Case Studies

Public videos are rich in information, but difficult to search. A video is not a spreadsheet: useful data may exist in the frames, audio, metadata, captions and visual context, but unless that information is extracted and structured, it remains hidden.

Fishplate applies this problem to UK railway videos. Enthusiasts, preservation groups and vehicle owners often want to find footage of specific locomotives, units, locations or journeys, but normal search only works when the uploader has described the video accurately.

Fishplate uses transcription, OCR, visual detection and railway reference data to turn scattered video evidence into structured, searchable records. The result is a dataset that can connect videos to locations, dates, operators, locomotive numbers, headcodes and even related journey movements.

This case study looks at how we built a fast, simple compliance tool for a client in the affiliate marketing industry. Facing strict new regulations, the client had an urgent problem: their ads were appearing on unknown websites with inconsistent or incorrect messaging. This created a major risk to their brand and their legal licenses.

The solution focused on answering one key question: Which external websites are actually sending traffic to the client? Rather than using slow manual searches or complex data integrations, we developed a "lightweight" JavaScript collector. This tool captures referral data in real time without slowing down the site or affecting the user experience.

We walk through the technical architecture and the trade-offs made to balance speed, privacy, and cost. You’ll see how a quick fix for regulatory compliance evolved into a powerful monitoring system—helping the client discover new traffic sources, prioritize their reviews, and stay ahead of legal risks.