Ask why a business has a CRM and the answer is usually pipeline visibility. The fields that exist are the fields a forecast needs: stage, value, close date, owner. The fields that would let anyone reconstruct what the buyer said are absent or free text nobody reads.
That configuration makes the system useful to one person a week and a tax on everybody else. Sellers fill it in because they are asked to, the data is thin because filling it in produces nothing they benefit from, and the forecast built on it is worse than the one in their heads.
The other job it could do
A record built as memory holds what the buyer said in their words, what they were worried about, what was sent and when, what they read, and what was promised. That record is what makes the second conversation different from the first, and the handover to delivery not feel like starting over.
It is also the input for everything in the fourth and fifth lenses. Objection frequency, follow-up adherence, the real reason opportunities were lost rather than the dropdown value, and the variance between what was predicted and what happened all come from a record that was kept for memory rather than for reporting.
Where automation belongs
Most of the capture is production work, which means it should not be a person's job. Routing, logging, enrichment and the mechanical parts of follow-up are exactly the layer a machine substitutes for headcount, and a seller entering data is a judgment seat doing production work.
What stays with the person is the reading of it. A system can record that a buyer hesitated at the proposal stage. It cannot tell you what that hesitation meant, and it should not be configured as though it could.