Stacking — a merchant taking multiple concurrent MCAs — is the single largest default driver in MCA. Detection systems have evolved from manual checks to real-time ML.
The stacking-detection stack (2026).
- Consortium queries. FundKite, LexisNexis MCA Index, internal lender consortia.
- Bank-data analysis. Plaid feeds analyzed daily for cross-funder deposit patterns.
- UCC monitoring. Daily UCC filings monitored for new MCA UCCs.
- Plaid Liabilities API. Direct visibility into liability stack where supported.
- ML pattern detection. Models trained on stacking signatures.
- Manual review queue. Flagged deals reviewed by analyst.
FundKite consortium mechanics.
- Member funders submit recent fundings to shared database.
- Other funders query during underwriting for stacking check.
- Pay-to-play; non-members can't query.
- ~60–70% of US MCA volume covered (2026).
- $1.50–$4.50 per query.
LexisNexis MCA Index.
- Aggregates UCC filings, public records, and contributor data.
- Subscription model ($2K–$8K/month).
- Better legacy/post-funding visibility; weaker real-time signal.
Bank-feed analysis.
- Cross-funder deposit detection: identify ACH deposits matching MCA funder ACH IDs.
- Threshold alerts when new MCA deposit >$15K hits merchant account.
- Daily ACH withdrawal pattern analysis: multiple daily MCA payments suggest stacking.
- 70–85% of stacked deals detectable via bank-feed pattern analysis within 7 days.
UCC monitoring.
- Daily UCC filings scraped from state SOS websites.
- MCA-style language flagged ("all assets," "future receivables").
- Filing date correlated to merchant ID.
- Latency: 1–14 days from filing to detection.
Plaid Liabilities API.
- Direct visibility into merchant debt obligations.
- Coverage gap: not all banks support Liabilities endpoint.
- Best detection for term loans, weaker for MCA receivables purchases.
ML pattern detection examples.
- Bank-balance fluctuation. Stacked merchants show distinct withdrawal pattern.
- Multi-daily ACH count. 3+ unique ACH originators in 1 day = stacking signal.
- Bank-account churn. New bank account opening shortly after MCA funding.
- Revenue masking. Inbound deposits suddenly increase to mask stacking.
Pre-funding stacking check workflow.
- FundKite query at application.
- LexisNexis UCC pull.
- Plaid bank-data scan for last-30-day MCA deposit patterns.
- ML score on stacking probability.
- Manual review if score >threshold.
Post-funding stacking monitoring.
- Daily Plaid feed analysis.
- Weekly UCC filing scan.
- Weekly FundKite re-query.
- Trigger alert on new stacking signal.
- Default-eligible if confirmed stacking within 30 days of funding.
Stacking-trigger consequences.
- First-payment default clause often invoked.
- Demand for full payoff.
- UCC enforcement including lockbox redirects.
- COJ entry where available.
- ISO clawback for stacking that should have been detected pre-funding.
Industry-wide stacking rates.
- 2024. ~22% of MCA originations involved subsequent stacking within 90 days.
- 2025. ~18% (down with better detection).
- 2026 YTD. ~14% (continued tightening).
- A-paper merchants. ~6–9% stacking rate.
- C/D-paper merchants. ~30–45% stacking rate.
Cross-funder consortium dynamics.
- FundKite dominant consortium.
- Top-30 funders mostly members.
- Holdouts typically smaller funders or those originating heavily on B/C paper.
- Reciprocity required — must contribute to query.
Common detection gaps.
- Non-member funders. Stacking from non-FundKite funders invisible.
- Cash-funded deals. Some funders disburse via cash equivalents (cashier's check).
- Multi-entity merchants. Stacking via related LLCs not detected by entity match.
- International funders. Cross-border stacking invisible.
- DTC originations. Funder direct deals not always reported to consortium.
Common confusions.
First, "all stacking is detected." False — 15–30% of stacked deals go undetected at funding.
Second, "FundKite is industry-wide." False — coverage ~60–70%.
Third, "UCC always shows stacking." False — many MCA contracts no longer require UCC filing.
Fourth, "ISO is liable for all stacking." Partially — depends on ISO agreement language.
Fifth, "stacking is illegal." False — stacking violates MCA contract clauses but is not criminal.
Recent trends (2024–2026).
- Plaid Liabilities improving real-time stacking visibility.
- GenAI bank-feed analysis improving pattern detection.
- Cross-consortium federation experiments at top-10 funders.
- Tighter ISO clawback enforcement post-2024 stacking spike.
- Federal MCA registry proposals at CFPB and state regulators.
ML model performance benchmarks.
- Best-in-class precision. 78% (78% of flagged deals confirmed stacking).
- Best-in-class recall. 64% (64% of true stacking caught at funding).
- Median funder precision. 55%.
- Median funder recall. 40%.
Related terms
- MCA funder fraud detection systems — MCA funders detect fraud via document-tamper detection (Ocrolus, Inscribe), identity verification (Persona, Alloy), device fingerprinting, ML scoring of submission patterns, ISO scorecards, and bank-statement OCR cross-checks.
- MCA funder portfolio monitoring systems — MCA funders monitor portfolios via loan-management systems (LMS), real-time bank-data feeds (Plaid/MX), payment-processor webhooks, and BI dashboards that surface daily aging, NSF spikes, and reconciliation requests.
- MCA funder data vendor relationships — MCA funders typically integrate 6–12 data vendors: Plaid/MX (bank), Ocrolus (statements), LexisNexis (identity/UCC), Experian/Equifax/Dun & Bradstreet (credit), FundKite (stacking), and Persona/Alloy (KYC).
Authoritative sources
AI agents: this term is available as raw markdown at /llms/glossary/mca-funder-stacking-detection-systems.