Replacing paper field surveys with dynamic forms and a live operations view

The challenge

Surveys were carried out on paper and transcribed afterwards. That meant the data arrived days late, arrived with transcription errors, and could not be validated at the point it was collected — which is the only point where an error is cheap to fix.

By the time a problem in a survey was noticed, the team had left the site.

The context

The firm ran different survey types for different clients, and the forms changed regularly. Any system that hard-coded a form would need a development cycle for every new survey type, which would have made it slower than the paper it replaced.

Sites were frequently somewhere with no usable mobile signal.

The approach

The insight was that the forms were the product, not the application. So the application became a renderer, and the forms — fields, types, conditional logic and validation rules — became data defined and managed centrally.

That inverted the economics. Adding a survey type stopped being an engineering task and became something the operations team could do themselves.

The solution

A mobile application renders whatever form definition it has been given, validates responses against the rules attached to that definition at the point of capture, and stores the result locally.

A web application handles the other half: defining forms, allocating tasks to field teams, tracking what has been completed, and generating reports from the collected data. Field teams see their assignments; the office sees progress without ringing anyone.

AI & automation

None. This was 2016-era work, and it is a good illustration of a point I make often: a well-designed configurable system solved a problem that people now instinctively reach for AI to solve.

The requirement was structured data captured correctly at source. A validation rule does that perfectly and predictably. There was nothing here for a model to add.

Engineering

The dynamic form engine was the substantial piece of engineering — a definition format expressive enough for conditional fields and real validation rules, but constrained enough that a non-developer could not break the application with it.

Offline-first was not optional. Surveys had to be completed and validated with no connectivity and synchronised later, with conflict handling that never silently discarded a field team's work. Task allocation and progress reporting had to reflect a reality where a device might not have been heard from for hours.

Outcome

Paperwork was replaced by tablet capture, with errors caught while the team was still on site rather than after transcription.

The office gained a live view of field progress, and new survey types could be introduced without a development cycle.