A CRM is usually bought as a software project. It works best when it is treated as a business design project: deciding what the business needs to know, and building the system around that.
This guide sets out where CRM implementations go wrong, what the good ones have in common, and how to prepare.
Key PointDecide what the business needs to know before you decide which fields the CRM should hold.
Where CRM implementations go wrong
A familiar pattern: the CRM arrives, the team is trained, historic data is loaded, and then usage drifts. Pipeline figures become unreliable, reports stop being opened, and the team drifts back to email and spreadsheets.
Before blaming the software, look at how the system was configured and what it was asked to represent.
Storing data, or modelling work?
A CRM can be built to store data or to model work. To tell which, ask a salesperson whether a pipeline figure is right.
- "Roughly, I think" means the CRM is storing data.
- "Yes, it is calculated from the records" means it is modelling work.
Stored data has to be kept up to date by hand. When people are busy, it slips, and out-of-date data looks authoritative when it is not.
Watch OutA figure that has to be updated by hand will drift. Calculate it from the records where you can.
Adoption is a design problem
When a tool is shaped around what the software can store rather than how the team works, it adds steps instead of removing them. A sales team will not fill in fifteen fields to log a note, and a project team will not manage complex work in a flat task list.
If people are not using the CRM, ask whether it makes their work harder or easier before you plan more training.
What good looks like
- Model work, not storage. Every field is something the business needs to know or act on.
- One source of truth for each fact. A customer name lives in one place, and statuses are calculated rather than typed.
- Built for handover. The people who use it daily can change stages, fields and views without a developer.
- Only the data you need. UK GDPR, as summarised on GOV.UK, requires personal data to be adequate, relevant and limited to only what is necessary, and accurate and, where necessary, kept up to date.
The ICO's guidance on data minimisation is a useful check when you decide which fields to keep.
Pro TipFor every field, ask who uses it and for what. If nobody can answer, leave it out.
Before you choose a platform
Map your own information flows first. Start with what the business needs to know to run itself, not what data you already have.
- Which figures do people have to go and find?
- Where does manual reconciliation happen?
- What does someone check before a client call, and how long does it take?
If the answers show the CRM must connect to finance, project or e-commerce systems, treat that as an integration question from the start. Microsoft's retry pattern guidance is a good test of how a supplier handles failed connections.
Next steps
Our articles on what breaks when systems don't talk and what an integration actually costs go further. See how we approach business systems and CRM, or book a free call.