The project guide

From “we need a better way” to a useful first release.

A practical guide to preparing a software project. Bring your knowledge of the business; the technical specification can come later.

Before the first conversation

Describe a real working day.

One recent example is more useful than a long wish list. Remove personal or confidential information from anything you share.

01 / THE TRIGGER

Where does the work begin?

A new enquiry, a booking, an uploaded file or an internal request? Follow it through to completion and list the people and tools involved.

02 / THE FRICTION

Where does it slow down?

Show where someone copies data, checks a rule, chases approval or fixes a mistake. Include what happens when the normal process fails.

03 / THE OUTCOME

What would better look like?

Choose a concrete result and how you will measure it. Separate essential changes from ideas that can wait until the next release.

Choose the right route

Configure, connect or build?

Configure

Best when a standard product already supports most of the process.

  • Review unused features
  • Simplify forms and permissions
  • Improve training and reporting

Connect

Best when the tools work well but the hand-offs do not.

  • Check API availability and cost
  • Choose the source of truth
  • Plan for failed or duplicate updates

Build

Best when important business rules or journeys cannot fit comfortably into existing products.

  • Define a focused first release
  • Confirm who owns the data
  • Budget for operating the system
Before approving a proposal

Six details worth getting clear.

Scope & acceptance

Which workflows are included? What concrete checks demonstrate that each one works? Record exclusions and how changes are priced.

People & decisions

Name a business owner and representative users. Agree who approves changes and when feedback is needed.

Data & migration

List source files, record volumes and cleanup needs. Agree validation, cutover and a fallback plan.

External dependencies

Confirm provider accounts, API access, usage limits and recurring charges. Identify anything another supplier must deliver.

Ownership & access

Set out source-code rights, licence restrictions, account ownership and export arrangements in writing.

Life after launch

Confirm hosting, backups, support hours, response targets and the boundary between bug fixes and new features.

Useful perspective

Plan around uncertainty.

How should we think about timing?

Development effort is only one part of elapsed time. Feedback, provider access, data preparation and testing also affect the launch date. Agree milestones and dependencies before treating a date as committed.

What if the existing data is messy?

Start with a small representative sample. Agree required fields, duplicate handling and which historic records are worth migrating. Keep an untouched source copy for reconciliation.

Where can AI help?

Consider bounded tasks such as drafting, categorisation or extraction. Define review rules, acceptable errors and what happens when the output is uncertain before including it in a critical workflow.

Start with one workflow.

The planner helps turn the problem into a sensible starting scope.

Open the project planner →