Integrate to remove entry, not to add fields
Every CRM integration should be judged by one test: does it reduce what a person types? Integrations that add fields for someone to populate make adoption worse, whatever the reporting benefit.
That test rules out a surprising number of proposals and rules in the ones that get used.
The four that pay
- Activity capture from email and calendar, logged against the right record automatically.
- Enquiry creation from your website forms, with source and campaign attached.
- Quote and proposal generation from the opportunity, rather than retyping into a document.
- Revenue back from finance, so the CRM shows what was actually invoiced rather than what was hoped for.
The fourth is the one most often missing and the one that makes pipeline reporting trustworthy.
Why two-way contact sync goes wrong
Two systems both allowed to create and edit contacts will diverge. Someone updates an address in one, someone else in the other, and the sync applies whichever ran last.
Pick one system as the owner of customer records. Everything else reads. If both genuinely need to write, you need conflict rules written down — and you will still be adjudicating them a year later.
Matching and duplicates
The practical difficulty in most CRM integrations is deciding whether the incoming contact is the same person as an existing one. Email is the strongest key and is frequently missing or shared.
- Match on email first, then on name plus company plus domain
- Automatic merge only above a high confidence threshold
- A review queue for the middle band, not automatic creation
- Never create silently on a weak match — that is how duplicates breed
Report on outcomes, not activity
Once revenue data flows back, report on enquiries that became customers by source, not on leads created. The second number flatters channels that produce volume and no revenue.
This single change usually redirects marketing spend more than any campaign optimisation.