The technology is the easy part. Implementations fail on behaviour, data, and sponsorship—in that order—and they fail in ways that are visible weeks before anyone admits it.
The twelve weeks that decide it
The five failure modes, and their early symptoms
| Failure | Early symptom | Fix while it is cheap |
|---|---|---|
| No real sponsor | Leadership skips the review or exports to a sheet | Move the review into the system; make leaders update records live |
| Dirty migration | “Which of these three ACME records is right?” | Stop the import; agree dedupe keys; reload from a clean extract |
| Scope creep | Custom objects before weekly usage is established | Freeze v1 scope in writing; queue everything else |
| Untrained managers | Coaching conversations happen outside the tool | Train managers first, on their own pipeline |
| No adoption metric | Nobody can say whether it is working | Pick two: same-day logging rate, deals with a future next step |
Data migration is a governance exercise
Clean before you import. Deduplicate accounts, normalise industry and stage values, agree what each field means, and decide the source of truth per field where two systems disagree. Bad data on day one demoralises users permanently—a seller who finds three versions of their own client stops trusting the system and never fully returns.
Import only what is live: active customers and open pipeline. The archive can wait, and often should never come across at all.
Name two owners
A business owner who decides what the system should do, and a technical owner who makes it happen. One person doing both is a bottleneck; nobody doing either is a stalled project. Publish a simple RACI for integrations, reporting, and field changes so requests have somewhere to go.
Measure adoption, not activity
Weekly active updaters per team, same-day logging rate, and the proportion of open deals with a future-dated next step tell you whether the system is real. Licence counts and login numbers tell you nothing—people log in to look at things they do not trust.
Work through the 30-point implementation checklist as a sponsor workbook, and read replacing spreadsheets for the migration narrative.
Frequently asked questions
- What is the #1 mistake?
- Importing messy legacy data without dedupe and field governance.
- Is there a checklist?
- Use our CRM implementation checklist.