AI readiness is not a count of licenses. A store can own several AI products and still be unprepared to operate any of them well. Readiness means the dealership can identify a useful problem, supply dependable inputs, protect customer information, assign ownership, manage exceptions, and measure a result.
The Dealer AI Skills readiness framework spans data, websites, visibility, compliance, departments, governance, training, KPIs, and roadmaps. For an executive first pass, these 12 questions expose most of the risk quickly.
1. Which business problem are we solving?
“Use AI” is not a business problem. “Reduce the time between an after-hours lead and a useful response” is. Require every proposed project to name a trigger, current failure, affected customer or employee, and measurable outcome.
2. What is the baseline?
Before a demo, record current performance. For a lead workflow, capture response time, contact rate, appointment set rate, show rate, and manual touches. For service, capture unanswered calls, scheduling abandonment, declined-work follow-up volume, and recovered appointments. Without a baseline, every post-launch story becomes opinion.
3. Who owns the process today?
The AI owner should usually be the manager accountable for the business outcome, not the person most excited about technology. IT, marketing, or a vendor can support the implementation. They should not become the substitute owner for sales, BDC, fixed ops, or F&I performance.
4. Where does the source data live?
List every system the workflow needs: CRM, DMS, inventory feed, phone platform, scheduler, website, OEM system, spreadsheets, or vendor portal. For each source, note the data owner, access method, update frequency, field quality, and export or API limitations.
If the answer is “someone downloads a report when asked,” the workflow is not ready for dependable automation. It may still be ready for a manager-assisted pilot.
5. Can the systems exchange status?
A tool that sends a message but cannot write the activity and outcome back to the CRM creates a second system of record. That increases duplicate outreach and reporting gaps. Car Dealership Guy has described the industry’s integration problem as a patchwork in which context gets lost and employees bridge systems manually. Before buying, map both directions: what the AI reads and what it writes back.
6. Which customer information will the tool touch?
Classify the minimum data required. Names and vehicle interests present a different risk than credit applications, Social Security numbers, bank information, or finance status.
The FTC’s automobile dealer Safeguards Rule FAQ explains that most dealers arranging financing or leases must maintain a written information security program for covered customer information and oversee relevant service providers. An AI pilot should not become an unofficial path around those controls.
7. What is the permission model?
Ask whether the tool supports role-based access, least-privilege permissions, multifactor authentication, and removal of access when an employee changes roles. Document who can view prompts, transcripts, exported data, model outputs, and administrative settings.
8. What may the AI decide or send?
Create three lists:
- Allowed without review.
- Allowed only after human approval.
- Prohibited.
Customer pricing, availability, financing terms, credit decisions, legal statements, employment decisions, and emotionally sensitive responses deserve strict boundaries. Even internal summaries need source links or a way for managers to verify the underlying record.
9. How does the customer reach a person?
Every customer-facing workflow needs an escalation path. Define the conditions: explicit request for a person, repeated misunderstanding, complaint, legal threat, pricing dispute, safety issue, vulnerable customer, or low-confidence answer. Define the response time and destination queue.
10. Is the team trained on the process?
Training is not a vendor webinar. Users should practice real tasks, review failures, and demonstrate the handoff. Managers should know how to audit adoption and quality. The store should also define prohibited inputs so employees do not paste sensitive customer or business information into unapproved public tools.
11. What will we measure weekly?
Choose one adoption KPI, one operating KPI, one outcome KPI, and one risk KPI. For example:
| Type | Example |
|---|---|
| Adoption | Eligible leads handled through the process |
| Operating | Median first-response time |
| Outcome | Appointments shown |
| Risk | Incorrect messages per 100 reviewed |
12. What is the exit plan?
Know how to export records, revoke integrations, retain required logs, delete dealership data, and continue the process if the vendor is unavailable. Contract renewal should not be the first time anyone asks who owns prompts, customer records, workflow logic, or performance history.
Score the store honestly
Give each question a score of 0, 1, or 2:
- 0: unknown or absent.
- 1: partially documented or inconsistently followed.
- 2: documented, owned, and verifiable.
A score below 12 means the store should fix the foundation before authorizing customer-facing automation. A score from 12 to 18 supports a bounded assistive pilot. A score above 18 supports a more connected pilot, provided the specific use case passes legal, security, and process review.
The score is not a certification. It is a way to replace enthusiasm with a useful management conversation.