Dealership CRM With AI Follow-Up

Ask any vendor what the AI is actually doing. The answer separates real automation from a relabeled drip campaign.

Follow up is where deals are won and it is where almost every store leaks. Salespeople make two or three attempts, the trail goes cold, and the customer buys somewhere else three weeks later from someone who was still politely present. Nobody disputes this. It has been the same for thirty years.

So every CRM vendor now sells AI follow up, and the phrase covers at least four different things. It is worth separating them before you sit through a demo.

Scheduled sequences. A cadence of texts, emails and tasks that fires on a timeline after a trigger. This is not artificial intelligence in any meaningful sense, and it is also the single highest value thing on the list.

Rule based branching. The sequence changes based on what the customer did: replied, opened, booked, went quiet. Still not AI, still valuable.

Generated message text. A model writes the message. Real AI, and the one to ask hard questions about, because a model writing to your customers unsupervised is a compliance and brand risk that belongs to you and not to the vendor.

Autonomous conversation. A bot holds a full exchange with the shopper, handles objections and books the appointment. Some vendors genuinely do this. Verify what it does when the conversation goes sideways, and ask to hear real recordings rather than a scripted demo.

We would rather be specific than fashionable, so here is the honest breakdown of our own product.

The follow up engine is deterministic. Automations, follow up processes, drip campaigns and task automation run on triggers and rules you define and can read. That is a feature and not an apology. A manager can open the sequence, see exactly what will be sent and when, change it, and explain it to a salesperson whose pay depends on it. Black box behavior is a liability in a business where every outbound message is regulated.

The genuine AI in the platform sits elsewhere. Call recording with transcription turns a twelve minute conversation into text a manager reads in thirty seconds. Document AI reads documents and performs VIN scanning from a phone camera, so a trade in gets captured by photographing the windshield instead of transcribing seventeen characters.

What we do not claim: autonomous negotiation, a bot that closes on your behalf, or predictive scoring that ranks which shoppers deserve attention. We do not score or rank leads at all, which is covered on the predictive lead scoring page.

If a competitor's AI does more than this, ask them the questions in the last section of this page and hold them to the answers. Then compare on what the follow up actually produces rather than on the label.

Here is the surface, named specifically rather than described vaguely.

Automations and automation lists define what happens when something changes: a lead arrives, a status moves, an appointment is missed, a deal stalls. Follow up processes hold the long form cadences, the ones that run for weeks. Drip leads with a drip editor let you build and edit sequences visually rather than through a support ticket. Task automation schedules the human steps, because some touches have to be a person and the mistake is leaving those to memory. Reminders and appointments tie into the calendar with its own settings.

Two design points matter more than the feature list. First, the cadence is attached to the customer rather than the salesperson, so when someone leaves in month four their pipeline does not leave with them. Turnover is the normal state of this business and your follow up should survive it.

Second, everything is editable by a manager. A cadence that requires a vendor ticket to change is a cadence that never gets changed, and follow up copy that never changes goes stale within a season. See the follow up cadence page for how to structure the timeline itself, and task automation for the human step side.

A cadence is only as good as the channels behind it, and this is where a lot of CRMs are thinner than they look.

SMS and MMS with real threading, so a photo of a vehicle lands in the same conversation as the message about it. RCS with automatic SMS fallback, which gives you richer branded messaging on handsets that support it without excluding anyone whose phone does not.

Email with a real inbox, a composer, campaigns, lists and bulk email with recipient management, plus email domain authentication so your messages arrive rather than landing in spam.

Voice. Click to call with a VoIP softphone, call forwarding, trigger calls, call logs, and voicemail drop with recipient lists, which is the underrated one. A pre recorded message dropped straight to voicemail lets one person cover fifty follow up touches in the time three calls would take, and the customer gets a human voice instead of a template.

On the inbound side AutoMail adds IVR with call routing and forwarding, number management and full call history, so the calls your follow up generates get answered and directed instead of ringing into a busy showroom.

Then the hygiene tools most dealers are surprised to find inside a CRM: a phone validator and an email validator, so a cadence is not spending three weeks messaging a disconnected number.

Two variables decide whether follow up works, and automation should serve both.

The first is time to first response. The relationship between how fast a store answers and how often it sells is one of the most consistently supported findings in this industry, and it is entirely inside your control. Minutes, not hours. The shopper who filled out your form filled out three, and the store that responds first defines the conversation.

Automation earns its money here in a specific way: an immediate, honest acknowledgement that a real person is picking this up, sent within seconds at any hour, followed by a human contact as fast as your staffing allows. Distribution rules put the lead in front of somebody working, missed lead alerts surface anything sitting untouched, and after hours coverage means a ten at night submission is not waiting for the morning meeting. The numbers are on the response time page.

The second variable is persistence. Most stores quit after two or three attempts while buyers commonly take weeks. A cadence that runs to twelve or fifteen touches over a month, mixing text, email, voicemail and live calls, is not aggressive. It is the minimum that matches how people actually shop, and it is exactly the kind of discipline software should carry so a human does not have to.

Automated outbound messaging is regulated, and the exposure lands on the dealership rather than on the software vendor. Any conversation about AI follow up that does not cover this is incomplete.

Consent has to be captured and stored, opt outs have to be honored immediately and across every channel, and quiet hours have to be respected. On our side, opt out status applies platform wide, so a customer who unsubscribes from a service message is not then contacted by a sales cadence. A blacklist with import handles numbers that must never be contacted. Messaging is consent based throughout.

The reason to be cautious about generated message content is the same reason. If a model composes an outbound text to a customer, someone still has to be able to answer what was said, why, and whether it complied. Templates a manager wrote and approved are auditable. Generated text is harder to defend after the fact, which is why our generated content sits on the reading side rather than the sending side.

Our page on TCPA compliance for dealership texting covers the rules in practical terms. None of this is legal advice, and a store running heavy outbound volume should have counsel review the program.

Not everything with the label is hype. Two applications in our platform are doing work no human process does better.

Call transcription. Most stores record calls and almost none review them, because nobody has forty minutes to listen to six calls. Transcription changes the economics completely. A manager scans a conversation in thirty seconds, sees exactly where the appointment request was fumbled, and coaches with a quote instead of an impression. That is a training tool disguised as a compliance feature, and it is the single most underrated thing in the platform.

Document AI. Reading documents and scanning VINs from a phone camera removes transcription errors at the point where errors are most expensive. A wrong digit in a VIN follows a deal all the way to the contract.

Both share a pattern worth noticing. The AI reads and structures information for a human decision. It does not talk to your customer and it does not decide who is worth calling. That is a deliberate line, and if a vendor crosses it in their product they should be able to explain exactly how they keep it safe. More on the transcription side at call transcription for dealerships.

The common failure is not that the automation does not work. It is that the store builds nine sequences in week one, cannot tell which is doing what, and quietly turns the whole thing off.

1. Build one cadence first. Pick your highest volume lead source and write a single sequence for it. Nothing else.
2. Write the copy like a person. Short, specific, signed by an actual salesperson. Templates that read like a mail merge get ignored, and worse, they train your customers to ignore your number.
3. Define who owns the replies. Automation creates conversations. If nobody is watching the inbox, you have built an expensive way to annoy people.
4. Run it for thirty days without editing. Changing it weekly means you never learn anything.
5. Then measure three things: time to first response, touches per lead before it goes cold, and appointments set. Reporting across activity, store performance and a management roll up will show all three.
6. Only then add the second cadence.

One honest caution. We cannot guarantee close rates, appointment volume or units from any of this, because that depends on your inventory, your pricing and your people. What automation reliably does is remove the excuse that somebody forgot. If you want to see the engine running on live lead flow, the demo is the fastest route, and contact us if you would rather just ask what it would look like on your volume.

Is your follow up actually AI, or is it automation?

The follow up engine is deterministic: automations, follow up processes, drip campaigns and task automation running on triggers you define and can read. The genuine AI in the platform is call transcription and document AI with VIN scanning. We would rather be precise than fashionable.

Does a bot talk to our customers on our behalf?

No. We do not offer autonomous conversation, generated outbound message content or a bot that negotiates. Messages come from templates your managers write and approve, which keeps every outbound message auditable and defensible.

How many follow up touches should a cadence have?

Most stores quit after two or three while buyers commonly take weeks. A cadence of twelve to fifteen touches over about a month, mixing text, email, voicemail drop and live calls, matches how people actually shop rather than how quickly staff get discouraged.

What happens to a salesperson's pipeline when they leave?

Cadences are attached to the customer record rather than the individual, so the follow up keeps running and the pipeline gets reassigned. Turnover is normal in this business and your process should survive it.

How are opt outs handled?

Immediately and platform wide, so a customer who unsubscribes from one type of message is not contacted by another cadence. Messaging is consent based, and a blacklist with import covers numbers that must never be contacted.

Do we need a DMS integration to run this?

No. No DMS integration or inventory feed is required to operate, which is why individual salespeople and brokers run on the platform. If you do have a feed, Inventory Link can ingest it separately.

See the drip editor, voicemail drop and call transcription working on real lead flow, with the rules visible rather than hidden. Month to month, no long term contract.