The Problem
Floor traffic is expensive to generate and almost impossible to replace once it walks out the door. Every up who leaves without a deal represents a real cost — ad spend, time, floorplan carrying costs — and the window to re-engage closes faster than most managers think. The problem isn't that your salespeople don't care. It's that they're managing active deals, working the lot, and handling trade appraisals. Systematic follow-up on unsold prospects falls through the cracks because nobody has the bandwidth to do it consistently, at the right time, with the right message.
- !Unsold ups go cold within 72 hours while your team is focused on working the floor
- !CRM tasks pile up — notes don't get logged, follow-up calls don't get made, hot leads age out
- !Internet leads from third-party sites like Cars.com and AutoTrader get slow initial responses, killing contact rates
- !Service-to-sales opportunities get missed because BDC and sales aren't talking to each other
- !Managers spend Monday morning reviewing what didn't get done instead of coaching what did
Where AI Fits In
AI for car dealerships means building automated systems that do the follow-up work your BDC can't keep up with — texting unsold ups the same evening they visit, responding to internet leads in under two minutes, and surfacing service customers who are equity-positive and due for an upgrade. These aren't generic chatbots. They're purpose-built workflows that connect to your DMS, your CRM, and your inventory system so every message is specific and timely.
Most Common Starting Point
Most dealerships start with automated unsold-up follow-up — a triggered sequence that goes out the same night a customer leaves without purchasing, continues through the 72-hour window, and routes hot responses directly to a salesperson's cell phone.
Unsold-Up Follow-Up Engine
Automated SMS and email sequences triggered the same evening a customer visits the floor, personalized to the vehicle they looked at, with hot-lead routing to the assigned salesperson.
Internet Lead Response Bot
Sub-two-minute AI responses to leads from your website, Cars.com, AutoTrader, and CarGurus — with real inventory details pulled live from your system, not canned templates.
DMS-Connected Equity Mining Workflow
Automated identification of service customers who are in positive equity or approaching lease end, with triggered outreach sequences routed to your sales team.
CRM Health Dashboard
A live view of tasks not completed, leads aging past threshold, and follow-up gaps — so managers see problems in real time instead of on Monday morning.
Other Areas to Explore
Every car dealership business is different. Beyond the most common use case, here are other areas where AI automation often delivers results:
What a High-Volume Sales Day Actually Looks Like — Before and After
Here's a typical Saturday at a busy store. The floor is running hot — eight, ten, maybe twelve ups by noon. Salespeople are doing their jobs: working deals, running numbers, babysitting the finance office. The BDC is handling inbound calls and trying to set appointments from internet leads. Nobody has time to systematically reach out to the six people who left without buying yesterday.
By Sunday evening, those six people have been sitting in the CRM with a "No Sale" status and a task assigned to someone who won't see it until Monday morning. One of them already bought somewhere else. Two more are comparing prices online. The window was open. It's closing.
Now picture the same Saturday with an automated follow-up system in place. The moment a customer's status is marked as unsold, a trigger fires. That evening — not Monday, that evening — they get a text from their salesperson's name: "Hey, just wanted to follow up on the F-150 you were looking at today. We got some flexibility on that payment if you want to revisit." It's specific. It's fast. It doesn't feel like a mass blast.
The manager's morning on Monday looks different too. Instead of running a report to see who didn't get followed up with, the dashboard shows which unsold ups responded, which ones haven't been touched, and which ones clicked a link in the email. The team starts the week working warm leads, not triaging last week's neglect.
What doesn't change: the salesperson still owns the relationship. The system doesn't replace the conversation — it makes sure the conversation actually happens. (Source: Cox Automotive, 2023) Research from Cox Automotive found that the majority of car buyers contact only one dealership before purchasing, which means your first follow-up isn't just courteous — it's often your only shot.
- Same-evening outreach replaces Monday morning catch-up
- Personalized messages reference the specific vehicle, not a generic "thanks for visiting"
- Hot response routing puts a reply in the salesperson's pocket within seconds
- Managers see the pipeline health live, not in a Monday morning report
What Your DMS, CRM, and Lead Sources Actually Need to Talk To Each Other
Integration is where most dealership tech projects stall. Not because the concept is complicated, but because the data underneath is messier than anyone admits at the start. Here's what a realistic integration actually involves at a car dealership.
Your DMS — whether that's CDK, Reynolds & Reynolds, Dealertrack, or DealerSocket — is the source of truth for customer records, deals, and service history. Any AI system worth building needs to pull from it. The challenge is that DMS APIs range from well-documented to actively hostile, and some require third-party middleware like DealerSocket's OpenTrack or a vendor like Tekion to expose the data cleanly.
Your CRM is where lead activity lives — VinSolutions, Elead, and DealerSocket CRM are the most common. This is where follow-up tasks get assigned, lead sources get tracked, and unsold-up statuses get logged. The AI system needs write access here, not just read access, so it can log touchpoints automatically without a BDC rep doing it manually.
Then there are your lead sources: your OEM website, your dealer website (usually on a platform like Dealer Inspire or DealerOn), Cars.com, AutoTrader, CarGurus, and any trade-in tools like Kelley Blue Book's ICO widget. Each of these fires leads in slightly different formats. Normalizing that data before it hits your follow-up logic is non-negotiable.
Before starting, a dealer principal should be able to answer these questions clearly:
- What CRM are you on, and do you have admin credentials available?
- What does your current lead routing look like — who gets what, and how fast?
- Are your customer records deduplicated, or does the same person appear six times with different phone numbers?
- Do you have a texting compliance setup — opt-in records, TCPA documentation — already in place?
- Who owns the DMS relationship internally, and can they pull an API key or connect a vendor?
Oaken AI builds on a stack — Python, FastAPI, PostgreSQL, and LangChain for orchestration — that's designed to handle the messiness of real dealership data. But the cleaner your records are going in, the faster you'll see results coming out.
The One Automation That Pays for Everything Else: Internet Lead Speed-to-Contact
If you're spending money on AutoTrader, Cars.com, or CarGurus leads, you already know the frustration: you pay for the lead, it comes in at 9 PM, and by the time someone responds the next morning, the customer has already heard back from two other dealers. Speed-to-contact is the single biggest driver of whether an internet lead converts. Research from the automotive industry consistently shows that leads contacted within the first few minutes are dramatically more likely to result in an appointment than those contacted an hour later. (Source: Dealers United / LeadsCon research, widely cited in automotive retail training — contact rate advantage for sub-5-minute response is well-established in dealer 20 Group benchmarking data)
Here's how the automation actually works. A lead comes in from any of your configured sources. The system — connected to your inventory feed via your DMS or a tool like vAuto — pulls the specific vehicle the customer inquired about: the stock number, current price, available photos, and any similar vehicles in stock. Within 90 seconds, the customer gets a text and an email that reads like a real person sent it, because the content is specific to what they asked about.
The message doesn't try to close the deal. It confirms receipt, surfaces one or two relevant vehicles, and asks a simple qualifying question — "Are you flexible on color, or is the white one the one you want?" That question does two things: it moves the conversation forward, and it signals intent. A customer who answers is a customer who's engaged.
When they respond, the system logs the interaction in your CRM and simultaneously routes an alert to the assigned salesperson or BDC agent. The human takes it from there. The AI's job was to make contact before the lead went cold — and it did that at 9 PM on a Saturday without anyone touching a keyboard.
On day one, dealers notice they're getting responses from leads they never would have reached. By month three, the pattern is clear: contact rates go up, appointment sets go up, and the Monday morning "why didn't we respond to this" conversation gets shorter. The system doesn't make your people redundant — it gives them warm conversations instead of cold dials.
- Sub-90-second response time, any hour, any day
- Inventory-specific content pulled live from your DMS or vAuto feed
- CRM logging happens automatically — no manual entry required
- Hot responses routed immediately to the assigned rep's phone
How It Works
We deliver working systems fast — no multi-month assessments, no slide decks. A typical engagement runs 3-5 weeks from kickoff to live system.
Week 1-2
DMS and CRM integration, data audit, and inventory feed connection. We map your lead sources, clean up your customer records, and confirm what's actually firing in your current follow-up process.
Week 2-3
Build and test the primary follow-up automation — unsold-up sequences, internet lead response, and routing logic. Run parallel with existing process so nothing breaks mid-month.
Week 4-5
Go live, monitor response rates, and tune messaging based on what customers are actually replying to. Managers get dashboard access and a walkthrough on reading the data.
The Math
Unsold-up conversion rate and internet lead contact rate
Before
Leads aging out, follow-up inconsistent, BDC chasing yesterday's problems
After
Every unsold up touched within hours, every internet lead responded to in minutes, manager attention on closing — not chasing
Common Questions
Will this replace my BDC team?
No, and that's not the goal. What it replaces is the gap between when a lead comes in and when a human can respond — which is often hours or days. Your BDC team handles conversations, builds rapport, and sets appointments. The AI handles the initial contact, the after-hours outreach, and the systematic follow-up sequences that would otherwise fall through the cracks. Most dealers who implement this find their BDC becomes more effective because they're working warmer leads instead of cold ones.
What CRM systems do you work with?
We've built integrations with VinSolutions, Elead (now part of Elead/CDK), DealerSocket CRM, and Tekion. If you're on a different system, the first conversation is about whether there's a usable API or data export. Some older CRM configurations require a middleware layer, which adds a week or two to the build — but it's rarely a dealbreaker.
How do you handle TCPA compliance for text messages?
TCPA compliance is your responsibility to document — we build the system, but we can't retroactively create opt-in records that don't exist. Before we go live with any SMS automation, you need to confirm that the customers being messaged have opted in to receive texts from your dealership. Most modern CRMs capture this, but it needs to be verified. If you're not sure where you stand on this, that's the first thing to sort out before any texting automation goes live.
How long before we see results from the follow-up automation?
The first signals come fast — usually within the first week of going live, you'll see responses from customers who would have gone cold under the old process. Trend-level data — contact rate improvements, appointment rates, unsold-up re-engagement — typically becomes clear within 30 to 45 days of consistent operation. Month three is usually when managers start to see the pipeline health shift in a way that shows up in unit numbers.
Can the system handle multiple rooftops?
Yes, but multi-point setups require more upfront planning — particularly around lead routing, which rooftop owns which customer record, and how to handle customers who have visited multiple stores. We've built for dealer groups before. The architecture is the same; the data mapping is more complex. That complexity should be scoped and priced before the project starts, not discovered mid-build.