A CRM never fails loudly. It degrades — a field nobody fills, an automation switched off during an incident and never switched back, a duplicate record set that quietly splits one account into three. By the time revenue is visibly affected, the system has been lying to the forecast for months.
Seven symptoms worth taking seriously
- 1Reps keep a private spreadsheet. The single clearest sign the system is not serving the work.
- 2Duplicate contacts and accounts. Split history means split judgement and embarrassing double outreach.
- 3Forecast accuracy is poor and nobody trusts stage definitions.
- 4Leads sit unassigned overnight or over weekends because routing has no fallback.
- 5Automations exist but nobody can say which are live or who owns them.
- 6Reporting is exported to a spreadsheet before anyone can read it.
- 7Support and sales cannot see the same customer timeline.
The one-week audit
- 1Day 1 — Data health. Count duplicates, records missing an owner, missing email or phone, and records untouched for 12 months. Express each as a percentage of the base.
- 2Day 2 — Pipeline integrity. Check every open deal has a next step and a realistic close date. Deals with neither are not pipeline.
- 3Day 3 — Routing and response. Trace ten real inbound leads end to end and time each step from submission to first human contact.
- 4Day 4 — Automation inventory. List every workflow, its owner, last edit and last successful run. Disable orphans.
- 5Day 5 — Adoption. Sit with three users and watch them work. Note every place they leave the CRM to do their job.
- 6Day 6 — Reporting. Rebuild the three reports leadership actually uses and validate the numbers against source systems.
- 7Day 7 — Prioritise. Rank findings by revenue impact against effort, and fix the top three before touching anything else.
Fix order matters
Clean the data before automating it, and fix routing before adding scoring. Automation applied to bad data simply distributes the errors faster and more confidently. Most teams find that deduplication, ownership rules and a working routing fallback recover more revenue than any new module.
Keeping it clean
- Required fields only where they change a decision — every other mandatory field trains people to type nonsense.
- A named owner for the CRM, with time allocated to it.
- Monthly data-health report with the same four percentages every time.
- Quarterly automation review, including a kill list.
The ReTectra View
How we think about this
- We start CRM engagements with the audit above, not a migration proposal. Most systems do not need replacing — they need ownership.
- Data cleanliness comes before automation in every plan we write, because automating dirty data multiplies the error rather than the output.
- Our CRM specialists run the maintenance layer continuously — deduplication, ownership, hygiene reporting — so the system stays trustworthy after the project ends.
- Migrations are run with zero-downtime cutovers and validated record counts; the reporting must reconcile before we call it done.