Ask a small dealer which marketing actually sells cars and you'll usually get a feeling, a vendor dashboard, and a story about one great month. Meanwhile the budget quietly funds whatever has always been funded. Enterprise groups solve this with attribution software; an independent store can solve most of it with discipline, a fixed source list and one spreadsheet — because at thirty to eighty sales a month, you don't need statistical modeling, you need honest bookkeeping. Here's the system.
Step 1: fix the source data, or nothing downstream matters
Attribution dies at intake. Three fixes, all free:
- One fixed source list in the CRM — a locked picklist, not free text: Website, Marketplace, each paid lead provider by name, Google (organic/profile), each ad platform, Phone-Up, Walk-In, Referral, Repeat, Service-to-Sales. Ten to fifteen values, no "Other" allowed without manager sign-off, and free-text entry turned off forever.
- Auto-capture what can be auto-captured. Internet leads should land in the CRM with the source already stamped by the integration. Humans mislabel; plumbing doesn't.
- Ask every up, every time. For walk-ins and calls, "what brought you in today?" is your attribution software. Train the exact question, log the exact answer. Tracking phone numbers per channel help if you have them; the trained question is the free version that catches most of it.
Step 2: the two-question interview at delivery
Here's the wrinkle honest stores discover: the first-touch source and the real reason often differ. A buyer logged as "Walk-In" watched your videos for a month; a "Website" lead was referred by a cousin. So capture both, and the delivery desk is the perfect moment — customers are happy and candid. Two questions: "How did you first find out about us?" and "What made you finally come in?" Log both fields. First answer credits the channel that created awareness; second credits what converted. You've just built two-touch attribution with a pen.
Step 3: the one-page monthly scorecard
One spreadsheet, one row per channel, updated monthly:
- Spend — the invoice, plus real internal labor where it's significant (the person who posts Marketplace all day is a cost).
- Leads — from the CRM, by source.
- Shown appointments — the first number that separates real channels from noisy ones.
- Sold units — matched from the DMS back to lead source at delivery. This match-back is the whole game; if you do nothing else from this article, reconcile deliveries against sources monthly.
- Cost per sold unit — spend divided by solds. The one number that lets a $500 channel and a $5,000 channel argue fairly.
- Gross per channel (optional but revealing) — some channels sell cheap cars thin; front-and-back gross by source occasionally reorders the whole ranking.
Read it quarterly for decisions, monthly for anomalies. One month proves little at small-store volume — a channel that sold two units in March and zero in April didn't die; it's small numbers being small numbers. Trends over three months are where truth lives.
The traps that fool small dealers
- Vendor dashboards grade their own homework. Every platform claims credit generously ("view-through conversions" means someone merely saw an ad and later bought). Count platform reports as directional; your delivery-desk match-back is the tiebreaker.
- Brand-name searches aren't a channel's win. A buyer who Googled your store name after watching your TikToks gets logged as "Google." The two-question interview is what catches this — expect the answers to redistribute credit away from last-click channels toward the ones doing the quiet work.
- Referrals and repeats get robbed. They cost almost nothing, produce your best gross, and vanish inside "Walk-In" at stores with sloppy intake. Give them their own source values and watch how large they really are — the case for investing in reviews and past-customer follow-up usually writes itself.
- Killing a channel for lead quality without checking your own speed. A source with fine volume and terrible shows might be a bad source — or might be leads answered four hours late. Check response time by source before executing the vendor.
- Ignoring halo effects when cutting. Cut a discovery channel and last-click channels sometimes sag a month later. Cut one thing at a time and watch the whole scorecard, not just the line you cut.
What to do with the answers
The scorecard's purpose is reallocation, not decoration: each quarter, shift real budget from the worst cost-per-sold channel to the best one, and re-test the loser later if circumstances change. Most stores that run this honestly for six months find the ranking they'd have sworn to on day one was wrong in at least two places — usually overrating whatever is newest and shiniest, and underrating the unglamorous compounding channels their happiest customers actually came from.
Want to see this working on your own inventory? UCallNow builds AI sales agents, BDC teams, Facebook Marketplace auto-posting and dealer websites for dealerships across the United States — in English and Spanish. Try SOPHIA live or see every solution and price.