Nobody in your store is falsifying data. But the lead-source field is filled in from memory, days late, by someone whose pay plan doesn't depend on it — and then you set a six-figure ad budget on the result.
The short answer: the lead-source field in most dealership CRMs is not a measurement. It's a recollection, typed after the fact, chosen from a dropdown that mixes real sources with lifecycle stages and catch-all buckets. That's why "Walk-In," "Lives in Area" and "Manual Add" quietly become your biggest channels — and why the channels that actually produce cars look underfunded. You fix it by capturing the source at the moment of contact and reconciling everything else against evidence, not memory.
Pull your last 90 days of sold units and group them by lead source. In almost every store, the top of that list contains at least two of these:
| What the CRM says | What it usually is | How you can tell |
|---|---|---|
| Walk-In | An online shopper who finally drove over | Their phone or email is already in your web leads |
| Lives in Area / Drive-By | The default a rep picks when they don't know | Clusters on one or two reps, never the whole floor |
| Manual Add | A phone or text lead typed in by hand | No form, no marketplace record, no call linked |
| Referral | Sometimes real. Often "keep it off the ad budget" | Spikes at month-end, or when a spiff changes |
| CRM / Opportunity / Internet | A system bucket, not a source at all | It grows every month and nobody knows what's in it |
None of those is a marketing channel. Four of the five are process artifacts. And they're at the top of the list because a catch-all always wins a popularity contest against twelve specific options.
The field is structurally set up to fail, in five ways:
A wrong source isn't a cosmetic problem. It moves money.
Every channel decision downstream — what you renew, what you cut, what you double — runs on this field. If a third of your web-originated sales are wearing a "Walk-In" badge, then your paid channels look expensive and your showroom looks magically productive. So you cut the ad spend, the showroom traffic falls two months later, and nothing in the report explains why.
It also corrupts the two numbers you actually manage the store by. Your cost per sale by channel is calculated by dividing real spend by mis-assigned units — real numerator, fictional denominator. And your close rate per source is worse than useless: a source stuffed with catch-all records will always look weak, and a source that only gets credited when a deal closes will always look like your best performer. Sources that only ever get tagged on winners cannot be compared to sources tagged on everything.
The single highest-leverage change is moving the capture point earlier — from "when the deal is worked" to "when the customer arrives."
That's the whole argument for a structured floor check-in. When every up is logged on a tablet at the door — who they are, what they came for, what brought them in — the source is recorded by the person standing in front of the customer, at the only moment when the answer is actually known. It stops being a memory test. It also gives you a denominator for walk-ins, which most stores don't have at all; the mechanics of that are in how to track walk-in traffic.
The rule to enforce: a lead's source is set once, by the first system that saw the customer, and nothing downstream may overwrite it. Later systems can add touchpoints. They don't get to rewrite the origin.
Capture-at-contact fixes new leads. It does nothing for the ones already in the system, and nothing for the shopper who was on your site for two weeks before walking in. For those you need reconciliation — matching each lead against the digital record instead of trusting the dropdown.
Three things make that work:
Worth saying plainly, because most vendor copy in this category won't:
Six things to require, of any system or any process:
Get those six and every channel report downstream — attribution, cost per sale, close rate by source, and the numbers on your daily board — becomes an argument about strategy instead of an argument about the data.
Why is my dealership CRM's lead source wrong? Because it's usually recorded from memory after the fact, not measured at the moment of contact. The source field is typically filled in when a deal is worked — sometimes days later, often by someone who didn't take the original up — and chosen from a dropdown that mixes real channels with lifecycle stages and catch-all buckets like "Manual Add" or "Lives in Area." Add handoffs between your website provider, marketplaces and the CRM (each of which can overwrite the origin with the last system that touched the lead), plus duplicate records for the same shopper, and the field ends up describing your process rather than your marketing.
Why do so many dealership leads show up as "walk-in"? Because "walk-in" is what the store observes, not where the customer came from. Most people who walk onto a lot have already spent days on your website and on third-party listings; the walk-in is the last step of a digital shopping process, not the first. When the source is recorded at the door with no lookup against prior online activity, every one of those shoppers is logged as a walk-in — which inflates showroom performance and makes the paid channels that actually created the visit look like they aren't working. The fix is to check each arrival against your own web and lead history, so a returning online shopper is recorded as one lead with an online origin and a showroom touchpoint.
How do you clean up lead-source data in a dealership CRM? In two passes. Going forward, move the capture point to first contact — log every up at the door in a structured check-in, restrict the source list to actual channels, and forbid any downstream system from overwriting an origin. Backward, reconcile existing leads against evidence: de-duplicate records for the same person (including across rooftops), match them to the durable visitor identity your own site assigns rather than click IDs alone, and report the reconciled source alongside what the CRM says instead of silently overwriting it. Track the gap between the two as its own metric — when it shrinks, your channel reporting is becoming trustworthy. Accept that a small residual bucket is genuinely unattributable and label it as such.
GhostDrive captures the source at the moment of contact, matches every lead to a durable visitor identity across sources and rooftops, and shows the reconciled source next to what your CRM says — so you can see the size of the gap before you set next month's spend.