Block Buddy
Live
Demo Mode — Sample Data
Know the number
before the gavel falls.
before the gavel falls.
Block Buddy prices every car on your run list overnight — real sold comps, condition, recon,
transport, fees — and hands you one number per car: the most you should pay.
Then it places the bids for you, up to your ceilings, never a dollar over. Run it fully
autonomous, or approve each bid yourself — your call.
Your best sellers, torn apart→
Your buy box→
Overnight scan→
Priced candidates→
Bids placed automatically→
Won under your number→
P&L
Use ← → or the buttons below to move through the walkthrough · 10 steps · ~12 minutes
Step 1 of 10 · The Problem
Sixty seconds. That's all the block gives you.
The auctioneer doesn't wait while you thumb through comps. Most buyers bid on instinct — and instinct is expensive.
Lanes running on a peak day300+across Manheim locations
Lanes one human can watch2–3the attention bottleneck
Time per car on the block~60 secto judge, price, and bid
Cost of emotional bidding$300–800per car, over the walk-away number
What actually goes wrong
- ✗Buy-box drift — you have rules in your head; on the block, you violate them.
- ✗Comp blindness — "it felt like a $24 car" is how aged inventory is born.
- ✗Recon surprise — the CR said 4.0; the shop bill says $1,800.
- ✗Landed-cost blindness — hammer price isn't your cost. Fees, transport, and recon are.
- ✗The missed car — the Trax at $1,200 back of the money ran in Lane 9 while you watched Lane 4.
"Most buyers do the homework on 20 cars and guess on the other 1,000.
We do the homework on all of them."
We do the homework on all of them."
Every number in this demo is real output from a live sale-day run and a real store's data — nothing here is mocked-up marketing math.
Step 3 of 10 · Buy Programs
The teardown becomes your buy box. In writing.
Everything the analysis found gets written down as buying programs: which cars, what condition, how far away, how many — and the pricing rule for the max bid. The Trax program below exists because the teardown said Trax. The bot follows it exactly — no drift, no exceptions, no "it felt right."
Chevrolet Trax
Live · AutonomousChevrolet Equinox
Approval Req'dToyota Corolla
ShadowProgram modes
Off — program disabled
Shadow — prices & alerts, never bids (how every program starts)
Approval Req'd — semi-manual: you tap Approve, it handles the rest
Live — fully autonomous: it bids up to your ceilings on its own, capped & kill-switched
Quotas are enforced against confirmed purchase invoices — when the second Trax is actually bought, the program stops bidding on Traxes automatically. It counts pending bids too, so it can never accidentally win three.
Step 2 of 10 · Store Analysis
Step one isn't software. First, we pull your best sellers apart.
Before anything gets configured, we take your CRM export and rebuild your store's complete sales history. This is the real one we delivered for Winner Chevrolet — every deal audited before we trusted a single number, then every model ranked on what actually makes money, and how fast.
Deals analyzed20,707every sale, 2014 → today
Used deals ranked8,382deduplicated retail deals
Data flaws found & corrected4before we trusted the averages
Independent re-checks2same rankings, re-derived from scratch
The teardown — five parts, delivered as a written report you keep
1 · Data audit
Can your numbers be trusted? We found & corrected 4 flaws here before using a single average.
2 · Store health
Volume, gross, front-end vs. F&I — the structural trends behind the monthly numbers.
3 · What sells
Every model ranked on gross per day of lot time, with a BUY / HOLD / PASS call and the why.
4 · How to buy
Per-model buy specs you can hand a buyer: target years, mileage band, price to pay.
5 · When to buy
A month-by-month calendar — stock each model 30–45 days ahead of its selling peak.
Best sellers — last 12 months, used retail (audited)
| Model | Units | Avg total gross | Median turn | $ / day | Call |
|---|---|---|---|---|---|
| Equinox | 72 | $2,065 | 32d | $64 | BUY |
| Silverado 1500 | 56 | $2,275 | 38d | $60 | HOLD |
| Traverse | 43 | $1,934 | 40d | $48 | BUY |
| Trax | 25 | $2,607 | 23d | $113 | BUY ▲ +83% |
| Tahoe | 24 | $2,781 | 60d | $46 | Selective |
| Suburban | 19 | $4,277 | 19d | $225 | BUY |
"$ / day" = gross per day of lot time — the number that actually compounds. The #2 earner (Silverado) is a HOLD because the units on the lot today average 123 days old.
Velocity beats margin. The Suburban makes $4,277 a copy and turns in 19 days — $225/day. Money that comes back fast gets reinvested. The ranking is built on $/day, not bragging-rights gross.
The Trax is the sleeper. Modest gross, but 23-day turns, the only model whose gross is holding vs. its 5-year average, and +83% sales momentum in the last 90 days. That's why it's the demo car for the rest of this walkthrough.
The trade pipeline dried up. This store's used volume fell from ~75 to ~45 units/month as new-car volume (and its trades) declined — and no auction pipeline replaced it. That's the gap this tool fills.
What the report looks like — executive summary, page one (actual report · store name removed)
Store Analysis · Executive Summary
the data, the store's health, what to buy, how to buy it, and when
Why the store is struggling — the causal chain, from its own numbers
- The store used to manufacture its own used inventory. Its used best-seller list is its new-car list, three years later — nearly half of all deals carried a trade.
- New-car volume fell from ~1,250 to ~850 a year, and the trades fell with it. The used department starved: ~75 units a month down to ~45.
- Total gross halved — $5.9M to $2.8M in three years. Roughly $260,000 a month of gross, gone.
- The front end is thin and F&I carries the store, so the winning play is fast-turning, financeable cars — bought right.
- No auction pipeline was ever built to replace the trades. That is the gap, and the reason for the acquisition program in Part 5.
Can the data be trusted?
Yes — after four flaws were found and corrected. Every claim is reproducible.
Yes — after four flaws were found and corrected. Every claim is reproducible.
How is the store doing?
Volume and gross are roughly half of peak; per-deal gross is back to pre-COVID levels.
Volume and gross are roughly half of peak; per-deal gross is back to pre-COVID levels.
What should it buy?
Velocity — the models that turn in under ~30 days and finance well.
Velocity — the models that turn in under ~30 days and finance well.
When?
Stock each model 30–45 days ahead of its seasonal peak; the trough months are for restraint.
Stock each model 30–45 days ahead of its seasonal peak; the trough months are for restraint.
Total gross booked, year over year — the whole story in one chart
'153.8
'164.0
'173.8
'183.1
'192.9
'203.5
'215.8
'225.9
'234.8
'243.1
'252.8
'26*≈2.6
$M total gross per year. And the why, from the data: the 2021–22 peak was the scarcity
windfall — $3,300–3,500 per deal, nearly double the norm. It receded on schedule, but the store's
expenses didn't. Underneath it, new-car volume fell ~1,250 → ~850/yr, taking the trades — and the
used department it fed — down with it. Half the gross against the same building. *2026 = pace.
This lands as a written report you keep, whatever you decide about the bot. For your store it's the first deliverable: your CRM export in, your best sellers torn apart, your buy list out — before any bidding is ever configured.
Step 4 of 10 · Candidates Board
Overnight, it does the homework on every car in the sale.
The night before a sale, Block Buddy scans the full run list — a real June run priced 1,110 cars before breakfast — and puts every match on this board with a max bid and the reasoning behind it.
Showing 4 of 1,110 priced · Sale: Manheim Pennsylvania · Buy program: Chevrolet Trax ×2
numbers from the real June '26 Trax study
Lane 6 · Run 042Strong
2024 Chevrolet Trax LT
Max Bid
$18,700
Retail $20,727 − costs − $950 profit · under MMR cap
BIDGreen lightClean title121 sold comps
Lane 6 · Run 088Good
2024 Chevrolet Trax 2RS
Max Bid
$20,650
Retail $22,558 − costs − $950 profit · under MMR cap
BIDGreen lightClean title207 sold comps
Lane 6 · Run 101Priced Out
2024 Chevrolet Trax LS
Ceiling $18,200 — bidding already past it
No Bid
Next bid $18,400 > our ceiling. We don't chase.
PRICED OUTGreen lightClean title
Lane 2 · Run 067Rejected
2024 Chevrolet Trax LT
Condition gate failed
No Bid
Red announcement light — hard reject, no price is cheap enough.
Red lightCR below 4.2SKIP
Real auction photos, straight from the sale. Every card links the full condition report and photo set — the recon dollars are priced from the CR grade and its listed damage, and your eyes get the photos before anything is approved. Every skip has a written reason; every max bid has an audit trail. Try the ✓ / ✗ pills — that's the whole morning workflow: coffee, scroll, approve.
Step 5 of 10 · The Market Study (Chevrolet Trax)
Before we bid on a car, we become the expert on that car.
It starts with the deepest source there is: every car this store has ever sold, since the day it opened. Then it layers on a 100-mile radius of the outside market — every sold transaction (not asking prices — sold prices), every live competing listing, and days-on-market. That's how it knows what the car retails for, how fast it turns, and how crowded the market is — before the first bid.
Market Report · 100 mi of your store
CHEVROLET TRAX
don't skip this —
the whole report, plots & all
Open the full Trax study →
the whole report, plots & all
Radius100 miaround your ZIP
Sold comps1,933actual transactions
Live listings321your competition today
Median DOM29ddays on market
Market readBalancednot flooded, not starved
By the time this car crosses the block, we know it better than the seller does. Then we go buy it. The button above opens the actual two-tab Trax study built for this store, in its entirety — retail market on one tab, the store's own Trax history on the other. Store name removed, everything else untouched.
Trax study · sold comps within 100 mi · June 2026
| Year / Trim | Sold comps | Retail value | Sold band | Days to sell | Trust |
|---|---|---|---|---|---|
| 2024 LT | 121 | $20,727 | $20,244–22,490 | 35d | High |
| 2024 2RS | 207 | $22,558 | $20,999–23,794 | 30d | High |
| 2024 ACTIV | 73 | $22,951 | $22,000–24,288 | 38d | High |
| 2024 LS | 110 | $20,238 | $19,297–21,068 | 27d | High |
| 2022 LT | 91 | $17,611 | $16,784–19,694 | 27d | Medium |
| 2025 (all trims) | < 8 | — | — | — | No number → no bid |
The trust rule is hard: fewer than 8 sold comps → the bot refuses to price the car at all. It also adjusts for mileage (this cell: about −$54 per 1,000 miles) and re-checks Manheim MMR live at bid time, never from a stale cache.
Rooted in the store's own books — every sale since it opened
The store's history isn't a footnote — it's an anchor. If this store's own Traxes retail below what the outside comps say, the study takes the lower number. What YOUR customers actually paid outranks what the market brags about.
Cross-check #2 — this exact car, not the average
Three independent sources, one answer. Sold comps say ~$20,700 retail for a '24 LT. The store's own history says these turn in ~23 days. Live supply says the market isn't flooded. When they disagree, the bot takes the most conservative number.
Step 6 of 10 · How the Max Bid Is Built
The number: work backwards from the exit, protect the profit.
Same 2024 Trax LT from the board. Start with what it sells for, subtract every real cost of getting it front-line ready, subtract the profit you insist on keeping — what's left is the most you can pay. Not a feeling. Arithmetic.
Ceiling #1 — the profit ceiling
Retail exit study value · 121 sold comps
$20,727
Auction buyer fee flat
− $400
Transport Manheim PA → your lot (real per-site table)
− $250
Recon priced from the CR grade (4.5)
− $400
Your minimum profit non-negotiable floor
− $950
Profit ceiling
$18,727
Ceiling #2 — the market redline
Even when the retail math says you could pay more, the bot won't bid meaningfully past what the wholesale market says the car is worth. Two ceilings, and the lower one always wins.
AdvisoryStrong
$18,700
Suggested max bid · min($18,727, $19,350), floored to Manheim's $50 increment
$950+ protected profit
plus every dollar the hammer lands under the max
Glass box, not black box. Every recommendation stores the exact operands — retail, comps count, fees, transport, recon, MMR — so you can audit any bid it ever suggested, months later.
Step 7 of 10 · Safety & Discipline
What makes it trustworthy is how often it says no.
Every unknown fails closed. No market study? No bid. Can't price the transport? No bid. Accident count unknown? No bid. The bot never fills a gap with optimism.
Hard gates — never overridden by price
- ✗Red announcement light
- ✗Frame damage
- ✗Any accident history — unknown counts as dirty
- ✗Branded / unclear title, odometer problem
- ✗Condition grade below your floor (CR < 4.2)
- ✗Farther than your transport radius
Spending caps — checked before every bid
- ✓Max dollars per car
- ✓Max bids per day
- ✓Max total dollars per day
- ✓Quota per model — invoice-confirmed, stops at "up to N"
- ✓Never two bids on the same VIN in a day
- ✓Priced out? It walks. It never chases past the ceiling.
Three independent kill switches
- ■The STOP button on this dashboard
- ■A kill file on the machine — halts everything instantly
- ■Per-program off switch
Go ahead — click it. Any one of the three halts all bidding, instantly, no questions asked.
Of the 1,110 cars priced that night, the board recommended bidding on a handful.
That ratio is the product. Anyone can build a tool that says "buy." This one earns its keep saying "walk."
That ratio is the product. Anyone can build a tool that says "buy." This one earns its keep saying "walk."
Step 8 of 10 · The Bid
It doesn't race the auctioneer. It places your ceilings hours before the sale — automatically.
When a car clears the math, Block Buddy places the max-bid ceiling into Manheim's pre-sale proxy bidding on your account — hours before the lane runs, never in the heat of the moment. Manheim's auctioneer then bids on your behalf up to your ceiling. Not a dollar over, ever.
The bid queue — the morning of the sale
2024 Chevrolet Trax LT · Lane 6 Run 042
Ceiling $18,700 · Trax program (autonomous) · 0 of 2 filled
2024 Chevrolet Trax 2RS · Lane 6 Run 088
Ceiling $20,650 · semi-manual — waiting on you
Top row: a program running fully autonomous — the ceiling went in on its own. Bottom row: semi-manual mode — same math, but it waits for your tap. Mix modes per program. Every bid gets an audit record written before it's placed.
How a bid goes in
- 1.The car clears the gates, the math, the caps, and the quota.
- 2.Autonomous mode: the ceiling is placed straight into Manheim's proxy system. Semi-manual mode: it waits for your Approve first.
- 3.An independent verifier re-reads the proxy list and confirms the exact ceiling stuck on the exact car — VIN-matched, to the dollar.
- 4.Manheim's auctioneer carries the bid for you when the car crosses the block. Above your ceiling, you're out — automatically.
No timing race, no snap judgments. Live in-lane bidding gives you 60 seconds. The proxy ceiling is set hours early, with the math already done — a disciplined buyer's process, executed on every car instead of the three you'd have time for.
Step 9 of 10 · Bought → P&L
The gavel isn't the finish line. The P&L is.
Block Buddy watches your purchase invoices, confirms every win, fills the quota, and books the deal — first as a projection, then as realized gross when the car actually retails.
Deal ledger — this car (illustrative exit)
| Hammer price | $18,400 |
| Buyer fee | $400 |
| Transport | ≈ $250 |
| Recon (CR 4.5 band) | ≈ $400 |
| Cost basis | ≈ $19,450 |
| Retail exit — sold day 32 · $20,995 + $1,214 F&I | $22,209 |
| Realized total gross | + $2,759 |
The ≈ marks estimates and is replaced by actuals as they land. If a number is missing, the ledger shows an honest floor or "exit unknown" — never a made-up zero.
Why the loop matters
- ✓Invoice truth. Wins are confirmed from your purchase invoices — the actual paper — not from watching the lane and guessing.
- ✓Quotas self-enforce. Trax 1 of 2: one more and that program stops bidding on its own.
- ✓Projected → realized. When the car retails, your CRM's sold record flips the deal to actual gross. Every buy gets a report card.
- ✓The studies learn. Realized results feed back into what the bot hunts next month.
This is the forensics most stores never do. "Did that auction car actually make money?" stops being a debate in the sales meeting and becomes a line on a ledger.
Step 10 of 10 · Recap
A disciplined buyer's process — on every car, every sale, without the fatigue.
Your CRM says what to buy→
Your buy box sets the rules→
Overnight scan prices the sale→
Sold comps + costs + your profit floor = the number→
Ceilings placed automatically→
Manheim's proxy bids to your ceiling→
Invoice-confirmed → P&L
What it is
A buy-side copilot that does the homework on every car in the sale and holds the line on your numbers. It watches everything, prices everything, and recommends rarely.
What it is not
Not a gamble. Even fully autonomous, it's boxed in: per-car, per-day, and per-model caps, invoice-confirmed quotas, and three kill switches under your hand. And if you'd rather tap Approve on every bid, semi-manual mode does exactly that.
What it takes to start
A CRM export so we can tear your best sellers apart first, your Manheim account, and about 45 minutes to turn the findings into your buy box. First sale runs in shadow — it prices everything and bids nothing, so you can watch it be right before you turn it loose.
Overnight homework. In-lane confidence.
The next sale is the demo. Run it in shadow and check its numbers against your buyer's gut — car by car.