Photos are the part of a Google Business Profile most likely to be handled by whoever happens to have a phone. They are also the part a machine now reads in detail. Understanding what actually gets extracted from an image changes what you ask a client to shoot, and it kills a few habits that have survived far longer than they should have.
How Google reads a photo
Image recognition on a profile photo produces labels: objects, scene type, whether there are people, whether text appears, rough quality and composition. Those labels are what let Maps show a "food" carousel, group interior shots together, or surface a photo of a specific dish when someone asks about it. Lens searches run against the same understanding.
What this means in practice is that the subject of the image is doing the work. A wide shot of a shopfront with a car in the way is understood as a street scene. A clean, well-lit photo of the treatment room is understood as a treatment room. Neither is captioned by you, so composition is the only lever you have.
The photo categories most profiles ignore
Every profile has a logo, a cover image and a pile of whatever was to hand. The category slots that go unfilled are the ones that answer real questions:
- Exterior - shot from the direction people approach from, with the entrance and any signage readable. This is the photo that stops someone walking past the door.
- Interior - the space as it looks on a normal day, not empty at 7am. People are checking whether they will feel comfortable.
- At work - staff doing the actual job. This is the strongest category for trades and services and the emptiest one on most profiles.
- Team - faces, named where possible. It matters more in categories where a stranger enters someone's home.
- Product or menu - individual items shot separately rather than one photo of a whole counter, because item-level images can surface on their own.
- Common areas and parking - underrated. "Where do I park" is a question the photo library can answer.
Customer photos fill these gaps eventually, and they carry weight precisely because the business did not control them. That is an argument for making the place photogenic, not for trying to suppress them.
File naming, size and metadata that survive upload
Most image optimisation advice is written for website images and copied across to profiles, where it does not apply. Here is what survives the upload and what does not.
| Attribute | Practical guidance |
|---|---|
| Format | JPG or PNG. JPG for photography, PNG only where flat graphics are involved. |
| Dimensions | 720 x 720 pixels is the sensible floor. Anything from a modern phone camera is already comfortably above it. |
| File size | Keep under 5MB. Heavy compression to squeeze files smaller is counterproductive; sharpness affects perceived quality. |
| Aspect ratio | Shoot landscape for cover and exterior, square or portrait for products. Crops get applied differently across surfaces. |
| File name | Descriptive names cost nothing and help your own library management. Do not expect ranking value from them. |
| EXIF and geotags | Stripped. Ignore. |
Shooting for search, not for the brochure
Brand photography and profile photography are different jobs. Brochure images are styled, empty, often heavily graded, and frequently carry text overlays. Those are exactly the images that read poorly here, and promotional text on an image can get it removed under Google's photo guidelines.
Plan the shot list before anyone picks up a camera
Write down the questions customers actually ask - where to park, what the waiting area looks like, who will turn up, what the work looks like finished. One image per question. Twenty deliberate photos beat two hundred random ones.
Shoot in daylight, no flash, camera level
Natural light through a window, phone held straight rather than tilted up, and nothing important in shadow. Turn on the grid overlay. Level horizons are the single biggest difference between a photo that reads as professional and one that does not.
Fill the frame with one clear subject
One dish, one room, one person working. Cluttered wide shots are hard to label and hard for a customer to read at thumbnail size, which is where most of them are first seen.
Edit lightly, then leave it alone
Straighten, crop, lift exposure if needed. Skip filters, borders, logos and price stickers. If an image needs text to make sense, it is the wrong image.
A photo cadence you can actually keep
The common failure is a hundred images uploaded on day one and nothing after. Profiles behave better with a steady trickle, and it is easier to sustain than a quarterly photo day nobody books.
A workable rhythm for a service business is two or three images a month, drawn from work that was happening anyway. Give the client a standing instruction: photograph the finished job, in daylight, before you pack up. For retail and hospitality, tie it to whatever already changes - new stock, seasonal menu, a refit. The point is that the library keeps acquiring recent dated material without anyone scheduling a shoot.
Handling this across many locations
The operational problem with multi-location photo work is that the good images live on individual phones and never reach the profile. A shared media library solves more of this than any optimisation tactic. In Ampli5 Pulse, images sit in one library, get pushed to the right profiles from the multi-location dashboard, and appear in the white-label report alongside performance insights, so a franchise manager can see which sites are actually contributing images and which have not uploaded anything since onboarding. Whether photo activity moved anything shows up in views, clicks and direction requests rather than in any keyword figure.