to the profile page

If you need a sanity check for why “rendered output” matters, not the presence of JS:

Avoid the two most common image mistakes

1) Multiple “official” photos across Tier‑1 profiles

If LinkedIn uses one photo, Instagram uses another, X uses another, and your site uses a fourth — you are asking Google to choose.

Make it the same portrait across Tier‑1. Save “creative variants” for content, not identity.

2) Using low-quality or heavily cropped thumbnails as the canonical candidate

If your only repeated image is a tiny crop from a video thumbnail or a screenshot, that becomes the “consensus candidate”.

Use a clean portrait with a neutral background for the identity layer.

A minimal, safe checklist (ship in one day)

Do not buy “panel image” services or spam image pins.
Do not spin up dozens of weak profiles and link them in sameAs.
Do not change the portrait every week.

Stability is a feature. Entity systems reward the lowest-entropy story.

Photo authority for Person entities (2026): how to make Google pick the right image

Share

Key takeaways

Table of Contents