A guide for lenders, fintech platforms and B2B risk teams — including how the scoring models actually differ, and what to do when two providers disagree about the same company.
The problem with a business credit report is that it looks like a fact
It isn't. A business credit report is an assembly of three very different things, and most buyers evaluate providers without separating them.
The first layer is identity: is this legal entity real, active, and the one you think it is. That comes from a government registry, and it is verifiable. The second is history: filed accounts, charges, mortgages, legal filings, payment experiences contributed by suppliers. Partly verifiable, partly voluntary. The third is a model output: a score between 1 and 100, or 101 and 992, or a letter-number composite. That is a vendor's opinion, calculated on the first two layers, weighted by a method they will not fully disclose.
When two providers return different scores for the same company — and they routinely do — it is almost never because one has "wrong data". It is because they are scoring different things, over different windows, with different inputs. A lender who does not know which layer is driving a decline is not managing risk; they are outsourcing it.
The three layers inside any business credit report
Only the bottom layer is independently verifiable. The top layer is the one most buyers shop on.
Verifiability is marked with a symbol and a word, not colour alone.
What is actually inside a business credit report
| Layer | Typical contents | Where it comes from | How verifiable |
|---|---|---|---|
| Entity identity | Legal name, registration number, status, incorporation date, registered address, activity codes | Government registry | Fully — cite the filing |
| Ownership & control | Directors, officers, shareholders, ultimate beneficial owners, group structure | Registry, beneficial ownership registers | Mostly, varies by jurisdiction |
| Financials | Filed accounts, turnover, net assets, liabilities, ratios, filing history | Registry deposits; modelled where not filed | Filed figures yes, modelled figures no |
| Payment behaviour | Trade lines, days beyond terms, payment index | Voluntary supplier contributions | Not independently — you trust the contributor pool |
| Public detriment | Judgments, liens, charges, insolvency events, strike-off notices | Courts, registries, gazettes | Yes |
| Scores & limits | Risk score, failure score, recommended credit limit | Vendor model | No — it is an opinion |
The distinction between filed and modelled figures is where most disputes start. In jurisdictions with full filing requirements — the UK, most of the EU — a private company's accounts are a public document. In the US, private company financials are largely not public, so US bureaus lean heavily on trade payment data and modelled estimates. A "US business credit report" and a "UK company credit report" are structurally different products that share a name.
How the scoring models differ, in numbers
This is the section most guides skip, and it is the one that decides whether your risk policy is coherent. Every score below runs on its own scale, with its own definition of "good".
Five business credit scores, drawn to their own scales
Each track is normalised to its full published range. The filled segment is the band generally considered acceptable; the marker is the published threshold.
| Score | Provider | Range | Good | What it measures |
|---|---|---|---|---|
| PAYDEX | Dun & Bradstreet | 1–100 | 80+ | Payment behaviour only. 80 = pays on terms; above 80 = pays early |
| D&B Rating | Dun & Bradstreet | Composite (5A1) | Higher letter, lower number | Financial strength band plus composite credit appraisal |
| Failure Score | Dun & Bradstreet | Percentile | Higher | Likelihood of ceasing operations with loss to creditors |
| Intelliscore Plus | Experian | 1–100 | 76+ | Blended delinquency risk, benchmarked by industry |
| Financial Stability Risk | Experian | 1–100 / class | Higher | Probability of financial distress over 12 months |
| Business Credit Risk Score | Equifax | 101–992 | 700+ | Likelihood of severe delinquency |
| Business Failure Score | Equifax | 1000–1880 | Higher | Bankruptcy risk |
| Payment Index | Equifax | 0–100 | Higher | Rolling 12-month payment trend |
| International Score | Creditsafe | 1–100 | Country-mapped | Blended risk, harmonised across countries |
Three consequences follow directly. A PAYDEX of 80 and an Intelliscore of 80 mean completely different things — the first says "pays on terms" and nothing about solvency. Equifax's 101–992 range is not comparable to anything else, so a policy written as "reject below X" without naming the model is ambiguous by construction. And payment-behaviour scores need a contributor pool to exist: if nobody reports trade lines for a company, PAYDEX does not exist for it.
The thin-file problem, and why it matters more than score accuracy
Ask any provider what percentage of the companies you will query return a full score. The answer is rarely volunteered, and it is the number that determines whether the integration actually works.
Trade-line-based scoring is dense where the contributor network is dense — large US suppliers, UK trade creditors — and thin everywhere else. A German Mittelstand supplier, a Cypriot holding company, a newly incorporated Dutch subsidiary: real entities, real risk, no score.
What happens when the score does not exist
Three routes out of a no-score response. Only one of them adds information.
How stale is the report you are reading?
"Real-time" describes a provider's retrieval frequency, not the rate at which the underlying facts change. Each layer moves at a different speed, and a report generated today can rest on a filing from nineteen months ago.
How often each layer actually changes
The report's generation date tells you nothing about the age of what is inside it.
The six providers
Global Database — registry-first, best for cross-border and thin-file coverage
Global Database builds from the source layer rather than the score layer: 600M+ company profiles assembled from 400+ government registries across 200+ countries, with up to 20 years of filing history where the jurisdiction publishes it.
Every field carries provenance. Registry name, a link to the underlying filing, and the retrieval timestamp. When a credit committee or an auditor asks why a limit was set, the answer is a document, not a number.
Coverage does not collapse at the border. The same schema returns a Dutch KvK record, a German Handelsregister filing and a Cypriot entity — removing the vendor-per-country problem that quietly consumes procurement budgets.
Credit sits next to KYB, UBO and sanctions in one call. Most teams are not running a credit check in isolation; they are onboarding a counterparty. Verifying the entity, resolving beneficial owners, screening them against OFAC, EU, UN and UK HMT lists and pulling filed accounts in a single request removes three integrations and three reconciliation problems.
Monitoring is registry-level, not score-level. You are alerted when the filing changes, rather than when a vendor's model happens to recalculate.
Delivery is unusually flexible. Web platform with bulk CSV, REST API, scheduled bulk feed, native CRM integrations, a browser extension, and an MCP server so AI agents can run checks directly. Pricing is per verification with volume tiers, not a seat licence.
Dun & Bradstreet — the trade-line standard, and the identifier everyone else references
The DUNS number remains the de facto entity identifier across procurement, government contracting and much of the credit ecosystem. PAYDEX is the most widely recognised payment-behaviour score in the US, and D&B's corporate linkage is genuinely deep. Strongest for: US trade credit decisions, supplier onboarding where DUNS is contractually required, large corporate hierarchies. Limitations: pricing scales aggressively with volume; coverage quality outside core markets is uneven; scores update on D&B's cadence, not the registry's; and self-reporting through the company's own portal introduces data your credit team did not source.
Experian Business — deepest US small-business file
Intelliscore Plus is industry-benchmarked, which makes it more useful than a raw score when comparing a construction subcontractor against a software reseller. Experian's blending of principal credit data for thin-file businesses is a real advantage on young US companies. Strongest for: US SMB lending, merchant onboarding, US-concentrated portfolios. Limitations: international coverage is a different product with a different shape; principal matching carries consumer-data compliance obligations that vary by jurisdiction.
Equifax Business — distinct scoring models, useful as a second opinion
Equifax separates delinquency risk (101–992) from failure risk (1,000–1,880), which is analytically cleaner than a single blended score, and its Payment Index tracks a rolling twelve-month trend rather than a point estimate. Strongest for: US small business, portfolios wanting an independent second model. Limitations: the non-standard ranges complicate policy writing; outside North America, coverage thins quickly.
Creditsafe — European payment data and fast onboarding
Creditsafe's strength is the breadth of its European payment-data network combined with a harmonised international score that maps to local scores. Reporting is fast and the self-serve experience is good, which is why it wins a lot of mid-market shortlists. Strongest for: European trade credit, mid-market teams wanting one subscription across several European countries. Limitations: depth varies significantly by country; the harmonised score simplifies genuine differences between jurisdictions; API maturity is adequate rather than developer-first.
Coface — risk intelligence attached to insurance
Coface is an insurer first. Its sector risk assessments and country risk ratings are excellent, and if you are buying trade credit insurance the underwriting view comes with the policy. Strongest for: exporters, sector-level exposure analysis, insured trade credit. Limitations: it is not a self-serve data product; the value is tied to the insurance relationship.
Where each provider sits
Two axes that actually separate these products: how far the coverage travels, and whether you are buying evidence or a judgment. Positions are qualitative.
Head-to-head: the comparisons buyers actually run
Creditsafe vs Dun & Bradstreet
The real distinction is network versus geography. D&B's trade-line network is deepest in the US and its identifier is embedded in procurement systems worldwide. Creditsafe's payment network is deepest in Europe, and its pricing and onboarding are friendlier to mid-market teams.
Choose Dun & Bradstreet when
- DUNS is a contractual or procurement requirement
- Your exposure is US trade credit
- You need hierarchies on large multinationals
- PAYDEX is already written into your credit policy
Choose Creditsafe when
- Your book is predominantly European
- You want several countries on one subscription
- Self-serve speed matters more than analytical depth
- Budget is mid-market rather than enterprise
Where both leave a gap: registry-level evidence for jurisdictions outside their core networks, and a fallback for companies with no trade lines at all. That is the slot a registry-first source fills — which is why plenty of teams run Creditsafe or D&B and a registry source, rather than debating which single vendor wins.
Dun & Bradstreet vs Experian
Both are strong in the US, and they measure different risks. PAYDEX is behavioural — how this company pays, right now. Intelliscore is predictive and industry-benchmarked — how likely serious delinquency is, relative to peers. For trade credit terms, PAYDEX maps more directly to the decision. For lending and limit-setting, Intelliscore's benchmarking is more informative. Portfolios using both usually do so deliberately: one for terms, one for approval.
Experian vs Equifax for business credit
Closer competitors than either is to D&B. Experian's advantage is file depth on US small business and the principal matching that rescues thin files. Equifax's advantage is model separation — delinquency and failure scored independently, plus a payment trend rather than a snapshot. If you can hold only one, Experian generally wins on coverage. If you can hold two, Equifax is the better second opinion precisely because its models are constructed differently: two correlated scores tell you less than two independent ones.
Global Database vs the bureaus
Not the same product, and the comparison should be framed honestly. The bureaus sell a judgment: a score, a limit, a recommendation, built on proprietary contributor networks. Global Database sells the evidence: the registry record, the filed accounts, the ownership chain, each with its source attached, across 200+ countries.
A credit team with only evidence has to build its own policy. A credit team with only judgment cannot explain a decision when the score is wrong or absent. Most mature teams end up with both — a bureau where the contributor network is dense, and a registry source everywhere else, for the thin files, and for the audit trail.
Comparison table
| Provider | Core strength | Geography | Primary score | Delivery | Best for |
|---|---|---|---|---|---|
| Global Database | Registry-sourced evidence with provenance | 200+ countries, 400+ registries | Registry risk signals, filed financials | Platform, REST API, bulk feed, CRM, MCP | Cross-border risk, thin files, auditable decisions |
| Dun & Bradstreet | Trade-line network, DUNS, linkage | Global, deepest US | PAYDEX 1–100, D&B Rating | Platform, API | US trade credit, procurement mandates |
| Experian Business | US SMB file depth, benchmarking | Deepest US | Intelliscore Plus 1–100 | Platform, API | US small business lending |
| Equifax Business | Independent delinquency and failure models | Deepest US | Risk Score 101–992 | Platform, API | Second opinion, failure risk |
| Creditsafe | European payment data, self-serve | Strong Europe | International score 1–100 | Platform, API | European trade credit, mid-market |
| Coface | Sector and country risk, underwriting | Global, insurance-led | Internal assessments | Advisory, policy-linked | Insured trade credit, exporters |
How to choose, by decision
Which source for which decision
Start from the decision you are making, not from the vendor shortlist.
Seven checks before you trust a report
- Read the date, not the score. Ask when the underlying filing was retrieved, not when the report was generated. A report generated today on accounts filed nineteen months ago is a 2024 opinion with a 2026 timestamp.
- Check whether the score exists at all. A returned "no score" is information — it usually means no contributor data, not low risk.
- Separate filed from modelled. Ask the provider to mark which financial figures are deposited and which are estimated, then set a policy that treats them differently.
- Confirm the entity, not the name. Lookalike names across jurisdictions are the commonest matching error. Match on registration number.
- Look up the group, not just the entity. A solvent subsidiary of a failing parent is a different risk from a standalone solvent company.
- Treat filing punctuality as a signal. Overdue accounts correlate with distress more reliably than many proprietary scores, and they are free to observe.
- Keep the evidence. If the decision is ever challenged, a link to the original filing settles it; a score does not.
Check a company against the registry, not against a model
Entity status, filed accounts, beneficial owners and sanctions screening in one API call, across 200+ countries — every field with the registry name, the filing link and the retrieval timestamp attached.
See business credit risk data →Frequently asked questions
What is a business credit report?
A compiled record of a company's identity, financial filings, public detriment and payment behaviour, usually ending in one or more risk scores. It differs from a consumer credit report in three ways: it is generally not protected by consumer credit legislation, it can be purchased by anyone with a commercial interest, and its underlying data is contributed voluntarily rather than mandated. In jurisdictions with public filing requirements, much of the report is a summary of documents anyone can retrieve from the registry.
What is the difference between a business credit report and a business credit score?
The report is the evidence; the score is a vendor's interpretation of it. The same company can hold a PAYDEX of 78, an Intelliscore of 84 and an Equifax Business Credit Risk Score of 690 simultaneously, because each model weighs different inputs over different windows. Credit policies should reference a named model and threshold, never "a score above X".
How do I get a business credit report on another company?
Buy a single report from a bureau, subscribe for ongoing access, or query an API if you need volume. For cross-border checks, confirm before you buy that the provider covers the jurisdiction at the depth you need — many advertise 200+ countries but return only identity data outside their core markets. If the company is in a jurisdiction with public filing, you can also retrieve the accounts directly from the registry, which is slower but definitive.
How do I check my own business credit report?
Request it from each bureau that holds a file on you — they are separate companies with separate data, so one clean report does not mean the others are clean. Check that the entity details match your registry record exactly, that trade lines belong to you, and that no closed accounts are still reporting. If you are a supplier being checked by customers, it is worth knowing which bureau they use before you try to fix anything.
Can I get a free business credit report?
Some bureaus offer limited free views of your own company's file, typically summary scores without the underlying trade lines. Full reports on third parties are paid. In the UK, the EU and other public-filing jurisdictions the underlying accounts are free at the registry — what you pay a bureau for is aggregation, scoring and speed, not exclusive access to the facts.
How often are business credit reports updated?
It varies by layer, and that distinction matters. Registry data can be synced daily. Filed accounts change once a year, whenever the company deposits them. Trade payment lines update as contributors submit, typically monthly. Scores recalculate on the vendor's own cycle. A provider claiming "real-time" is describing its retrieval frequency, not the rate at which the underlying facts change.
What is a PAYDEX score and what is a good one?
Dun & Bradstreet's payment-behaviour score, on a 1–100 scale, calculated purely from reported payment experiences weighted by dollar amount. 80 means the company pays on terms; above 80 means it pays early; below 80 means increasing days beyond terms. It says nothing about solvency, profitability or ownership — a company can pay on terms right up until it fails.
What is a D&B Rating?
A composite of two parts: a financial strength indicator derived from net worth or tangible assets, and a composite credit appraisal from 1 to 4. A rating like 5A1 means high financial strength with the best credit appraisal. It is a useful summary for large companies with filed net worth, and often unavailable or estimated for small ones.
What is a good business credit score?
It depends entirely on the model. On the 1–100 scales, 80 or above for PAYDEX and 76 or above for Intelliscore Plus are the usual benchmarks. On Equifax's Business Credit Risk Score, above 700 on a 101–992 range. Any policy that does not name the model alongside the threshold is ambiguous, and ambiguity in a credit policy eventually becomes an inconsistent decision you cannot defend.
How do I run an international business credit check?
Either assemble country-by-country coverage from local bureaus, or use a provider with direct registry connections across the jurisdictions you need. The first gives depth in each market and costs a vendor relationship per country; the second gives one schema, one contract and one integration, with registry-level evidence everywhere. Most teams start with the first and migrate to the second once they pass roughly five countries.
Can one platform cover both domestic and international credit checks?
Yes, but examine what "cover" means in each market. Ask for a sample report from the least important country on your list — coverage claims are always true of the best market and frequently thin in the worst. Depth, not breadth, is the number that predicts whether the integration works.
Why do two providers give the same company different scores?
Because they hold different inputs. One may have trade lines the other lacks; one may have retrieved the latest filing and the other not; the models weigh payment behaviour, financial strength and public detriment differently. Disagreement is normal and informative — a company scored well by a payment model and poorly by a failure model is usually a company paying on time out of a shrinking balance sheet.
What is a thin file, and what should I do about it?
A company with too little contributed data to score — common for young companies, small companies, and companies outside a bureau's core network. Options are to decline, to accept with a conservative limit, or to fall back on registry evidence: incorporation date, filing history, filed net assets, director history and group structure. The third is the only one that adds information rather than removing revenue.
How do I dispute an error on a business credit report?
Contact the bureau that publishes the report and supply documentary evidence — a registry filing carries more weight than a letter. Each bureau handles disputes separately, so an error corrected at one may persist at another. Because business credit is not covered by the same statutory dispute timelines as consumer credit in most jurisdictions, resolution speed varies, and following up in writing matters.
How is business credit reported, and how does a company build it?
Suppliers and lenders contribute trade experiences voluntarily to one or more bureaus. A company builds a file by trading on credit terms with suppliers who report, paying on or before terms, and keeping its registry filings current and punctual. There is no way to force a supplier to report, which is why two companies with identical behaviour can have very different file depth.
Which business credit cards report to Dun & Bradstreet?
Card issuers differ, and the policy changes — some report business card activity to business bureaus only, some to personal credit as well, some to both. Check the issuer's current terms rather than a list, and verify by pulling your own file a few months after opening the account. For a risk team the relevant point is the reverse: the absence of card trade lines in a file tells you little about a counterparty.
Does taking a business loan affect the owner's personal credit?
It depends on whether the facility was personally guaranteed and on the lender's reporting policy. Many small business lenders report to consumer bureaus when a personal guarantee exists. For a credit team assessing a small company this is why principal data can be informative on thin files — and why using it carries consumer-data compliance obligations that differ by jurisdiction.
What is trade credit, and how does it differ from a loan?
Trade credit is the deferral a supplier grants a customer — goods now, payment in 30, 60 or 90 days. It is the largest source of short-term financing for most businesses and it is extended without a formal credit agreement, which is precisely why payment-behaviour scores exist. A loan is priced for risk up front; trade credit is usually priced at zero and monitored afterwards, if at all.
What is business credit monitoring, and what should it alert on?
Continuous observation of a counterparty after onboarding. Score-level monitoring tells you when a vendor recalculated. Registry-level monitoring tells you when the fact changed — a director resigned, a charge was registered, accounts went overdue, the status moved to strike-off. The second is earlier, and earlier is the entire point.
Can I automate business credit checks in my own systems?
Yes, via API. The questions to ask are the ones that determine whether it works at scale: does the response include reason codes you can act on programmatically, what is the latency, what happens on a no-match, is there a sandbox, and is bulk re-checking of an existing portfolio priced differently from one-off lookups. An API that returns a score with no reason codes cannot support an automated decision you will later have to explain.