In late 2024, I redesigned Motorway's document-verification flow, improving first-time approval rates by 7% and reducing average verification time by 3 hours. With no PM or backend support, I delivered a lightweight, content-led solution that reduced operational load and strengthened seller trust.

Document verification was one of the highest-cost steps in Motorway's seller journey. As the business scaled and invested in machine-vision systems, our small team had no backend support and no dedicated product manager. Failures were common, guidance was unclear, and sellers blamed Motorway when documents were rejected.

Rather than wait for engineering capacity, I led a fast, design-led sprint that reframed verification as a guided, confidence-building experience.

I started analysing submitted photos and most common failure reasons, and prioritised open questions I needed answering, all this identified recurring failure patterns and existing technical and operational constraints:

1. Sellers were hiding sensitive data, making verification impossible.
2. Low-quality, cropped, dark photos and missing pages led to frequent rejections.
3. Mismatch details (e.g names not aligning with records).

I framed the work around one question: What are the smallest, most impactful changes we can make to improve the upload experience?

I saw an opportunity to drive quick improvements through content-led solutions. This approach allowed us to refine clarity, guidance, and usability with minimal FE engineering effort. I looped in our UX content designer and together, addressed the user needs.

One key improvement was ensuring photo tips were discoverable. Previously, sellers were forced to dismiss a persistent modal every time they added a photo (up to four times for core documents alone). I replaced this with a tile that remained visible as a reference point. On mobile, I simplified the UI further, using a bottom sheet that only appeared on the first upload. These changes mattered because:

1. I empowered sellers: Keeping photo tips always visible allowed sellers to cross-reference guidance if their photo failed, rather than relying on memory.
2. I reduced friction: Instead of blocking sellers with repeated modals, we surfaced help only when needed, reducing friction.
3. We improved clarity and scalability: by ensuring consistency across UI, help centre, and support.

Alongside key UX improvements, myself and a contend designer developed a scalable guidance framework and content structure. We clarified what does is needed, why this is required, and what are the specifics we need (sometimes this could be things like rules around expiration date.

This UI pattern was quickly adopted by other product teams, ensuring consistency across the product experience, help centre articles, and phone/email support. As a result, sellers received clear, aligned guidance on what we needed to see—and why it mattered.

Lastly, informed by seller feedback and support CS insights, I replaced generic contact links with a support tile featuring the 3 most frequent seller concerns from support logs, paired with detailed help centre articles. Additionally I added a universally recognised lock icon to the submission button, aligning with seller expectations around document privacy.

The redesign proved that lightweight UX and content changes could deliver material business value, influencing how future ops-led projects were approached.