A CRM nobody updates is an expensive address book.

Buy, unless your pipeline has stages no product models, per-user pricing has overtaken build cost, or you need deep integration with an operational system. For a conventional sales process the available products are good and inexpensive, and building would waste your money.

Adoption, almost always. The sales team finds it slower than what they did before, so entry becomes patchy, then the reports stop matching reality, then nobody trusts it. The cure is making the common actions fast rather than adding features.

Through the official WhatsApp Business API, yes. In India this matters more than in most markets because a large share of sales conversation happens there. A CRM that cannot see those conversations holds a systematically incomplete record.

They should, and it is the first thing to build. Website forms, ad platform lead forms and WhatsApp enquiries all arriving automatically, assigned and timestamped. Manually entering a lead that already exists digitally is where CRM discipline collapses.

A focused first version covering your actual pipeline is usually weeks. I would rather ship something small that the team uses and extend it than spend months building against a specification written before anybody used anything.

Then it is too slow, or it asks for information they do not have at the moment they are asked for it. Both are design problems rather than discipline problems, and both are fixable — but not by mandating usage.