Building a CRM clients actually use
Most CRMs fail not because of missing features, but because they model someone else’s workflow. Here is how I model the real one first.
A CRM is only as good as the day-to-day workflow it captures. The generic ones lose because they force a sales team to bend around a model that was designed for someone else.
Start with the pipeline, not the schema
Before writing a single migration, I sit with the people who will live in the tool and watch them work. The pipeline stages they describe out loud are almost never the ones the off-the-shelf product assumes.
Modeling that real pipeline first means the database mirrors how the business actually thinks — and adoption stops being a fight.
Automate the boring, surface the important
The fastest win in any CRM is removing manual follow-up. A small rules engine — “if a deal sits 5 days in negotiation, nudge the owner” — saves hours a week and earns trust immediately.
- Make every automation visible and editable by the user.
- Never hide why something happened — log it.
- Default to fewer, sharper notifications.
Software people enjoy using is software they keep using. Adoption is a design problem, not a training problem.
Related posts
Predicate pushdown, explained simply
How moving filters closer to storage turned a 90-second query into an 8-second one — without buying a single extra server.
WebGL without killing performance
You can ship a beautiful animated hero and still hit a 100 Lighthouse score. Here is the budget I work within.
ERP modules that share one ledger
The difference between an ERP and four spreadsheets in a trench coat is a single shared ledger. Here is why that matters.