DealScanGreat Deal≥ 85
Strong buyer-friendly signals; still verify title, VIN, records, and condition.
Methodology
How the deal score works.
Every scan produces a 0–100 score and a verdict: Great Deal, Decent Deal, Proceed with Caution, Red Flags Present, or Avoid. The score comes from eight factors read from the listing itself. Here is what moves it and what the result cannot verify.
How the analysis works
DealScan first creates a structured local baseline from the listing text. When configured, Groq applies the scoring rubric with broader language understanding. If Groq is unavailable, DealScan can use configured Gemini access and then the local heuristic so the user still receives a clearly labeled result. The result screen identifies when local analysis was used.
Rubric limitations
- Keyword-based heuristics: The local analyzer (and the AI when it follows the same rubric) matches phrases — it does not see the car, verify documents, or run a history report.
- Absent information is not a defect: A short listing lowers confidence, not the score. Missing details appear in the checklist so you can ask the seller.
- Green-flag cap: Positive keywords are capped at +20 points so a keyword-stuffed ad cannot reach a perfect score.
- Heuristic ceiling: The local analyzer never exceeds 92 so AI-enhanced scoring always has room to add nuance.
- Price estimates are rough: Fair-value ranges use basic depreciation curves, not licensed market data. Treat them as conversation starters, not appraisals.
Verdict thresholds
The score maps to one of five verdicts using these exact cutoffs:
Decent Deal70–84
Workable listing; score depends on seller proof and an inspection.
Proceed with Caution55–69
Enough uncertainty to justify asking for proof and offering below asking.
Red Flags Present40–54
Red flags detected. Do not move forward without title proof, VIN history, and a mechanic inspection.
Avoid< 40
Major risks. Avoid unless a trusted inspection and title check resolve them.
Factor 1
Asking price
The listed price is weighed against the vehicle's age, mileage, and condition language to judge whether it sits inside a reasonable range.
Factor 2
Mileage for age
Around 12,000 miles per year is typical. Far above that lowers the score; unusually low mileage with no explanation raises a question instead of a bonus.
Factor 3
Title and ownership
Clean title with the seller's name on it is the baseline. Salvage, rebuilt, lien, missing, or "title in hand soon" language is one of the heaviest penalties.
Factor 4
Mechanical risk language
Phrases like "needs work," "runs rough," "check engine light," or "as-is" are scored against the asking price — a cheap car that admits problems can still score poorly.
Factor 5
Seller transparency
Listings that volunteer the VIN, service records, accident history, and clear photos score higher than vague ads that ask you to "text for details."
Factor 6
Missing information
Every key fact the ad leaves out — price, mileage, title status, location — lowers confidence and shows up in your missing-info checklist.
Factor 7
Positive proof points
One owner, maintenance receipts, recent tires or brakes, and a clean history report offered up front can add points back, with a cap that limits keyword stuffing.
Factor 8
Negotiation opportunity
Time on market, price-drop language, and condition admissions feed the suggested offer range and the negotiation tip you get with each scan.
What the score cannot know
The score reads the listing — it cannot see the car. It does not replace a vehicle history report, a pre-purchase inspection by a mechanic, or a title check with your DMV. Price ranges are estimates based on listing details and basic market heuristics, not licensed market valuations. A high score means the listing looks honest and fairly priced on paper; it is the start of your homework, not the end of it.
Scan a listing