The five phases of a CRM implementation
- Assess the process, not the tool. Map how the sale happens today: who handles leads, where they come in, what moves a deal to the next stage and where records get lost. Without this, configuration will just reproduce in the system an operation nobody follows.
- Choose the tool against written criteria. The decision between keeping the current CRM and switching comes from a list of requirements the operation needs to meet, not from vendor preference. In most cases the problem is process and integration, and switching tools fixes neither.
- Migrate with a cutoff rule. Migrating everything is rarely the best call: an old database carries duplicates and dirty data, and moving it unfiltered transfers the problem to the new system. Decide what comes over, what gets archived and what gets discarded before the first row moves.
- Configure and integrate with the real channel. Funnel, fields, automations and permissions, connected to where the sale actually happens. In Brazil that's almost always WhatsApp — if the CRM doesn't talk to it, logging becomes manual transcription, which is the first thing dropped in a busy week.
- Measure adoption, don't assume it. Training is the beginning. What closes out the implementation is tracking the percentage of deals with complete records for a few weeks and fixing whatever is costing the people who fill it in. An implementation without an adoption plan measures the wrong thing.
Why CRM implementations fail
The failure is rarely technical. The CRM gets configured, the data gets migrated, the team gets trained — and three months later the fields are empty, the funnel is out of date and the spreadsheet is back.
The reason is usually simple: the CRM was implemented so the manager could get reports, not so reps could sell better. When filling in the system is pure cost for the person doing it, the data entry dies.
- Too many required fields that nobody uses afterward
- A funnel that mirrors the org chart instead of the buying journey
- No integration: reps type into the CRM what they already typed on WhatsApp
- A tool chosen before the process was defined
The order that works
Process first, tool second, integration third. Reverse that order and you get an expensive, poorly filled-in tool.
In practice, a CRM only lasts when it gives something back to the people who feed it: follow-up that triggers on its own, history that's already there, a proposal that doesn't have to be redone.
Integration is what decides adoption
In Brazil, most selling happens on WhatsApp. If the CRM doesn't talk to the channel where the sale actually takes place, logging becomes manual transcription — and manual transcription is the first thing the team drops in a busy week.
Nexo builds CRPRO, an AI CRM integrated with WhatsApp, and also works with the CRM the company already has. The decision between keeping and switching comes out of the assessment, not out of vendor preference.
What goes into each phase
The scope varies with the starting point, but these are the deliverables — and they're how you verify that a phase is done.
- Assessment: a map of the current process, a list of losses by stage and written requirements for the tool
- Selection: a comparison of the options against the requirements, with the recommendation and the reason each option was ruled out
- Migration: a plan for what comes over, what gets archived and what gets discarded, plus the migrated and verified database
- Configuration: funnel, fields, automations, permissions and integrations in production
- Adoption: a usage playbook, team onboarding and the complete-records metric tracked for a few weeks
CRM implementation from Porto Alegre, delivered remotely
Nexo is based in Porto Alegre, Brazil, and runs CRM implementations for companies across Brazil and Latin America. The work is remote — assessment, configuration, migration and adoption support happen in online meetings and inside the tool itself — with in-person meetings when the project justifies it.
Do it in-house or hire someone
Implementing a CRM in-house works when there's someone with protected time to run the project, the authority to decide the process and the experience of having done it before. If any of the three is missing, the implementation doesn't fail outright — it drags on, and the cost shows up as an outdated funnel instead of an invoice.
The detailed comparison, with decision criteria, has its own page: implement a CRM in-house or outsource it.