Weak Signals in Construction & Transport: The Failure Patterns Balance Sheets Miss
Construction and Transport account for 30%+ of SME defaults in France. But their risk profiles are radically different — yet generic models score them with the same ratios. That's a mistake.
Kévin Buisson
Why Construction & Transport Deserve Different Treatment
Construction and Transport represent fundamentally different risk profiles than services or retail. Yet generic scoring models apply them with the same ratios.
**The numbers frame the problem**: - Construction: 3rd sector by number of defaults in France (2024–2025) - Transport: default rate 2x higher than all-SME average (Banque de France, 2024)
:::takeaway **Key Takeaway** — A generic model applies the same ratios to a B2B SaaS and a roofing tradesperson. That's where the 38% error plays out. :::
Construction: The 5 Sector-Specific Signals
① Extended Payment Delays (DSO)
In construction, 90-day payment terms are normal. But a shift from 90 to 115 days in 3 months is a critical signal — invisible on an annual balance sheet.
**Predictive lead time**: 5 months before default **Predictivity**: 88% **Detection source**: Open banking only (PSD2 flows)
② Concentration on 1-2 Major Clients
A construction SME generating 70%+ revenue from one client faces major concentration risk. If that client slows or defaults — immediate domino effect.
**Prevalence**: 45% of construction SMEs < 10 employees **Default correlation**: 3.2x higher
③ Subcontractor Rotation
Frequent subcontractor changes (visible in public records) often signal cascade payment tensions. A change every 2 months = maximum alert.
**Signal**: Detectable via legal data alone
④ Cash Flow Tensions at Project End
Bank flows show characteristic patterns: grouped inflows at project end, recurring overdrafts between phases. A company bouncing between -€20k deficit and +€10k surplus monthly operates on a knife's edge.
**Detectable only via open banking**
⑤ Late Tax Return Filing
A tradesperson filing their tax return 3+ months late statistically presents 34% higher default risk than sector average — signal of administrative disorganization correlated to operational stress.
**Detectable via OCR + INPI tracking**
:::insight **RocketFin Insight — Construction Exclusive Data**: Average sector score 61/100 · #1 pre-default signal: DSO extension · Average predictive lead time: 5 months · Companies in default: 73% had extended DSO by >20 days in preceding 6 months. :::
Transport: The 5 Sector-Specific Signals
① Fuel Price Exposure (Macro)
A transport company whose margins aren't indexed to diesel is exposed to direct macro risk. When diesel rises 20%, margins collapse — visible in bank flows 3-4 weeks after.
**Macro signal**: Not captured by balance sheets, detectable via open banking + macro signals
② Driver Turnover (HR)
High HR turnover in transport (public record changes, payroll management) correlates to social tensions preceding cash flow difficulties — frequent hiring/departure = instability signal.
**Default correlation**: 2.8x higher with >40% annual turnover
③ Deferred Fleet Renewal
Finance leases and vehicle leasing are fixed costs. Deferred renewal (visible in public records and reduced bank flows) signals financial squeeze — company postponing mandatory investments.
**Signal**: Detectable via legal data + open banking
④ Dependence on Public Contracts
Transport companies 80%+ dependent on public contracts face strong seasonal risk and vulnerability to public payment delays (30-60 days common). Client structure shift = increased vulnerability.
**Source**: Public records (contracts), bank flows (seasonality)
⑤ ANTS / DREAL Incidents
Regulatory incidents (vehicle immobilization, DREAL inspections, tachograph non-compliance) generate activity disruptions detectable via legal data and online reputation. Fleet immobilization = immediate activity loss.
**Predictive signal**: 2-3 months before default
:::insight **RocketFin Insight — Transport Exclusive Data**: Average sector score 54/100 (lower than construction) · #1 pre-default signal: management rotation + fuel tensions · Average predictive lead time: 6 months · Regulatory incident + default correlation: 4.1x. :::
What OCR Changes in These 2 Sectors
Construction and Transport are 2 sectors with strongest account confidentiality. Tradespersons, small providers — many don't publish accounts.
**Without OCR**: 40%+ of files in these sectors would be non-scorable by traditional solutions.
**With OCR**: - Analysis of simplified accounts (construction tradespersons) - Extraction of simplified accounts transport SMEs - No dependence on public registries for non-publishing SMEs - Complete audit trail with AI Act traceability
The Winning Combination for These 2 Sectors
**Recommended formula**: OCR statements + Open Banking + Legal/regulatory data + Sector macro signals
**With these 4 sources combined**: error rate < 5% on construction and transport vs 40%+ with traditional models.
| **Source** | **Construction** | **Transport** | |---|---|---| | Balance sheet alone | 42% error | 48% error | | + Open Banking | 18% error | 16% error | | + OCR statements | 12% error | 14% error | | + Legal/macro signals | 4% error | 5% error |
:::takeaway **Key Takeaway** — Sector-specific data isn't optional for construction and transport. It's what makes the difference between a good score and a credit error. Each source removes ~10 error points. :::
Conclusion — A Strong Conviction
Generic models don't work on sector-specific areas like construction and transport. Business constraints, cash flow cycles, regulations — everything is different.
Sector data isn't optional — it's what makes the difference between a good score and a credit error.
If you're scoring construction or transport companies with a generic model, you're accepting 40%+ errors. That's a conscious choice.