Bad product photos never reach your store,
the agent catches them first
A catalogue with a few hundred SKUs usually has photos shot by different people, on different days, with different lighting, and nobody has time to open every image before it goes live. We build an agent that scores every incoming photo against your own rules, blur, exposure, background, crop, missing angles, and sends anything that fails straight back with the reason attached, instead of letting it reach the storefront.
The process today
Product photos arrive from in-house shoots, outside photographers and sometimes suppliers, and someone on the catalogue team has to open each one, check it against the brand’s style guide, and either approve it or send it back with notes. On a few dozen photos a day this is manageable; on a few hundred during a seasonal push it becomes the thing that delays launch, because the bottleneck is a person’s time to look at every image, not the photography itself.
The second cost is inconsistency. One reviewer is strict about background color and lighting, another lets things through under deadline pressure, and the result is a storefront where some product pages look professional and others look like they were shot on a different day by a different team, because they were.
The third is the delay loop with external photographers: a batch gets rejected days after it was shot, by which time the set is gone, the model has left, or the season has moved on, and the business either reshoots at extra cost or ships photos it knows are below standard.
None of this shows up as one dramatic failure. It shows up as a steady drag: product photo quality check work that should take minutes stretching into a backlog item, a quality bar that holds on a quiet week and slips on a busy one, and a team that knows the fix is mechanical but never has a free afternoon to build it themselves.
What the agent does
The agent watches your upload folder, PIM or DAM for new product photos and scores each one against a checklist built from your style guide: blur and sharpness, exposure, background color and cleanliness, crop and framing, and whether the required set of angles is present for that product type. Photos that pass move straight into your catalogue pipeline; photos that fail are held back and returned with the specific reason attached, not a generic rejection.
For categories with strict visual rules, apparel on a model at a fixed angle, electronics against a pure white background, food shot under consistent lighting, the checklist can be as strict or as lenient as your team decides, and different categories can run different rules. The agent also flags near-duplicates and photos that do not match the SKU they were uploaded against, which is a common source of catalogue errors that nobody catches until a customer complains.
Typical integrations: a shared upload folder, Shopify or your PIM for where approved photos land, and Slack or Telegram for the daily batch of rejections sent back to whoever shot them.
What stays with humans
Deciding whether a borderline photo is good enough to ship is a judgment call your team keeps, especially for anything stylistic rather than technical, is the mood right, does this angle sell the product. The agent catches technical failures reliably; it does not have the final say on brand taste, and any photo it is unsure about goes to a human reviewer rather than being auto-approved or auto-rejected on a close call.
Guards
Every photo scored is logged with the checklist result, so you can audit why a batch was approved or rejected weeks later. Thresholds start conservative, meaning more borderline photos go to human review than will eventually be needed, and are only loosened once we can show the checklist matches your team’s own judgment on a sample batch. A kill switch pauses auto-rejection in one message if a new product type is producing false rejects.
Before it runs unattended, we run a side-by-side dry run against a sample of your own product photo quality check material so your team can see exactly what it would have done. Every build ships with a short written runbook so your team can pause it, adjust a threshold, or roll it back without waiting on us, and the running-cost estimate below is a starting budget you set, with an alert built in before it is crossed.
Price and timeline
| Option | Price | What it covers | Timeline |
|---|---|---|---|
| Single automation | from $500 | Quality scoring on blur, exposure, crop and background | 3 to 8 days |
| Department package | from $2,500 | photo QA, background removal and packshots across your catalogue team | 2 to 4 weeks |
Running cost is usually $10 to $80 a month in model usage depending on volume, with a budget cap set before launch.
Related
Pair this with background removal and packshots at scale so approved photos go straight into a clean, consistent packshot, and with AI photo retouch for e-commerce for the final polish before publishing. For the image-generation side of product cards, see image generation for product cards. The full package breakdown is on the AI agents service page and the automation-everything overview; for a real build of a product-card pipeline, see the AI product card designer bot case study and the Thailand D2C store rebuild case study.
Ready to stop opening every product photo by hand? Get in touch and we will build the checklist from your style guide in the first call.
Tired of doing this by hand? We can take the whole routine off your team, not just this step: Routine takeover, from $400 →
FAQ
How much does product photo quality checking cost?
from $500 for one product category and one scoring checklist, live in 3 to 8 days. A department package covering photo QA, background removal and packshots together starts at $2,500.
How does the agent know what a good photo looks like for us?
We build the checklist from your existing style guide and a sample of your best photos; it scores blur, exposure, crop, background and required angles against that, not a generic standard.
Does it work with photos from outside agencies and freelancers?
Yes, that is the most common use case, since quality is hardest to control when the photographer is not in-house. The agent scores whatever lands in the upload folder regardless of who shot it.
What happens to a photo that fails the check?
It is held back from publishing and returned with the specific reason, wrong background, missing angle, too dark, so the photographer or agency can fix exactly that instead of guessing.
Can it also fix minor issues instead of just rejecting?
Simple fixes like exposure and crop can be auto-corrected if you want that; background removal and packshot generation is available as a separate automation if you need the full pipeline.