Instacart's core image specs, in one table
Instacart's catalog documentation is precise about the file itself. Get these five things right and most rejections disappear before a human ever reviews the asset.
| Spec | Requirement | Why it matters |
|---|---|---|
| Dimensions | 1000 x 1000 px, square | Grid tiles are square; non-square files get cropped or rejected |
| Resolution | 300 PPI | Supports zoom on the product detail page |
| Background | Pure white, no props | Added colors or graphics are a common rejection reason |
| File format | JPG, JPEG, or PNG | Other formats are not ingested |
| File size | Under 3 MB | Oversized files fail at upload |
Two rules do more damage than the rest combined. First, the image must be perfectly in focus — soft product edges are grounds for rejection, and phone shots of a shelf are the usual culprit. Second, the photo must show the actual in-store product the customer receives. A refreshed package design that hasn't hit shelves yet is a mismatch, and mismatches drive refunds.
If your carton art changed six months ago but stores are still selling through old stock, the image that matches the shelf wins. Update the catalog when the old SKUs clear, not when the design is approved.
Why grocery images fail differently than Amazon images
An Amazon shopper browsing a listing gets seven images, a zoom viewer, and A+ content. An Instacart shopper gets a small square tile in a scrolling aisle, often on a phone, often while comparing four near-identical cartons. The image is doing nearly all the work.
Amazon-style thinking
- Hero image plus six supporting shots
- Text overlays and infographics carry the pitch
- Shopper reads the title and bullets
- Angled hero shots for visual drama
Grocery-shelf thinking
- One tile decides the click
- Package front is the infographic
- Shopper pattern-matches against the physical product
- Straight-on front-of-pack, no ambiguity
That difference explains a mistake I see constantly: CPG brands reuse the angled, styled hero from their Amazon listing. It looks better in isolation and performs worse in an aisle grid, because the front label is foreshortened and the shopper can't confirm it's the item they buy in-store.
Shooting for the grid: a practical setup
You do not need a studio to hit Instacart product image requirements. You need a repeatable setup that produces sharp, square, white-background frames of a package front.
- Straight-on, eye level with the label. Put the camera on a tripod at the vertical midpoint of the package. Any downward tilt introduces keystoning that makes the front panel look trapezoidal.
- Shoot at f/8 to f/11. Grocery packaging is deep — a cereal box is 3 inches front to back. Wide apertures leave the side edges soft, which reads as "blurry" to a reviewer.
- Two diffused lights at 45 degrees. Even, shadowless illumination on the front panel. Glossy cartons and foil pouches need larger, farther modifiers to kill hotspots on the barcode area.
- Frame with room to breathe. Shoot slightly wider than final crop so you can square-crop to 1000 x 1000 with the product filling roughly 85% of the frame.
- Lock white balance. Grocery packaging lives and dies on color — a shopper who expects the red carton and sees an orange one scrolls past.
Photograph one representative sample from each production run, not just the pre-production sample. Print color drifts between runs, and the shelf version is the version that has to match.
Multi-packs and variety packs
A 12-pack should be photographed as the shopper receives it: the outer case or shrink-wrapped bundle, front facing, with the count clearly visible. Photographing a single can when the SKU is a 12-pack is one of the fastest routes to a customer complaint and a refund — and refunds on grocery marketplaces are attributed straight back to catalog quality.
Fixing a legacy catalog without reshooting everything
Most brands arriving on Instacart don't have a clean shoot problem — they have a backlog problem. A few hundred SKUs exist as mixed-quality JPEGs collected over a decade: some 600px, some on grey seamless, some with a marketing badge burned into the corner.
Triage before you reshoot. Sort every image into three buckets:
| Bucket | Condition | Action |
|---|---|---|
| Recoverable | Sharp, correct packaging, wrong background or size | Background cleanup, square recrop, upscale |
| Borderline | Sharp but heavily shadowed or color-shifted | Relight and color-correct; reshoot if it fails review twice |
| Dead | Blurry, obsolete packaging, or missing entirely | Reshoot — no amount of editing fixes stale artwork |
The recoverable bucket is usually the largest, and it's where AI editing earns its keep. Background removal, white-background compositing, square recropping, and resolution upscaling are exactly the mechanical, high-volume operations that AI tools handle well — this is the kind of catalog cleanup Retouchable was built for. What AI cannot do is invent packaging that doesn't exist yet, which is why the dead bucket stays a camera job.
A pre-submission checklist
Run every asset through this before it goes into a catalog inventory file. It takes about thirty seconds per SKU and saves the multi-week loop of rejection, diagnosis, and resubmission.
- Exactly 1000 x 1000 px or larger, and perfectly square
- Pure white background with no gradient, shadow spill, or prop
- JPG, JPEG, or PNG, under 3 MB
- Sharp from edge to edge — zoom to 100% and check the package corners
- Packaging matches what is physically on the shelf today
- Pack count and net weight legible without zooming
- No added text, badges, promotional callouts, or borders
- Color verified against a physical sample under neutral light
- Product fills roughly 85% of the frame, centered
- File named to the SKU or UPC so catalog mapping doesn't drift
View the final tile at about 150 px wide on a phone. That's roughly how it appears in an aisle grid. If you can't identify the flavor or size at that scale, the image isn't done — usually the fix is a tighter crop, not a better photo.