A top product problem
40% of all products in our ecommerce app and point of sale system had no photo - making it hard for customers and branch staff to identify the correct item. This was consistently one of our top two complaints in NPS surveys for over two years.
This issue directly impacted product findability and customer trust. Solving it was a clear opportunity to improve our customer experience and reduce a known friction point in both digital and physical environments.
A two-pronged approach
Declutter the app
The PM and I analysed product purchase history and received buy-in to hide any product with no sales in 12+ months. This cleared 72,000 irrelevant items from the customer experience, many which had no photo.
Crowdsource missing photos
We added an image-capture feature to a branch-facing iOS app already used for stocktaking. We called it Cyclepic. Staff could take photos of in-stock items using a branch iPhone.
iOS tools removed the photo background behind products
Photos were routed to our Digital Assets team for review
Approved photos were uploaded to our Product Information System and then immediately surfaced in the app and POS system
How might we ensure the most relevant products have images, so that more users can identify products?
Validation and iteration
I tested a Figma prototype on an iPhone directly with five branch staff, from drivers to counter workers.
All users completed the task easily, but skipped the instructional text. I replaced the text with visual cues and added error handling for invalid submissions.
Based on feedback, I also added a small 'moment of delight' to the confirmation screen to motivate frequent participation
One major simplification: I removed an unnecessary review screen after discovering iOS could auto-remove photo backgrounds
Challenges
Getting buy-in from stakeholders
The success of Cyclepic hinged on our Digital Assets team reviewing and uploading branch-submitted photos. But they were short-staffed and concerned about time, effort, and photo quality.
We ran a pilot with one branch: 5 photos taken with an iPhone. Quality was acceptable. iOS tools removed the background automatically. We preserved file naming conventions to match their current process. Outcome: no more effort than working with photographers - and full buy-in.
Uncertain branch engagement
Testing showed enthusiasm. I added a 'moment
of delight' confirmation screen to encourage continued use.
Risk of inappropriate submissions
iOS nudity detection helped safeguard content flow before human review. (Ask me about this one!)
Impact
The first part of the solution increased photo coverage from 60% to 78% in one go, just by hiding irrelevant products, most of which had no photo.
We have defined clear success metrics the Cyclepic part, and it is so far tracking well.
Primary goal: increase product photo coverage from 78% to 85% within 4 months.
Supporting indicators: NPS findability score to increase from 3.8 to 4.2.
