In my last newsletter, I shared three disasters that happen when dealerships run CRM platforms claiming CDP capabilities: a divorce exposed, a viral Facebook post, and a VIP lost to another dealership.

Today, I’ll share why these failures are architectural, not operational – and why ‘running a nightly batch process’ doesn’t fix them.

Consider who has ‘write’ access to your CRM on any given day:

Salespeople – creating leads from lot ups, phone calls, and internet inquiries

Sales managers – adjusting records, reassigning leads, merging what they think are duplicates

BDC agents – logging every call, creating new records when they can’t find existing ones fast enough

Service advisors -creating customer records for walk-ins

Service managers – managing workflows with their own record-keeping habits

That’s 50 to 100+ users writing to your CRM simultaneously, all day, every day. Each person has different data entry habits, different accuracy standards, and different incentives.

The salesperson trying to protect a lead creates a new record, not wanting to risk linking to someone else’s customer. The service advisor, in a rush, abbreviates the customer’s name and skips their email address. The BDC agent can’t find the record fast enough, so they create a fresh one.

This is the environment a CRM-based CDP would need to unify into a golden customer record – in real-time, at scale, without errors.

It can’t be done.

This is the inevitable objection from CRM vendors. “We run nightly identity resolution. Problem solved.”

It’s not solved and here’s why:

1. The damage happens during business hours.Remember the divorcing customer from Part 1? Her purchase confirmation went to her ex-husband’s email at 11:20am — minutes after the salesperson created the record. The Lexus owner received the insulting email about her vehicle by 9:15am the next morning. By the time any overnight process runs, customer trust has already eroded.
2. Records change mysteriously overnight.When records merge at 2am, fields get overwritten, tasks get orphaned, and salespeople arrive to find their leads mysteriously changed. They didn’t make these changes. They can’t explain them to customers. They lose trust in the system and start maintaining shadow records in spreadsheets.
3. Matching quality doesn’t improve.No matching rules work well on chaotic, human-entered data. Conservative rules miss obvious duplicates. Aggressive rules incorrectly merge unrelated customers. Middle-ground rules deliver both problems, just less frequently.
4. Real-time marketing can’t wait until midnight.Website personalization, ad suppression, behavioral triggers — modern marketing runs continuously. A system that only unifies at midnight can’t support any of it effectively.

Nightly batch processing means running on fragmented, unreliable data for 24 hours every single day.

A CRM must allow users to create and edit records freely. That’s its entire purpose. Operational speed demands it.

But you cannot simultaneously optimize your CRM for ‘let 50 people write to this system as fast as possible’ and ‘maintain pristine data integrity for identity resolution.’

The moment you enable the chaotic data entry a CRM requires, you’ve destroyed the controlled environment that identity resolution needs.

These capabilities are mutually exclusive. No amount of clever engineering changes this.

If CRMs can’t become CDPs, can CDPs become CRMs? (Spoiler: No – and the reason reveals what actually works.)

Next: the architectural feature that makes CDPs valuable is the same one that makes them useless as CRMs – and the clean architecture that actually solves the problem.