Rob Rigby

Rebuilding the research process

Monthly output went from around 400 records to around 1,500, on the same staff hours.

The problem

The oldest records had not been updated for five years, in a product sold as a live, managed database. Work went out on paper to freelancers whose output could not be measured, and nobody could see how big the backlog was.

What I did

Replaced the paper forms with an online research form that checks entries as they are made. Made the case for employing researchers instead of using freelancers, then recruited and inducted them. Split the database into priority bands by company size, so large companies are researched four times a year and the smallest once, and built a research calendar around it.

Where AI helped

I used AI tools as an aid to build the research site. The decisions about priority bands and how often each is researched were mine, based on which errors customers notice.

Also

Printing and postage of about £500 a cycle stopped.

A checking platform that keeps people in charge

Checking moved from me alone, against a 50-page guide, to two senior researchers working a 3-page process.

The problem

Every record needed checking before it went live, and the only instructions were a 50-page manual that one person worked through by hand.

What I did

Designed a platform that takes a batch of around 100 records through five stages: data quality, comparing updates, refusals, companies that have gone, and completion. A person makes the decision at every stage; the platform tracks progress and makes sure nothing is skipped. A batch takes about an hour.

Where AI helped

AI tools helped me build it. The stages, and what is left to human judgement, came from how the checking actually worked.

Cashflow forecasting at short notice

A £20,000 shortfall found four months out, halved through supplier savings, and closed before it arrived.

The problem

The colleague who handled cashflow for two companies left at two weeks' notice, with little usable documentation. Nobody could say with confidence that staff and suppliers would be paid at month end.

What I did

Paused my other work, shared a first spreadsheet forecast within a day, then rebuilt it on the dates customers actually pay, working through twelve months of invoices and bank statements. That showed a gap caused by our largest customer moving to longer terms. Later I built a forecasting tool linked to our accounts software, including a fix for card payments that arrived in combined batches and were landing on the wrong dates.

Result

The directors now decide which suppliers to pay, and when, from evidence.

My own project: a job application tracker