CIBIL Rank and CMR for MSMEs: What the 1 to 10 Scale Really Signals
CIBIL Rank — the CIBIL MSME Rank, or CMR — is a rank, not a score. TransUnion CIBIL grades an MSME on "a scale of 1 to 10, CMR-1 being the least risky MSME and CMR-10 being the most risky", from its credit profile, credit behaviour and firmographics, and it predicts probability of default over a one-year horizon. It is ordinal, bureau-only, and it tells you where a borrower sits — not how far apart two borrowers are.
Key facts
- The scale is inverted relative to a score, and steeply non-linear. One-year default probability runs 1.6% at CMR-1 against 71.3% at CMR-10 (TransUnion CIBIL, CMR asset sheet). Most of the model's discriminating power sits in the bottom bands.
- YuSight has processed 10 Mn credit journeys, and the Bureau Analyzer reads the tradelines that produced the rank — not just the rank field — so a CMR movement can be decomposed instead of quoted.
- The exposure ceiling is ₹50 crore, and the old ₹10 lakh floor is gone. CIBIL's model covers businesses "with aggregate commercial credit exposure lesser than ₹10 Lakhs to the Medium segment having aggregate commercial credit exposure of ₹50 Crores" (CMR asset sheet). The extension below ₹10 lakh was announced on 22 April 2020 and brought roughly 55 lakh very small businesses into scope (TransUnion CIBIL newsroom). Pages still quoting "₹10 lakh to ₹50 crore" are six years stale.
- CMR discriminated on restructuring. In the TransUnion CIBIL–SIDBI MSME Pulse of August 2022, restructured accounts were 57% in CMR-7 to CMR-10, 28% in CMR-4 to CMR-6 and 16% in CMR-1 to CMR-3 (TransUnion CIBIL newsroom, 8 August 2022).
- A newer rank covers more borrowers. CreditVision CIBIL Commercial Rank (CV CMR) uses a 36-month behavioural window and "expands borrower base by ranking 10-12% more commercial borrowers, which were erstwhile unranked" (TransUnion CIBIL).
What drives a CIBIL Rank?
CIBIL describes CMR as built on three families of input: credit profile, credit behaviour and firmographics. In practice, and reading the commercial CreditVision characteristics CIBIL publishes for its commercial report — "Credit Activity, Payment Behaviour, Leverage, Liquidity Profile, Credit Quality" over a 36-month window (Commercial CIR asset sheet) — the drivers a credit analyst can actually see on the report are:
Driver family | What the bureau sees | What moves it |
|---|---|---|
Payment behaviour | DPD grid across every tradeline, asset classification | Any slip past 30 days; SMA tagging; write-offs and settlements |
Credit activity | Facility open dates, enquiry volume and recency | New facilities, a cluster of enquiries, new lender types |
Leverage | Total sanctioned and outstanding, funded and non-funded | Fresh sanctions; guarantees given |
Liquidity profile | Revolving limit utilisation, month-to-month behaviour | A cash credit line moving from 70% to 98% utilisation |
Credit quality | Mix of lender types, facility types, vintage | Bank limits static while NBFC and fintech limits grow |
Firmographics | Constitution, vintage, industry, location | Rarely; these are slow-moving |
What the model does not see is at least as important. CMR is computed from bureau-reported data. It does not read the borrower's audited financials, GST returns, bank statements, order book, promoter net worth or the group's consolidated position. A borrower can be CMR-2 with an interest coverage ratio below 1.0, and nothing in the rank will tell you.
How does CMR map to risk?
CIBIL publishes a one-year default probability against each rank band. The two endpoints are unambiguous:
Band | One-year default probability |
|---|---|
CMR-1 | 1.6% |
CMR-2 to CMR-9 | Published band-wise in CIBIL's asset sheet — pull the current values directly before quoting them in a policy document |
CMR-10 | 71.3% |
Source: TransUnion CIBIL, CMR asset sheet.
Three consequences for how you use the number:
1. The gaps between bands are not equal. The distance from CMR-1 to CMR-2 is nothing like the distance from CMR-9 to CMR-10. Any policy that treats "one band better" as a constant improvement — a uniform pricing step per band, for instance — is pricing a curve as though it were a line.
2. Most lenders' cut-offs cluster in the middle. Common practice is to treat CMR-1 to CMR-3 as acceptable, CMR-4 to CMR-6 as acceptable with conditions, and CMR-7 to CMR-10 as declined or referred. That is a convention, not a CIBIL definition, and each lender's board-approved policy sets its own bands.
3. A rank is a position in a population. Because CMR ranks borrowers relative to a modelled population, a borrower's band can move when the population moves. In a benign year — commercial delinquencies hit a five-year low of 1.87% as at December 2025 (MSME Pulse, April 2026) — holding your rank is a weaker signal than it looks.
When is a CMR not available?
More often than analysts expect. A file arrives with the rank field blank or marked not available for four reasons:
- Aggregate exposure above ₹50 crore. The MSME model's ceiling. Larger borrowers are handled through CIBIL's other commercial ranking products, including CV CMR, rather than CMR.
- No commercial credit history. A new-to-credit entity has nothing to rank. New-to-credit borrowers were 47% of total new MSME loan originations as at March 2025 (TransUnion CIBIL, 22 May 2025), so this is a large and growing population.
- Insufficient tradeline history or vintage. Where reported facilities are too new or too thin to support the model.
- Entity-matching failure. The most under-diagnosed cause. If the enquiry did not match to the entity's bureau record — because of a name variant, a mistyped PAN or a post-conversion CIN — the report comes back thin and unranked. Reconcile CIN, PAN and date of incorporation before you conclude the borrower has no history, as set out in the section-by-section guide to the commercial CIR.
Note what changed: the ₹10 lakh floor no longer applies. Since April 2020 the model covers entities with aggregate exposure below ₹10 lakh. A blank rank on a very small borrower is not automatically an eligibility problem.
Worked example: what a CMR movement actually signals
Illustrative quarterly record for Vaayu Auto Components Private Limited, drawn from four consecutive commercial CIR pulls. Constructed example, not a real borrower.
Pull date | CMR | Total exposure (₹) | CC utilisation | Facilities added in quarter | Worst DPD | Commercial enquiries, trailing 6m |
|---|---|---|---|---|---|---|
30 Sep 2025 | 3 | 4,10,00,000 | 71% | — | 000 | 2 |
31 Dec 2025 | 3 | 4,35,00,000 | 78% | — | 000 | 3 |
31 Mar 2026 | 4 | 5,60,00,000 | 89% | NBFC WCDL ₹75,00,000 | 000 | 6 |
30 Jun 2026 | 5 | 6,10,00,000 | 98.9% | NBFC unsecured ₹50,00,000 | 000 | 9 |
Step 1 — size the exposure movement. 6,10,00,000 − 4,10,00,000 = ₹2,00,00,000 added in nine months 2,00,00,000 ÷ 4,10,00,000 = 48.8% growth in total exposure
Step 2 — size the liquidity movement. Cash credit utilisation: 71% → 98.9% = 27.9 percentage points in nine months
Step 3 — size the demand movement. Trailing six-month enquiries: 2 → 9 = 4.5x, against 2 conversions — 7 enquiries that did not become facilities
Step 4 — read what the DPD column says. 000 in every quarter. Zero days past due throughout.
Step 5 — form the judgement.
The rank moved two bands, from CMR-3 to CMR-5, without a single missed payment. That is the whole lesson. Everything that moved the rank moved before delinquency: exposure up 48.8%, the working capital line filling to 98.9%, new money coming from NBFCs rather than the incumbent banks, and seven lenders looking and not lending.
Three readings a credit analyst should take from this:
- A two-band CMR fall with clean DPD is a leading indicator, not a lagging one. It is the model reacting to activity, leverage and liquidity — which is exactly what you want it to do, and exactly the movement that gets waved through when the covering note says "no overdues, no defaults".
- The absolute band is less informative than the path. CMR-5 as a first observation is unremarkable. CMR-5 arrived at from CMR-3 in three quarters is a file that needs a conversation.
- The rank did not tell you any of the five numbers above. They came from reading the tradelines. The rank told you that something had changed; the report told you what.
What this file justifies is not a decline on rank. It is: get the ₹2 crore of incremental borrowing explained against a use-of-funds statement, test drawing power against the latest stock and book-debt statement, compare the last six months of bank statements against the reported tradelines to confirm nothing is being serviced that the bureau has not seen, and re-pull the CIR within seven days of the credit committee date. That reasoning then belongs in the risk-factors section of the credit assessment memo, with the four quarterly ranks and the utilisation series shown, not summarised.
What CMR cannot tell you
Be blunt about the limits in the memo. A rank quoted without them is a rank the credit committee will over-trust.
- It is ordinal. CMR-4 is not "twice as risky" as CMR-2. Convert to the published default probability before doing any arithmetic with it.
- It is bureau-only. No financials, no GST, no bank statements, no order book, no promoter net worth, no collateral.
- It is entity-level. A group's aggregate stress does not enter the borrower's own rank. You have to pull the group entities' reports yourself.
- It inherits every reporting lag. Lenders report fortnightly, submit within seven calendar days and CIBIL ingests within five (RBI, Credit Information Reporting Directions, 2025). The rank is computed on data that can already be weeks old.
- It inherits every reporting gap. Facilities from lenders outside the reporting perimeter, trade credit and promoter loans never reach the model.
- It is a prediction, not a diagnosis. It gives a probability. It does not tell you what is wrong, and it will not survive a credit committee question about why.
FAQ
What is CIBIL Rank?
CIBIL Rank, or CIBIL MSME Rank (CMR), is a 1 to 10 grade TransUnion CIBIL assigns to an MSME from its bureau credit profile, credit behaviour and firmographics. CMR-1 is the least risky and CMR-10 the most risky, and the rank predicts default probability over a one-year horizon.
What is a good CMR for an MSME loan?
Most lenders treat CMR-1 to CMR-3 as acceptable and CMR-7 to CMR-10 as a decline or a referral, with the middle bands going through with conditions. That split is market convention rather than a CIBIL rule — your own board-approved credit policy sets the bands that bind you.
How is CMR different from a CIBIL score?
A CIBIL Score is a 300 to 900 score on an individual, where higher is better. CMR is a 1 to 10 rank on a business entity, where lower is better. The two run in opposite directions, and mixing them up in a rating grid is a silent error.
What exposure range does CMR cover?
Businesses with aggregate commercial credit exposure from below ₹10 lakh up to ₹50 crore. The often-quoted ₹10 lakh floor was removed in April 2020, when CIBIL extended the model to smaller entities.
Why does my borrower have no CIBIL Rank?
Usually one of four reasons: aggregate exposure above ₹50 crore, no commercial credit history at all, tradelines too thin or too new to model, or the enquiry failed to match the entity's bureau record. Check the entity identifiers before assuming the borrower is new to credit.
Can a CMR fall without any missed payment?
Yes, and it frequently does. Rising exposure, a working capital line filling up, a burst of enquiries and a shift toward more expensive lenders all move the rank while the DPD grid stays clean. That is the most useful signal the rank gives you.
Does a CMR movement mean the borrower's behaviour changed?
Not necessarily. CMR is a rank within a modelled population, so a band can shift when the population shifts. Decompose the movement against the borrower's own tradelines before writing it up as a behaviour change.
Is CMR the same as CreditVision CIBIL Commercial Rank?
No. CV CMR is a separate product built on a 36-month behavioural window that CIBIL says ranks 10 to 12% more commercial borrowers than were previously rankable. Both use a 1 to 10 band structure, so check which one your report is printing.
Should a CAM quote the CMR or the default probability?
Both. The rank alone is an ordinal position that invites over-reading; pairing it with the published one-year default probability turns it into something the credit committee can price against.
How often should I re-pull the CMR?
At application, again within seven days of the credit committee date, and monthly for anything on the watchlist. For an existing borrower, a quarterly series of ranks is far more informative than any single pull.
Key takeaways
- CMR is a rank, not a score: ordinal, inverted relative to a CIBIL Score, and steeply non-linear from 1.6% at CMR-1 to 71.3% at CMR-10.
- The exposure range is below ₹10 lakh up to ₹50 crore. The ₹10 lakh floor was removed in April 2020.
- A blank rank most often means no history, exposure above the ceiling, or an entity-matching failure — check the identifiers before concluding anything.
- A CMR movement with a clean DPD grid is a leading indicator. Decompose it against exposure, utilisation and enquiries.
- The rank is bureau-only and entity-level. It knows nothing about financials, group stress or unreported borrowing.
Run one borrower through the analyzer — bring a commercial CIR and see the rank sitting alongside the tradelines, utilisation series and enquiry history that produced it.
Related: how to read any commercial bureau report end to end, every section of a CIBIL Commercial Report, and where the bureau reading sits in a credit appraisal memorandum.