The right tool is the one your team will update, that matches how you actually plan, and that someone will administer on a wet Wednesday. The wrong one is the platform that won a demo. Choose on constraints, not on feature count, and do not expect software to supply the discipline you skipped.
Fit the work, not the demo
Start from how work arrives and how you sequence it. A flow team living on a board needs a first-class kanban, WIP limits, and cheap card movement. A delivery with vendors, freezes, and a critical path needs dependencies and a timeline that is not a toy. A Scrum team needs a backlog that can be ordered this afternoon, not a Gantt with sprint stickers. If the tool fights the method, people will open a spreadsheet. The spreadsheet will win.
Team size changes the problem. Five people can live in a lightweight tracker and a shared doc. Fifty people need permissions, a shared taxonomy, and reporting that does not require a priest. Five hundred need administration as a real role. Buying an 'enterprise' suite for a team of six is how you spend two months configuring workflows nobody asked for. Buying a sticky-note app for a programme is how you lose the audit trail and then commission a second tool in a panic.
Integrations are part of fit. If engineering lives in one system and you invent another that will never sync, you have chosen dual entry. Dual entry lasts until the first busy week. Then your tool is decorative. Prefer the tool the doers already open, or a thin layer that reads from it. A PM tool that the team treats as 'your system' will be updated when you nag, which is to say: not enough.
Admin cost is the price you pretend is free
Every tool needs an owner: fields, permissions, the board template, the graveyard of old projects. If nobody is named, the first enthusiastic person becomes unpaid admin, then they leave, then the workflows rot. Estimate the hours before you buy. A cheaper licence with a day a week of admin is not cheaper. Put the name in the decision record.
Count the switching cost honestly. Export quality, training, parallel-run time, and the month where both tools are 'source of truth.' Moving house mid-project is a delivery risk. Sometimes it is still right — the current tool cannot do dependencies and you are about to run a migration. Say that. Do not switch because a senior person saw a product tour and felt a feeling.
Security, residency, and procurement are constraints, not afterthoughts. If your organisation cannot put customer data in a given cloud, the delightful startup tool is a hobby. Check that before you run a trial with real work. Trials with real work become production. Production without a data conversation becomes a conversation with legal, which is a ceremony nobody enjoys.
Decide, then stop shopping
Run a short trial with one real slice of work and the people who will click every day. Score a handful of must-haves you wrote down first: board or timeline, permissions, reporting, admin load, cost at your actual seat count. Do not score fifty nice-to-haves. Vendors are good at nice-to-haves. Nice-to-haves are how you buy a cockpit for a bicycle.
Write the decision: what you picked, what you rejected, and what you will not customise in year one. Custom fields multiply like rabbits and then nobody knows which 'priority' is the real one. Start with a default workflow and change it only when a weekly pain is obvious. Configuration is not progress. It is a project that pretends it is enabling the project.
Revisit only when the constraint changes — team size, method, or a compliance rule — not when a new logo appears. Tool-chasing is a way to avoid the harder work: a ranked backlog, an honest timeline, and a team that updates the board. Software will not attend the standup for you. If that is the feature you wanted, you wanted a different career problem.
Course complete — what's next
Choose for method, team size, integrations, and admin cost. Trial with real work, write the decision, and stop shopping. A tool the team will not update is a licence fee attached to a lie.
You have the Tools 101 spine: why tools exist, boards and backlogs, timelines that admit they are models, a single source of truth, and a selection approach that is not a vendor parade. Pick one improvement on your current stack this week — usually killing a shadow tracker — and let the official system become true by use.