← Back to work
Case study

Image management system

An automated system that keeps campaign imagery in step with live product availability, without anyone having to check.

Manual checking removed from every seasonal campaign refresh

The situation

Category and campaign images often showed several products together in one lifestyle scene, and those images ran across the site. They were made with the products available at the time. As ranges sold through, individual products inside an otherwise good image became unavailable or were discontinued.

What was missing

There was no reliable link between the products in an image and live availability. Finding a problem meant someone recognising the product, finding every image it appeared in, and deciding whether each one needed changing. Usually only one item in a composition was the problem, but nothing flagged it, so it could stay live until someone happened to notice.

My role

I identified this as a problem worth solving properly rather than working around indefinitely, and designed the automated system that replaced the manual check: the business rules, the exceptions that still needed a human decision, and the internal view for spotting anything the system flagged.

What got written down

Two things: the link itself, connecting each product in an image to its live product record, and the rules for what happens when that product becomes unavailable. The image is flagged and the item identified. The team can then update the composition by hand, swap in an approved alternative, or use AI-assisted editing to replace the product within the existing image.

The rules the system worked from

  • If a product shown within an image becomes unavailable, flag the image and identify the affected item.
  • Where an approved replacement exists, allow the product to be substituted within the existing composition.
  • Where no suitable replacement exists, send the image for manual review.
  • Do not trigger a replacement for temporary stock dips of less than 48 hours.
  • Treat discontinued products differently from temporarily unavailable products.

Automated image lifecycle

Campaign imagery updates itself

One flagged product, one image, no manual check

  1. 1

    Category / lifestyle image

    One image used across the site

  3. 2

    Availability change detected

    Live product availability monitoring

  5. 3

    Action

    Replace the unavailable product within the existing composition

    Manual update Merchandiser selects a suitable replacement and updates
    or
    AI replace in situ AI suggests and swaps the best fit automatically, preserving the scene

    Oversight rule: if no suitable replacement exists, the image is sent for manual review.

  7. 4

    Updated image

    Same composition, one product swapped

What changed

The manual checking step disappeared from the seasonal campaign process entirely. Maintenance overhead dropped, and what customers actually saw became more accurate, especially at the biggest seasonal peaks, when availability changed fastest and the old manual process was under the most strain.

It's the same underlying problem Batch blindness names for AI-generated imagery: content that's individually fine and collectively wrong, invisible until something is actually checking for it.

The check used to depend on someone remembering. Now it depends on a record.

---
title: Image management system
slug: image-management-system
updated: 2026-10-03
summary: An automated system that keeps campaign imagery in step with live product
  availability, without anyone having to check.
stat: Manual checking removed
stat_label: from every seasonal campaign refresh
---

If a problem like this sounds familiar, the audit is where I'd start: one workflow, one 90-minute session, and a written brief on what's missing and what to fix first.

Book the audit

Prefer to talk first? Get in touch — no obligation.