V-Score Card

TICKER:         BILL
V-SCORE:        2.30
VERDICT:        AT_RISK  [2.0, 3.0)
κ (conviction): (2.30 − 3.0)⁺ = 0
GATE 1 (E>1):   PASS  (E=2 > 1)
GATE 2 (A>1):   PASS  (A=2 > 1)
FAST SCREEN:    1/3 — regulatory lock-in only (no proprietary data moat, no transaction embedding)

DIMENSIONS
  C (w=0.25):  3  Risk engine ($1T+ TPV) + 6 two-way accounting integrations; but core AP/AR is AI-native — BILL's own agents eliminate 90% of steps (10-K ln 393-396, Q2 FY26 transcript)
  E (w=0.22):  2  Money transmitter licenses all states + FinCEN + card networks, but FUNGIBLE: 14% annual churn, NRR 94% declining, RPO 4% of revenue, 3-6wk switching (10-K ln 912-918, 4781-4802)
  U (w=0.18):  3  8-11 workflows (AP, AR, Spend/Expense, Payments, Financing, Cash Mgmt, Vendor Mgmt, Procurement) concentrated in finance/accounting (10-K ln 417-616)
  A (w=0.12):  2  APIs functional, Embed 2.0 live (NetSuite, Acumatica, Paychex); agents route bank→ACH, QuickBooks→native AP, bypassing BILL middleware (10-K ln 660-681)
  M (w=0.15):  3  8.3M network members, 9,500 accounting firms, $1.65B revenue (5-10x nearest AP competitor); switching in weeks, not years (10-K ln 328-330, 713-715)
  F (w=−0.06): 2  Self-serve, mobile-native, AI reducing friction (self-serve 13%→40%); "several months to adapt" and Divvy credit underwriting add moderate overhead (10-K ln 633-636, 4777-4778)

CALCULATION:
  Raw = 0.25(3) + 0.22(2) + 0.18(3) + 0.12(2) + 0.15(3) − 0.06(2)
      = 0.75 + 0.44 + 0.54 + 0.24 + 0.45 − 0.12
      = 2.30
  Gates: PASS · PASS = 1
  V = 2.30

REGIME CONTEXT (T = 15wk, Dec 16 – Mar 27):
  IR = α̂ / σ_idio = 0.47   (t = 0.24, p = 0.81 — NOT significant, zero signal)
  ρ_intra = 0.41            (falling: 0.55 → 0.37, exiting correlated selloff)
  δ = V − V_market ≈ 0      (no idio discount — decline is 100% factor-driven)

THERMODYNAMIC SUMMARY: Half-infrastructure (licensed payments), half-exposed (AI-native AP/AR software). The regulatory stack prevents COLLAPSED; the collapsible software layer prevents EMBEDDED. Intelligence flows around BILL, not through it.

DURABLE REV:   ≈50% — payment processing fees requiring licensing (ACH origination, card interchange, cross-border, invoice financing) + float income on $4B deposits
EXPOSED REV:   ≈50% — AP/AR subscription/transaction fees (invoice capture, approval routing, accounting sync), expense management software margin, commodity payment routing

Dimension Analysis

E — Irreducible Infrastructure: Scored 2, Not 3

The original batch scored E=3. Stress testing downgrades to E=2. The evidence:

The infrastructure is real but fungible. Money transmitter licenses across all U.S. states, FinCEN MSB registration, card network compliance (Visa/MC/Amex), NACHA, lending licenses — this is a 2-3 year regulatory stack to replicate. An AI agent cannot process payments without routing through a licensed entity. c_ℓ(payment_processing) = ∞ for unlicensed entities.

But BILL is A licensed entity, not THE licensed entity. Tipalti, Stampli, Ramp, Brex, Melio, and every major bank hold equivalent licenses. No regulation mandates routing through BILL specifically — contrast NRSRO designation (S&P Global, Moody's) or state statistical agent mandates (Verisk). Multiple competitors offer substitutable licensed infrastructure.

The retention data confirms fungibility. NRR 94% and declining (111% → 92% → 94% over FY23-25). Logo retention 86% — 14% annual churn (10-K ln 4781-4802). RPO $65M on $1.65B revenue = 4% contractual lock-in (10-Q ln 832). Infrastructure companies with truly irreducible positions show NRR >110% and GRR >95% (ADP, Workday, ServiceNow). BILL's retention profile matches the E=2 rubric calibration: "Cloud wrapper with some lock-in. NRR declining."

Switching timeline: 3-6 weeks. Re-collect vendor bank details (2-4wk), re-integrate accounting software (1-3d), migrate history (1wk), retrain staff (1-2wk), rebuild card program (1-2wk). Compare Workday (12-18mo), Okta (6-12mo). BILL's switching cost is an order of magnitude lower than E=3 comps.

E=2 is the honest score. The regulatory stack provides a floor above E=1 (local model can't process payments). But the fungibility of that infrastructure — demonstrated by 14% churn, 3-6wk switching, declining NRR, and thin RPO — places BILL firmly at E=2.

C — Compound Cognition: Scored 3, Under Pressure

C=3 holds today. It won't hold in 18 months.

What's load-bearing: The risk engine trained on $1T+ TPV (stopped 5.3M fraud attempts in H1 FY26, reduced manual reviews 40%). Six two-way accounting integrations (QuickBooks Online/Desktop, NetSuite, Sage Intacct, Xero, Dynamics 365). Inter-module compound effects: AP data → risk models → invoice financing eligibility; W-9 agent → vendor management → 1099 generation.

What's collapsing: BILL's own agents are the prosecution's star witness. The coding agent reduces invoice processing 90%. The W-9 agent automates 80% of manual steps. BILL Assistant tripled self-serve from 13% to 40% (Q2 FY26 transcript). If BILL's own AI can eliminate 90% of its core workflow steps, an external agent with the same accounting API access achieves equivalent results without the platform.

Re-derivation timeline by component: Core AP/AR workflow (60% of value): weeks — AI-native tasks, proven by BILL's own agents. Risk engine (25% of value): 6-12 months — domain-specific fraud patterns require training data, but JPMorgan and Visa have larger corpora. Accounting integrations (15% of value): 3-6 months per provider at human pace, compressing to weeks with AI-assisted development.

C=3 today, trending C=2. The risk engine + integration breadth genuinely exceed DocuSign/UiPath (C=2 comps). But frontier models close the gap on fraud detection and API integration within 12-18 months. The core is already AI-native.

U, A, M, F — Stable

U=3 holds cleanly. 8-11 workflows concentrated in finance/accounting. The switching cost superlinearity is real within the domain (replacing AP + AR + expense + payments + vendor management simultaneously is harder than replacing any one). But single-department concentration caps U at 3.

A=2 is correct. BILL has functional APIs and Embed 2.0 partners (NetSuite, Acumatica, Paychex — "3 partnerships alone unlocked potential close to 1 million businesses," Q2 FY26). But agents route bank → ACH, QuickBooks → native AP, Plaid → bank connections, Stripe → payments. BILL is middleware agents skip. The accounting firm channel (90/100 top firms) is human distribution, not agent routing.

M=3 is correct. 8.3M network members with vendor bank details on file create real migration friction. 9,500 accounting firm partnerships create relationship gravity. $1.65B revenue at 5-10x nearest AP competitor. But: switching in weeks, NRR <100%, no developer ecosystem or app marketplace. Meaningful gravity, not deep gravity.

F=2 is correct. Modern SaaS, clean APIs, AI actively reducing friction.

Sensitivity Analysis

No scenario reaches V ≥ 3.0. BILL cannot enter the basket under any plausible dimension movement. The path to EMBEDDED would require upgrading 3+ dimensions simultaneously (E→3, A→3, M→4), which requires Embed 2.0 to fundamentally change agent routing AND NRR to recover above 110%. Neither is probable within 12 months.

Regime Analysis (T = 15 Weeks)

Factor regression (BILL = α + β₁·SPY + β₂·IPAY + ε):

Intra-sector correlation:

ρ is falling. The indiscriminate phase is behind us. Dispersion is emerging — but BILL shows no idio signal in either direction.

The zero-idio finding: BILL declined -31.7% over 15 weeks. Cumulative idiosyncratic return = 0.00%. The entire decline is factor-driven: β_SPY(1.0) × SPY(-6.3%) + β_IPAY(1.45) × IPAY(-20.7%) explains the full move. The market is not penalizing BILL specifically — it's penalizing fintech uniformly.

δ = V − V_market ≈ 0. No structural mispricing detected. The market prices BILL as a high-beta fintech stock in a fintech selloff, which is exactly what it is. V = 2.30 (AT_RISK) and the market agrees through factor pricing.

Conviction Weight

κ = (V − 3.0)⁺ = (2.30 − 3.0)⁺ = 0

w_BILL = 0

Zero weight. BILL does not enter the basket at any allocation.

Basket Verdict: FILTER

V = 2.30 | AT_RISK | κ = 0 | FILTER

BILL is half-infrastructure (licensed payments, ≈50% of revenue protected by regulatory requirements) and half-exposed (AI-native AP/AR software, ≈50% of revenue vulnerable to agent substitution). The regulatory stack prevents collapse to V < 2.0. The collapsible software layer prevents elevation to V ≥ 3.0.

The regime confirms: no idio signal, no δ, no edge. The -31.7% decline at RSI 23 with 14% short interest LOOKS like a BILL-specific opportunity. It isn't. It's fintech beta in a fintech drawdown. Fintech recovery exposure is available via IPAY at a fraction of BILL's 75% idio vol.

BILL sits in the dead zone: too much infrastructure to die, too much exposed software to thrive. κ = 0. Move on.

Evidence Table