Beyond understanding technical literacy levels, I needed to map the actual workflows farmers used to record their operations. The workshop revealed fragmented, often analogue processes for tracking farm data. Some kept paper notebooks tucked in tractor cabs, jotting down seed varieties and planting dates. Others maintained spreadsheets, manually entering yield data after harvest. Soil sampling results arrived as PDF reports, filed away and rarely referenced again.

What struck me wasn't the lack of data, it was the lack of connection between data points. Farmers had information about their soils from testing labs, crop performance figures from harvest records, and field management notes scattered across formats. But nothing linked these together. They couldn't easily compare how different soil types affected yields, or which varieties performed best on their specific land. The data existed in isolation.

This insight shaped the platform's core value proposition: not just collecting data, but connecting it in ways that made individual records meaningful. I designed the information architecture around these actual farming workflows, ensuring the system could accept data in whatever format farmers already had it whilst automatically creating the relationships between soil characteristics, crop choices, and yield outcomes that had previously required hours of manual analysis.

Wireframing revealed a systems design challenge: the platform needed data from farmers to provide value to farmers, creating a chicken-and-egg problem. How do you provide valuable comparisons without existing data? This was the central tension in creating a national data infrastructure. The vision was integration and interoperability of software tools and datasets across UK agriculture, providing consistent, national-scale analytics. But you can't analyse what doesn't exist yet.

The solution required solving the cold start problem with external data sources. Working with the data science teams at Reading and Rothamsted, I integrated open source soil APIs that provided baseline field characteristics without requiring farmer input. This meant new users landed in a platform already populated with actionable data about their fields. They could explore soil variations across their acreage immediately, creating value before entering a single crop record.

These APIs became the foundation layer upon which farmers built their proprietary data. Instead of staring at empty forms demanding information, they saw their fields visualised with real environmental context. This approach transformed the first-use experience from "I need to do work before this is useful" to "This already knows something valuable about my farm." Strategically, it also meant I could start delivering on the promise of Big Data insights from day one, even with a small pilot cohort.

I designed the information architecture to provide immediate individual value before requiring participation in the collective benefit. This sequencing was critical: farmers would invest in data entry for personal ROI before contributing to the network. But farmers were protective of their proprietary yield data, commercial information they'd never shared before. Rather than hiding anonymisation in terms of service, I made it a visible UI element in comparison views, showing farmers exactly what others could and couldn't see about their data. This transparency became a trust-building feature that addressed the primary adoption concern.

These architectural decisions weren't just about interface layout; they solved the core business model challenge of building a multi-sided platform that needed to serve individual farmers, national policy objectives, and the broader research community simultaneously.