Mark Meacher
Home
CASE STUDY
Multi Image Upload
Speeding up the car listing process
Year: 2026
Category: Seller experience
AI Assisted
Overview
Sellers listing a car on Carwow were adding photos one at a time. Slow, and a common point of drop-off before a listing was even complete. In a survey, sellers told us directly they wanted to add photos in bulk rather than individually. I designed on a batch upload flow with AI photo tagging, letting sellers select every shot at once and have the system sort each one into the right slot automatically.
User Testing
I tested the Batch Uploads MVP with an unmoderated Maze study across 10 sellers. The core flow scored well: multi-photo upload rated 4.4/5, confidence photos landed correctly rated 4.3/5, and clarity on what still needed uploading hit 4.7/5, the highest score in the study. Testers repeatedly praised the AI auto-sorting ("you could really feel the difference the AI-powered magical matching brought") and the reference diagrams for each photo slot.
The wheel & tyre step was the one clear blind spot, two of ten testers couldn't complete it, though 80% said they'd still add those photos here given the choice. After investigation, this was due to a small UI inconsistency that was preventing users finding the option and was fixed for laucnh. Testers also flagged that the mobile layout hides key controls below the fold, and asked for a confirmation moment after editing a photo. Several had nothing to add at all, calling the experience "already super user friendly" — the flow is close to launch-ready once the wheel & tyre step and mobile layout are fixed.
Approach
I replaced the one-photo-at-a-time uploader with a single "select all photos" action letting sellers see exactly what's needed, required exterior and interior shots, plus optional wheel and tyre angles, then pick them all from their camera roll in one go. AI tagging automatically sorts each photo into the right slot, with a review screen that shows progress and nudges sellers toward the optional shots that strengthen a listing. Before building, I validated the concept with a fake-door test, unmoderated usability testing, and an in-product survey, nearly a third of sellers said batch uploading was the single most helpful change we could make to the photo step.
WORKING WITH CLAUDE AI
I also used Claude as a working partner to support with prototyping and test setup. For usability testing, I worked with Claude to map out what to test on Maze against each block type and then ran the test analysis through Claude to spot key trends and map critcial launch blockers.
Next Project
Mark Meacher
mark.meacher@gmail.com
Mark Meacher
Home
CASE STUDY
Multi Image Upload
Speeding up the car listing process
AI Assisted
Year: 2026
Category: Seller experience
Overview
Sellers listing a car on Carwow were adding photos one at a time. Slow, and a common point of drop-off before a listing was even complete. In a survey, sellers told us directly they wanted to add photos in bulk rather than individually. I designed on a batch upload flow with AI photo tagging, letting sellers select every shot at once and have the system sort each one into the right slot automatically.
User Testing
I tested the Batch Uploads MVP with an unmoderated Maze study across 10 sellers. The core flow scored well: multi-photo upload rated 4.4/5, confidence photos landed correctly rated 4.3/5, and clarity on what still needed uploading hit 4.7/5, the highest score in the study. Testers repeatedly praised the AI auto-sorting ("you could really feel the difference the AI-powered magical matching brought") and the reference diagrams for each photo slot.
The wheel & tyre step was the one clear blind spot, two of ten testers couldn't complete it, though 80% said they'd still add those photos here given the choice. After investigation, this was due to a small UI inconsistency that was preventing users finding the option and was fixed for launch. Several testers had nothing to add at all, calling the experience "already super user friendly" — the flow is close to launch-ready once the wheel & tyre step and mobile layout are fixed.
Approach
I replaced the one-photo-at-a-time uploader with a single "select all photos" action letting sellers see exactly what's needed, required exterior and interior shots, plus optional wheel and tyre angles, then pick them all from their camera roll in one go. AI tagging automatically sorts each photo into the right slot, with a review screen that shows progress and nudges sellers toward the optional shots that strengthen a listing. Before building, I validated the concept with a fake-door test, unmoderated usability testing, and an in-product survey, nearly a third of sellers said batch uploading was the single most helpful change we could make to the photo step.
WORKING WITH CLAUDE AI
I also used Claude as a working partner to support with prototyping and test setup. For usability testing, I worked with Claude to map out what to test on Maze against each block type and then ran the test analysis through Claude to spot key trends and map any critical launch blockers.
Next Project
Mark Meacher
mark.meacher@gmail.com