the store — The Complete CRM Analysis

One consolidated report: the data, the store's health, what to buy, how to buy it, and when

Assembled 2026-07-13. This document merges every analysis produced from the store's CDK/eLead records into one file. Sources consolidated (all in dealer_crm/reports/ unless noted):

Data basis: 20,707 sold deals extracted from the store's own eLead CRM (March 2014 → June 25, 2026). After removing duplicate bookings and deals with data-entry errors: 20,596 clean deals (12,174 New / 8,427 Used as flagged in the CRM), of which 8,382 deduplicated used deals drive the used-car analysis. Where earlier reports and the 2026-07-13 audit disagree, the audited numbers are canonical and the difference is footnoted.

PART 1 — EXECUTIVE SUMMARY

Why the store is struggling — the causal chain, from its own numbers

1. The store's business model broke, and it was never replaced. the store never really bought used inventory — it manufactured it. Sell a new Chevy, take the customer's 3-year-old Chevy on trade, recondition it, sell it to a payment buyer (45% of all deals carry a trade; the used top-10 is the new top-10 shifted three years). That loop needs new-car volume to feed it. New sales fell from ~1,250/year in 2015–18 to ~850 now, so the loop throws off fewer trades — and the used department starved from 75 cars a month down to 45. The replacement pipeline (going out and buying cars at auction) was never built, so the shrinkage compounded. This is the single biggest thing.

2. The store is hung over from the COVID windfall. 2021–22 paid $3,300– 3,500 per deal — nearly double the historical norm — and $5.9M a year in gross. That kind of money papers over every operational problem. It's now back to ~$2,000 per deal and $2.8M a year, which is below even the 2015–2019 level ($2.9M–$4.0M) on two-thirds of the old volume. A store's expense base — payroll, floorplan lines, the building — gets set during the fat years and doesn't shrink on its own. Half the gross against the same cost structure is the loss, mechanically.

3. The new-car franchise now costs money at the front. In 2026, more than half of the store's new-car deals lose money on the vehicle itself, and the EV inventory GM requires it to carry has lost money on 68–83% of copies (~$169k cumulative on Blazer EV/Equinox EV/Silverado EV). The franchise used to be the profit engine; now it's closer to a customer-acquisition expense that only pays through F&I and loyalty.

4. The store's customer got priced out, and GM discontinued the answer. This store's bread and butter was the payment buyer — the $12–18k Cruze/Malibu/Impala/cheap-Equinox customer at 72 months. Rates went from ~5% to ~9% at the 2023 peak, and GM killed every cheap sedan, so the bottom rungs of the payment ladder are gone. Trax absorbed some of it (which is exactly why it's the fastest riser in the data), but not all.

5. The store is surviving on one leg. The F&I desk now produces 60–80% of every deal's gross. That kept the lights on, but it means the entire store's profitability rides on finance penetration and product sales — one lender-program change or one soft month at the desk and there's nothing behind it. June 2026 is what that looks like: the front end went negative in aggregate and F&I alone couldn't cover a building.

The honest caveat: this analysis sees the store's sales gross, not its expense ledger — so the revenue side of the squeeze is precise and the cost side is inferred. But the shape is unambiguous: fewer cars × thinner margins × mandated losers, against a cost base built for the 2022 store. The fixable piece — the one that doesn't require GM or the rate environment to change — is feeding the used lot, and that is precisely what the auction-acquisition program exists to prove.

Can the data be trusted? Yes — for deciding which models to buy and at what velocity. The extraction is complete (every deal the endpoint claimed, no missing months since Dec 2014, money fields >99.6% populated). It is not deal-level perfect: ~1.5% duplicate bookings, a handful of impossible gross figures, and days-on-lot only exists for deals sold after June 2024. All three are identifiable and were corrected before any number in this report.

How is the store doing? Badly, and structurally — not suddenly. Total sales-department gross fell from $5.9M (2022) to $2.8M (2025). Front-end profit is gone: the typical 2026 new-car deal loses money on the vehicle, and used front gross collapsed a year later on the same curve. The F&I office now carries 60–80% of every deal. The mandated EV lineup (Blazer EV, Equinox EV, Silverado EV) has lost ~$169k cumulative. Recent loss months are the arithmetic destination of those curves, not sudden events.

What does that mean for buying? The store's own history says velocity beats front margin now: buy fast-turning, finance-friendly units thin on the front and let the back end pay. The buy list, in order:

1. Equinox — the volume backbone (72 used/yr, 32-day turn). Buy to replace, don't pay up.
2. Silverado 1500 — #2 gross engine, steady. Watch aged lot units before adding.
3. Trax — the fastest riser (+83% 90-day momentum), 23-day turn, $113/day. Best buy-box match.
4. Suburban — best dollars-per-day in the store ($4.3k/car, 19-day turn). Take every clean 1–3-yr one you can.
5. Traverse — volume with momentum; thin front, back carries it to ~$2k.
6. Tahoe (selective) — $2.8k/car but turns have slowed to ~60 days; buy only sharp, 1–2 at a time.

When? March is the most reliable month in the book (+13%, 9 of 11 years); Oct–Dec is the trough. Buy small/cheap stuff Dec–May, big iron Jul–Aug for the Sep–Jan run, and go light in Sep–Oct.

PART 2 — THE DATA, AND WHETHER IT CAN BE TRUSTED

(Audit performed 2026-07-13, read-only, from the files on disk.)

2.1 What was extracted

2.2 What checks out

• Completeness: 0 duplicate deal IDs; all 20,695 detail records join 1:1 to their deals (the 12-deal gap is exactly the 12 rows missing the join key). Dates run continuously Dec 2014 → Jun 25 2026, no empty months, no future dates. Sale price 99.6% populated; front/back/total gross each missing on exactly one row.
• New/used separates cleanly (12,174 / 8,427); the two files agree within one row in the analysis window.
• The store's numbers cross-validate: three independent reports (best sellers, acquisition board, audit) computed the same model rankings within 1–2 units.

2.3 The four known flaws (all corrected in this report)

1. Duplicate bookings (~1.5%). 99 same-VIN-same-day deal pairs (co-buyer / re-booked deals), 8 pairs inside the last-12-months used window — one a Malibu pair where the duplicate carries an impossible $16,214 back gross. Fix applied: dedup on VIN + sale date.
2. Booking errors in gross. ~47 deals where front gross exactly equals the sale price (or 2× it) — e.g. a Trax showing $28,600 front gross on a $14,300 car. Left in, that one deal inflates the 12-month Trax average by ~$1,200/unit. Fix applied: excluded from money averages (87 deals total with extreme-outlier rule), never from unit counts.
3. Days-on-lot only exists after June 2024 — the DOM report was pulled for a ~2-year window (the endpoint accepts any window; older history was simply never pulled). Coverage is ~95% inside the last-12-months used window, ~0% before mid-2024. 58 rows show negative DOM (re-acquired VINs joined to the wrong stocking cycle). Fix applied: DOM medians use non-negative, in-window values only; anything DOM-based about pre-2024 is not claimed.
4. Every analysis script hardcodes its as-of date (2026-06-25/27 across five scripts), and the data itself ends June 25. All "last 12 months" numbers silently age until a re-pull. Not yet fixed — the single most valuable improvement is a fresh pull plus deriving as-of from the data.

What the data cannot say: wholesale disposals (this endpoint is customer deals only), the store's expense side (payroll, floorplan, fixed ops), and anything about models the store doesn't sell (it's a Chevy store — Camry and Corolla have 1 used sale each in 12 months, so the Manheim quotas for those rest entirely on market studies, not this data).

PART 3 — STORE HEALTH: THE STRUCTURAL TRENDS

3.1 June has shrunk four straight years (identical June 1–25 windows)

In June 2026 the front end lost money in aggregate (−$7,343 front gross for the month); every gross dollar came from F&I.

3.2 The gross engine was cut in half before June happened

Volume is down about a third from the 2017–19 pace, and per-deal gross round-tripped to pre-COVID levels after the 2021–22 scarcity windfall. Roughly $260k of gross per month that existed in 2022 no longer does, while the building's expenses didn't shrink with it.

3.3 The front-end collapse came in two waves

• Wave 1 (2024): new collapsed first. By 2025–26, average front gross per new car is ~$100, the median is negative, and 51% of new deals lose money on the vehicle. The F&I desk answered — new back nearly doubled.
• Wave 2 (late 2025 → 2026): used followed. In 2026, 45% of used deals have front ≤ $100 and 17% lose money on the car and are saved by F&I. Used total gross is genuinely down ~$1,100/unit vs 2024 — the back end cushions but doesn't fully replace the front.

3.4 The mandated EVs lose money on most copies

The current mandated lineup has cost ~$169k in cumulative negative gross. The store doesn't choose this inventory; the franchise agreement does.

3.5 What still works — the case for auction acquisition

• The used department used to retail ~75 units/month (2018–19); it now does ~45 — same rooftop, same brand, bigger repeat-customer base. The capacity is proven; the acquisition pipeline is what shrank. Trades alone no longer feed it (45% of deals carry a trade — the trade-cycle loop that used to manufacture the used inventory).
• Finance penetration is back to 85–86%, and the desk reliably books $1,500–1,850 of back gross per financed deal — which is exactly why velocity-buying front-thin units works.
• "Previous buyer" is 16–20% of every year's deals since 2019; equity-mined deals tripled to 5.4% — the repeat base is real.
• Concentration risk: the top-5 salespeople write 77–81% of used deals.

PART 4 — WHAT SELLS: THE BEST-SELLERS EVIDENCE

4.1 The decade view — why the buy list is stable

The same six vehicles have been the store's best sellers for a decade — Equinox, Silverado 1500, Traverse, Tahoe, Trax, Suburban — because the store manufactures its own used inventory. Every top used seller is a top new seller ~3 years earlier: sell it new, take it back on trade, recondition, sell again to a payment buyer. The median used unit sold is 3 years old for every core model. What changed over the decade is not which cars sell: (1) the cheap sedans (Cruze, Malibu, Impala) died when GM killed them — their payment buyers moved into Trax and TrailBlazer; (2) the money moved from the front of the deal to the F&I office; (3) volume shrank ~40%.

4.2 The ranked table — last 12 months of used retail, audited

(Full sheet: best_sellers_audit_2026-07-13.csv. Windows end 2026-06-25; duplicates removed; booking errors excluded from money averages. Momentum = units last 90 days vs the 90 before — note the prior window includes tax season, so mild negatives on volume models are partly seasonal.)

*On 6 and 3 units respectively — directional at best. Rows 13–20 in the CSV (Yukon XL, Odyssey, RAM 2500, etc.) are PASS: gaudy per-unit gross on 1–4 sales a year is not a demand signal.

Long-window color: almost every model's 12-month gross sits below its 5-year average (the COVID margin washing out) — Trax is the only volume model whose current gross ($2,607) beats both its 3-year and all-time averages. Malibu (565 all-time) and Cruze (501) show why the long windows matter: giant historical sellers that GM discontinued — history alone would overrate them; the 12-month column is the truth.

4.3 Margin compression per model (avg total gross per used unit)

PART 5 — HOW TO BUY: THE ACQUISITION PLAYBOOK

(From the 2026-06-29 Acquisition Board, updated where the 2026-07-13 audit moved a number. The thesis: the desk makes ~$1,500–1,850 back gross on nearly every financed car (85%+ penetration), so a thin front is fine if the unit turns fast. Velocity, not front margin, is the game.)

5.1 The buy-now board (velocity-of-profit ranked)

Tahoe reconciliation note: the June board showed a 28-day median turn (computed over a 2-year DOM sample); the audited last-12-months median is 60 days. Tahoes are turning slower now than their two-year history — consistent with the board's own warning that 4 of 6 on the lot were already aged >60 days. Treat Tahoe strictly as a pick-off, not a program. (Suburban's $/day also differs across reports — $476 vs $225 — because the board took the median of per-deal $/day while the audit divides average gross by median DOM. Both orderings agree: Suburban is the best capital-efficiency buy in the store either way.)

5.2 Spec sheets — hand this to the buyer

The pattern that holds for every model: the 3-year-old unit is the #1 seller every time (Trax 54% of all its sales, Traverse 45%, Equinox 44%); ages 1–3 are ½–¾ of every model's volume; ages 4–6 carry fatter front but far less volume. Buy 1–3-year-olds for velocity; pick off the occasional clean 5–6-year-old for margin.

• Suburban — buy 1–3-yr at ~38–43k mi; into them ~$43–47k; they sell ~$44–49k for ~$3,400 total gross in 3–4 weeks. The rare model strong on BOTH front (~$1,500–2,000 even in the volume band) and back. Clean 5–6-yr units throw $4,600–6,400 — grab them.
• Trax — buy 3-yr at ~24k mi (lowest miles of any core model); into them ~$14,500; sell ~$14,800. Front is only ~$400 but the desk's 85–88% finance penetration makes ~$1,400 on the back. The clearest "pay up to win the car, the back end pays you back" case — and the Manheim buy-box match.
• Tahoe — buy 1–3-yr at ~39k mi; into them ~$37–51k for ~$2,500–3,000 total. 5–6-yr clean units are the margin play ($4,200+). One or two at a time until the aged stack clears.
• Traverse — buy 2–3-yr at ~30–34k mi; into them ~$24–27k for ~$2,000. Thin front, back carries — same playbook as Trax.
• Equinox — buy 2–3-yr at ~26–30k mi; into them ~$17,500 for ~$1,900. The always-liquid staple: replace one-for-one as they sell, never pay up.

5.3 Hold / stop board

5.4 The key tension, stated plainly

Equinox and Silverado make the most TOTAL money; Suburban and Trax make money the FASTEST. A dollar of gross collected in 19–23 days and redeployed beats a dollar that sits 40+. That's why high-total-profit Silverado is a HOLD and lower-total-profit Trax is a BUY.

PART 6 — WHEN TO BUY: SEASONALITY

(11 full years, 2015–2025; index 1.00 = a flat month. Only months consistent ≥8 of 11 years are stated as reliable.)

6.1 The used-volume shape

March is the most reliable month in the book (+13%, above average 9 of 11 years); May–Aug are strong; Oct–Dec is the trough; April dips (the tax bump here is Feb–Mar, April is the hangover). Selling into Q4 also costs ~$300–400/unit of front gross on top of the volume drop.

6.2 The model calendar

Plainly: cheap small stuff sells Feb–Jul; big SUVs and trucks sell Sep–Jan; nothing sells great in Oct–Nov.

6.3 The buy calendar (stock 30–45 days ahead of the sell peak)

Non-effects worth knowing: no month-end gross giveaway and no day-of-week gross pattern (Saturday is just the volume day) — desk discipline on timing is fine. The store sells the same age/miles profile all year (median 3yr/~40k); seasonality changes how many of each model, not what.

PART 7 — RISKS, CAVEATS, METHOD

1. DOM is solid for the last 12 months only (~95% coverage there; the metric doesn't exist before mid-2024). Turn-speed claims about earlier years are not made anywhere in this report.
2. Momentum on small samples is directional. Corvette +500% is 6 units; Blazer +200% is 3. Any model under ~15 units/yr: the units column is the signal, the momentum column is noise.
3. The 2026 front-gross collapse is real in direction, wobbly in level — monthly cells ride on 26–58 deals. One open question for the GM: whether the 2024 ownership change moved pack/recon policy (which would shift booked front without market change).
4. Gross ≠ P&L. This is sales-department variable gross as booked in eLead. No expenses, no floorplan interest, no fixed ops, no wholesale disposals. Part 3 is framed accordingly.
5. The dataset ends 2026-06-25 and five scripts hardcode that date. Re-pull before any decision made after July; make as-of derive from the data's max sold date.
6. Method, one line each: money parsed with accounting negatives (($1,136) = −$1,136); dedup on VIN+sale-date; 87 booking-error deals excluded from money averages only; medians quoted wherever a mean could be skewed; every window anchored to 2026-06-25; momentum = last 90 days vs prior 90; $/day = avg total gross ÷ median DOM.

APPENDIX — REPRODUCTION & SOURCES

• Ranked sheet: dealer_crm/reports/best_sellers_audit_2026-07-13.csv
• Print-ready acquisition board: dealer_crm/reports/used_acquisition_board/the store_Chevrolet_Used_Acquisition_Board_2026-06-29.pdf
• Model deep-dive dashboards (HTML): dealer_crm/reports/{camry_combined,trax_combined,chevrolet_trax}/
• Scripts: dealer_crm/scripts/ (build_acquisition_report.py, seasonality_deep_dive.py, auto_ledger_engine.py) — note the hardcoded as-of dates before re-running.
• Independent verification: paste codex_review/crm_bestsellers_recheck_2026-07-13.md into Codex to re-derive the audit numbers and top-10 ranking from scratch.