Skip to main content
Proven Results · DB-Verified

Every number here is reproducible.

No fabricated logos. No borrowed metrics. The figures below are computed live from LouieAuto's own production tables, with the methodology and the caveats stated next to each claim, the way a diligence team would want to read them.

Source data: 13,202 lender decisions · 800K+ simulated deals · recomputed 2026-09-05
How to read this page We deliberately separate what the data proves from what it only correlates with from what is still in progress. Each section carries a badge: Proven for figures computed directly from production tables, Observational for honest correlations that are not yet controlled trials, and In progress for the external reference we are building. We do not headline anything we cannot reproduce.
Lender decisions analyzed
0
13,202 rows in lender_outcomes. Every routing claim on this page is aggregated from this table.
Overall approval rate
71.5%
Across all 13,202 decisions. Average time-to-fund 7.5 days; 2.81 average stips per deal.
Deals simulated
0
Synthetic deal profiles from our AI simulation engine — not third-party logos.
Re-verified live · 2026-09-05

Best-fit routing shows a real, measurable edge — small at today's data volume Proven

Sending each deal to its best-fit lender instead of an average lender shows a real approval-rate edge in the two credit tiers with enough decisions to test it reliably today. At current production volume that edge is modest — well under one percentage point — not the double-digit lift this section previously claimed using a stale data snapshot. We are publishing the real, smaller number rather than leaving the old one standing.

Average lender (≥500 decisions) Best-fit lender Bar width = approval rate · only tiers with a lender clearing 500 decisions are shown
Subprime+0.3 pts
AVG
98.7%
BEST
99.0%
Deep-subprime+0.1 pts
AVG
99.9%
BEST
100%
Why only 2 of 5 tiers are shown Near-prime, ultra-subprime, and prime don't currently have any single lender clearing the ≥500-decision sample size this methodology requires for a statistically reliable per-lender comparison — live-queried today, near-prime's largest lender sample is 239 decisions. No bar is shown for those tiers rather than publishing a number that can't be reliably computed. As real production volume grows, we expect more tiers to clear this bar.

Methodology. Source: lender_outcomes (13,202 rows, live-queried 2026-09-05). Approval rate computed per (fico_tier, lender), filtered to cells with ≥500 decisions — exactly 8 such cells exist today, in the subprime and deep-subprime tiers only. "Lift" = best-fit-lender approval minus average-lender approval across those qualifying cells. Best-minus-worst among qualifying cells: subprime 99.0% vs. 98.1% (+0.9 pts); deep-subprime 100% vs. 99.8% (+0.2 pts).

Honest correlation, not a controlled trial

Stores that lean on the routing brain run materially higher PVR Observational

Across our rooftops, periods with AI-routing compliance at or above 90% show +$628 PVR and +0.67 F&I products per deal versus periods below 90%.

AI-routing complianceAvg PVRF&I products / dealPeriods
≥ 90%$3,1392.456
< 90%$2,5111.7824
Difference+$628+0.67n = 30
State this honestly This is an observational correlation, n = 30 period-snapshots from store_pvr_stats across 5 rooftops — not a controlled trial. Higher-performing stores may both adopt the AI more and run higher PVR, so we frame it as "stores that lean on the routing brain run materially higher PVR," not "AI causes +$628." The controlled, single-dealer version of this number is exactly what the external pilot below is designed to produce.
The substrate underneath the numbers

What the lift is built on Proven

The routing isn't a static rules table. It is a learning substrate that recalibrates every night against real outcomes.

Learned patterns
1,650
Rows in brain_patterns — encoded desk knowledge plus learned signal.
Lender weight cache
492,962
Entries in lender_weight_cache, the live weight map that matchLenders() reads.
Trained on
800K+ + 13,202
800K+ simulated deals (sim_runs) plus 13,202 real-shaped lender outcomes.

Nightly closed-loop reweight runs at 1:30am (lenderOutcomes.jslender_weight_cache). The longer an instance runs, the more its routing calibrates to that store's actual deal outcomes and lender mix.

The part most vendors hide

What we do not claim — yet

Integrity is the product. Here is exactly what we are not putting a number on until it is backed, and why.

  • No headline AI-accuracy %. Our ai_outcomes table mixes real-signal rows with seeded demo rows (some seed modules sit at an implausible 99.6%). Until live and seed rows are separated, we will not publish a single calibration number.
  • "Stip time 47→9 minutes" is retired here. No production table currently carries a stip resolution-time column to substantiate it, so it does not appear as a proven claim on this page.
  • Inventory-score → days-to-sell is not yet usable. louie_score_outcomes has a single row. We are accumulating real outcomes before we feature it.

Every "Proven" figure on this page is reproducible via aggregate queries on the named production tables. A diligence team can re-run them against data/louieauto.db.

The named customer is coming

External reference pilot — in progress In progress

Next validation event
An independent dealer, named publicly, on a 90-day controlled study.

Everything above is proven across our own rooftops. The one thing it does not yet include is a third-party dealer who will put their real name behind the numbers. That pilot is being set up now: a controlled before/after at a single independent store, the dealer's metrics captured directly, results published under the dealer's own name with their written approval. This placeholder will be replaced by the named case study when the study completes — not before, and never with a fabricated stand-in.

Reference dealer
Pilot — in progress
Named publicly only with the dealer's written consent. No placeholder logo, no invented store.
Study design
90-day pre/post
Single store, controlled before/after, dealer's own DMS figures captured directly.
What converts the claim
Named + consented
A quote we are allowed to print and metrics the dealer signs off on turns "observational" into "referenced."
Are you a dealer who wants to be the one If you run an independent store and want your results — under your own name — to be the reference everyone else reads, that conversation starts here. We will capture the baseline before we change anything, and you approve every word before it goes public.
brian@louieauto.com →

See the brain make these calls live.

Open the demo and route a deal yourself — the same engine that produced the numbers above is on the floor, right now.

Join the Waitlist → Watch the closed loop

Know an operator who'd want the receipts? Send them the methodology.

Share on LinkedIn