How research-led design doubled supply from key customer segments

The most impactful thing we did for Newsflare's highest-value contributors wasn't redesigning the upload experience. It was eliminating the need to upload at all. For everyone else, we consolidated five fragmented flows into one.

Role Design lead & product lead

Collaborators Engineering, Video Desk, Sales, Support

When Q2 2022

Newsflare analytics and A/B test results alongside mobile upload UI

Context

Newsflare is a video marketplace connecting videographers and members of the public to media buyers across news, entertainment, and advertising. At the time I was both Product and Design Lead, working across design, engineering, customer-facing teams, and senior leadership.

Buyer demand was strong. The supply side wasn't keeping up - particularly for a small number of high-value contributors joining at a growing rate. Five separate upload flows existed within the portal, generating inconsistency and significant technical debt.

Architecture diagram showing the five separate upload entry points: Guest Homepage, Mobile App Upload, Main Upload, Video Brief, Consent Form, and Upload Legacy/China — The five upload flows that existed when the project began - each a separate journey with its own technical stack, creating inconsistency for users and compounding debt for the engineering team.

The problem

Five separate upload flows generated inconsistency and compounding tech debt. 90% of uploads came from mobile but the experience was desktop-first. Contributors had no visibility into what happened after they uploaded. Many described it as throwing videos into a black hole. And Newsflare had never formally defined its core contributor segments or how their needs differed.

Research and segmentation

I spoke directly with users through interviews, focus groups, surveys, and contextualised their attitudes through platform analytics and sales rate data. A research session with contributors was used to surface attitudes about the platform, identify pain points, and gather sentiment around the upload experience.

In-person research session with a contributor presenting findings on a wall of sticky notes and printed screenshots — In-person research with contributors - gathering attitudes about the upload experience and identifying pain points that analytics alone couldn't surface.

Four filmer archetypes emerged from the research:

Chancers - opportunistic earnings, low upload volume
Chasers - reliable income, high upload volume
Creators - audience and platform growth, moderate volume
Content Partners - sales and syndication, very high volume

Stacked bar charts showing Member Types (Creator, Chaser, Chancer) breakdown by Count Users, Count Videos, and Saleables percentage over time — Segment data showing the breakdown of user types, video volume, and - critically - percentage of saleable video by segment. Chasers dominated saleable supply despite being outnumbered by Chancers.

A product-market fit survey across each segment found two groups breaching the 40% threshold: Chasers and Content Partners. Chasers produced roughly 65–70% of all saleable video. Content Partners contributed the overwhelming majority of the remainder.

PMF survey results for the Chancer segment showing responses across what type of member would benefit, main benefits, how to improve, and other feedback — PMF survey results for the Chancer segment - one of four archetypes surveyed. The contrast between Chancers (below threshold) and Chasers (above) shaped every subsequent design decision.

Many Chasers and Content Partners were filming significantly more footage than they uploaded - submitting only what they thought would sell. Because the upload process was effortful, they were leaving money on the table. Until the perceived cost of supplying a video dropped, supply would remain constrained.

"I like to shoot proper footage, interviews with good audio, sometimes A and B roll. The effort that goes into that can feel wasted when, for the same price, someone's wonky phone footage that's carelessly shot gets sold simply because it's first past the post."
Newsflare Chaser, user interview

Solution - Content Partners: MRSS pipeline

Through deeper conversations with Content Partners, I discovered many hosted video on private systems that generated MRSS feeds: structured data files that could be scraped automatically to import entire video libraries without any manual upload from the partner. Working closely with engineering and a key Content Partner, I built a proof-of-concept that did exactly that - a first for the company.

Raw MRSS XML feed from a Content Partner showing video metadata, titles, descriptions, categories and Amazon S3 media URLs — The MRSS feed from the first Content Partner we onboarded - a structured XML file we could scrape automatically to ingest entire video libraries without any manual upload effort from the partner.

Content Partner MRSS Upload Workflow swimlane diagram showing Newsdesk Actions, Content Partner Actions, and Platform Actions for processing MRSS feeds — The MRSS ingestion workflow - mapping what the Content Partner needed to do once (add video to their feed), what the platform did automatically, and how Newsdesk handled edge cases like category mismatches.

The result: a near-doubling of video received from Content Partners overnight, from a one-time setup. The architecture was then scaled to all similar partners.

Solution - Chasers: consolidated upload flow

With the MRSS pipeline underway, I mapped all five existing upload journeys, identified opportunities to redirect users toward a single core funnel, and eliminated duplication. A series of A/B tests on the consolidated flow informed the final design direction.

Supply flow swimlane diagram showing the outbound consent form process with Staff Actions and Contributor Actions lanes — The outbound consent form flow - one of the fragmented sub-journeys mapped and consolidated as part of the architecture work.

Supply flow swimlane for video scoring and green queue showing how videos are scored by the Trust Algorithm and routed to Green, Amber or Red queues — The video scoring and queue flow - understanding how uploaded videos moved through the platform after submission helped identify where feedback loops were missing.

Google Optimize A/B test results table showing experiment sessions, conversions, calculated conversion rates, and modelled improvement ranges for eight variants — A/B test results from the upload flow experiments - the winning variant (A1-B0-C1) showed a modelled conversion rate of 7.7–11.7%, with 99% probability of beating the original.

Line chart showing CSAT scores trend from January to May with daily data points — CSAT tracking across the project period - monitoring sentiment as changes were rolled out to ensure improvements didn't come at a cost to user satisfaction.

The redesigned upload experience was built mobile-first: chunked inputs, removed non-critical fields, and clear progress states bookending the journey.

Mobile upload screen showing 'Your video reaches further with Newsflare' hero, Show the world your video section, and Upload a video now CTA — The mobile upload entry point - restructured to lead with value (reach) and reduce cognitive load before the user had committed any effort.

Mobile upload in progress screen showing 64% circular progress indicator, source file details, and files uploading list — Upload in progress - a clear progress indicator and file status replacing the previous "black hole" experience that left users uncertain whether anything was happening.

Video uploaded success screen showing celebration state with What's next section and Upload video CTA — Upload success - a moment of acknowledgement and encouragement, with an immediate path to upload again. Small changes to the emotional register at key moments.

Supply flow diagram for the consolidated upload journey showing Pre-Upload and Upload phases with Contributor and Platform Actions — The consolidated upload flow - five journeys reduced to one. A single entry point, shared required fields, and a clean platform handoff replacing the fragmented architecture.

This became the only upload flow the company needed.

Outcomes

Stacked bar chart showing Uploads by Partner growth over time, doubling from approximately 1,000 to 2,500-3,000 monthly uploads — Uploads by partner over the project period - the step-change visible in the data corresponds to the MRSS pipeline going live for Content Partners.

+50% Increase in Chaser uploads

>200% Increase in Content Partners uploads

+8/10 Consistent CSAT in usability testing

5 to 1 Upload journeys consolidated

Reflections

The MRSS pipeline was the most impactful thing we did, and it emerged entirely from talking to Content Partners in depth rather than assuming their problem was the same as Chasers'. The architecture solution - scraping rather than uploading - wouldn't have been visible without that conversation.

One test we ran prototyped a one-to-many upload model, allowing multiple videos to share a single metadata set. This produced a 7× increase in uploads during testing and pointed to directions worth pursuing: multi-session publishing, agent-based uploads via messaging platforms, and dynamic metadata by location.

The broader lesson: success came from segmentation and a willingness to build distinct architecture for the highest-performing cohorts, rather than treating all users the same. The two solutions - MRSS pipeline and consolidated upload flow - look completely different because the underlying user needs were completely different.

← Onboarding All work →