The Committee Never Reads the Document You Submitted
A borrower prepares a project report, submits it, and waits. What happens next is almost never what the borrower imagines.
The document isn't circulated to the committee. It's received by a credit officer, frequently junior, always carrying several files at once.
Their task is to read it, test it, and write a shorter document about it. That shorter document is the appraisal note, and it's what the committee actually reads.
The project report sits in the annexure, available to anyone who wants it, consulted by almost nobody.
This single structural fact governs everything else.
The borrower's document is written to persuade. The appraisal note is written to be defended.
The officer who signs it will be asked, possibly years later, why they recommended a facility that went bad. Their protection is that they identified the risks and stated them.
That incentive runs opposite to the borrower's.
A report that anticipates it, surfacing its own weak points and showing how they were tested, makes the officer's note easier to write. Counter-intuitively, it makes the proposal more likely to pass.
A report that buries them makes the officer look for what's missing. An officer who is looking will find something.
This article sets out how a credit proposal is actually assessed: the order the committee reads in, how the appraisal note differs from the borrower's document, why debt service cover is tested as a minimum across the tenor rather than an average, what the security package and covenant schedule are really being asked, and what changes when several institutions must agree.
The borrower writes to persuade. The credit officer writes to be defended. A document that helps with the second is far more likely to achieve the first.
A Credit Paper Is Read Backwards From Repayment
Borrowers write forward: the promoter, the company, the market, the opportunity, the project, the cost, the funding, the returns.
Credit reads backwards.
The first question in the room isn't what the business does. It's how the money comes back, from where, and what happens if it doesn't.
Everything else is context for that question.
A document sequenced as an investment story asks the committee to hold seven pages of background before reaching the only thing it came for. In practice the reader simply jumps, so the background gets skimmed on the way past rather than read on the way in.
That doesn't make the story sections unnecessary. It means they're read in a different order, for a different purpose than the borrower intended.
The market section isn't read to establish that the market is attractive.
It's read, after the cash flow has been seen, to test whether the revenue line is supported. Specifically, it's read for the sentence saying where the volume assumption came from.
Four pages of market description that never state the basis of the volume assumption have failed at the only job they were given.
So each section should be written to answer the question it will be read for, not the question its title implies.
In the mandates our advisers run, a project report for an appraiser is structured so the repayment case is stated early and each later section visibly supports one element of it.
| Pass | What is read | The question being asked | What fails it | |
|---|---|---|---|---|
| 1 | The ask and the structure | Amount, tenor, moratorium, pricing, security offered, promoter contribution — usually on one page. | Is this proposal even within policy? Does it fit the bank's exposure limits, sector appetite and tenor norms? | A structure outside policy is stopped here, before merit is considered. It does not reach the committee at all. |
| 2 | The repayment case | The cash flow projection, the debt service schedule, and the cover in each year of the tenor. | Where does the money come back from, in which years, and how much room is there in the worst of them? | A cover figure quoted as a single number with no year-by-year schedule behind it. The reader cannot see the minimum. |
| 3 | The assumptions under the cash flow | Volume, price, capacity utilisation, input cost, working capital cycle — and the stated source of each. | Is the revenue line an assertion or a derivation? What is it derived from, and is that source independent of the borrower? | An assumption with no stated source. This is the single most common cause of a first-round query list. |
| 4 | The downside | Sensitivity analysis — what happens to cover if volume, price or cost move against the case. | At what point does this facility stop servicing itself, and how far away is that point from the base case? | Sensitivities that move one variable a token amount and conclude the project remains viable. The reader runs their own instead. |
| 5 | Security and its realisable value | The security package, its basis of valuation, and the cover it provides against the exposure. | If the repayment case fails, what is actually recovered, on what timescale, and at what discount to the stated value? | A going-concern valuation offered as security cover. Security is realised when the going concern has stopped. |
| 6 | The promoter and the track record | Experience, existing exposures, group structure, contribution, and conduct on prior facilities. | Has this borrower done this before, and what happened? What else is the group carrying? | An undisclosed group exposure discovered independently. This changes the character of the file entirely. |
| 7 | The market and the project description | Sector context, competitive position, technology, implementation plan and timeline. | Does this support the assumptions already read, or contradict them? | A market narrative that cannot be reconciled to the volume assumption in the model. The contradiction is the finding. |
A generalised description of common credit practice across commercial lending. Institutions differ, credit policy is not published, and no particular bank's internal process is asserted here. The sequence is offered as the pattern a document should be able to survive, not as a rule any named lender follows.
Two features of that sequence are worth drawing out.
The first is that the earliest pass isn't about merit at all.
A proposal outside the institution's policy on tenor, exposure limit, sector or structure is stopped before anyone assesses whether it's a good project. The borrower frequently never learns this happened. The file simply doesn't progress.
Establishing the structural parameters before the document is written is therefore worth more than any amount of quality in the document itself, and that conversation costs nothing to have early.
The second is that the market section, which borrowers treat as the centrepiece, is read last and read adversarially.
By the time the reader reaches it they already know what the model assumes. They aren't reading to be convinced the sector is growing.
They're reading to see whether the sector description supports the specific number in the specific line of the cash flow. A beautifully written market chapter that never connects to that number reads as evasion rather than context.
Your Document Becomes Someone Else's Summary, and the Summary Is What Decides
The appraisal note is a different genre, written by a different person with a different incentive. Where the project report advances a case, the note assesses one.
Where the report presents assumptions, the note attributes them. "The borrower states." "Management projects." "The promoter has represented."
That attribution marks the boundary between what the institution has satisfied itself of and what it has been told.
The note also has a section the borrower's document usually lacks: a statement of risks and mitigants, written as a list.
This is the officer's protection and the committee's agenda. Whatever appears in that list will be discussed. Whatever doesn't, won't.
A borrower who leaves an obvious risk unaddressed doesn't avoid it. The officer identifies it, states it in their own words, and characterises the mitigant as absent.
A borrower who has addressed it supplies the officer with a mitigant to record, in language the borrower chose. The risk appears in both versions. Only one contains an answer.
The note is short, written under time pressure by someone reading several files, so anything important but hard to extract won't make the transition.
A key contract, an offtake arrangement, a completed approval, a technical certification: described in a paragraph on page sixty rather than stated where the relevant test is answered, it may never reach the committee.
Our advisers place each supporting fact adjacent to the assertion it supports. The test isn't whether the document contains the evidence. It's whether the evidence survives being summarised by someone who has forty minutes.
The note's register is flat and declarative, and treats adjectives as noise. A promotional document has to be rewritten into it, and in rewriting the officer decides what the claims amount to. That's a decision the borrower handed over.
What an appraisal note commonly contains that a borrower's project report commonly does not. Each of these is a place where the borrower's framing is replaced by the officer's, and each can be pre-empted by supplying the material in a form the officer can adopt.
-
Attribution of every material assumption
The note distinguishes what the bank has verified from what the borrower has stated. A figure carried through as "management projects" has been flagged as unverified, whatever its merit. Supplying an independent source converts the attribution, and the attribution is what the committee reads.
-
A risks-and-mitigants list, in the officer's words
Usually the most-read part of the note. Risks the borrower did not address appear with the mitigant column empty. That empty column is the agenda for the meeting, and filling it in the room, verbally, is far harder than filling it in the document.
-
The bank's own sensitivity, not the borrower's
Where the submitted sensitivities look undemanding, the officer runs their own — commonly a larger movement, and frequently two variables at once. A borrower who has already published a genuinely severe case controls which downside the committee sees.
-
Group and connected exposure
Aggregated across the borrower's group, whether or not the borrower disclosed it. Exposure found rather than disclosed changes how everything else in the file is read, because it raises a question about completeness rather than about credit.
-
A recommendation with conditions attached
Rarely a plain yes. The recommendation typically carries conditions precedent and covenants, and those conditions are where the officer's residual discomfort is recorded. Reading them tells the borrower exactly what was not fully resolved.
The Cover That Matters Is the Worst Year, Not the Average One
Debt service coverage is the most examined figure in a credit proposal, and the one most often presented in a form that can't be assessed.
A stated cover of 1.6x could mean the first year, the average, or the worst year. Only the minimum is the test. A facility isn't repaid on average; a year where cash falls short of the instalment due is a default in that year.
An average conceals precisely the year that decides the outcome. Strong cover in years four through eight and thin cover in year two produces an average that reads well and a year two that is the entire risk.
Cover that starts strong and deteriorates describes something else: a contract expiring, an input cost escalating, competition arriving. A committee reads a declining profile as a question about tenor rather than amount.
Where our financial due diligence team rebuilds a model for an appraiser-facing report, the schedule is presented year by year with the minimum identified and its year named, before any average is mentioned.
Leading with the average and disclosing the minimum later reads as an attempt to anchor on the flattering figure, and it invites a rebuild of the model rather than an assessment of it.
Cash available for debt service isn't profit and isn't EBITDA. It's the cash the business actually has after tax, after the working capital the projection requires, and after maintenance capital expenditure the asset cannot operate without.
A cover ratio computed on EBITDA without stating what was deducted will be recomputed by the reader on their own definition. The recomputed figure is the one that goes in the note.
A facility is not repaid on average. It is repaid on a schedule — and the year the schedule is tightest is the whole of the risk.
Security Is Valued for the Day the Business Has Already Failed
Security is the second way out, assessed on the assumption the first didn't work. A going-concern valuation describes what an asset is worth to an operating business; security is realised when there is none.
The gap between those figures isn't a haircut applied for conservatism. It's the difference between two genuinely different questions.
A security section that answers the first while claiming to answer the second is the most common substantive weakness in an otherwise competent proposal.
Specialised plant with a single realistic class of buyer realises very differently from general-purpose land and buildings. Receivables are worth what remains collectable once the customer relationship has broken. Inventory in a distressed sale rarely realises anything close to carrying value.
Our valuation team prepares security valuations on the basis the security is actually realised on, and states that basis explicitly. An unstated basis is read as the more favourable one and then discounted accordingly.
Covenants are the third element, and borrowers misread them more often than either of the others.
A covenant is an early-warning instrument, not a penalty. That's why levels are set against the base case rather than the point of distress: one that only breaks when the business is already failing has given nothing.
A borrower who accepts a tight covenant to secure approval has bought approval with future flexibility. A technical breach in an otherwise fine year consumes management time, triggers a waiver process, may reprice the facility, and appears on the file permanently.
The interaction is where the real exposure sits. A covenant that trips in a moderate downside forces renegotiation exactly when the security has lost most of its realisable value.
One cover ratio, read four ways
A borrower submits a proposal for a facility of 120 over an eight-year tenor, and the summary page states debt service cover of 1.62x. Below is what that single figure becomes as the credit officer works through it. All figures are illustrative only, and the treatment of any particular item depends entirely on the facility, the institution and the evidence available.
- As stated Cover of 1.62x, as presented
The summary page quotes a single figure with no schedule behind it. Nothing about it is yet wrong; it simply cannot be assessed. The officer's first request is the year-by-year schedule, and the request itself delays the file by a cycle.
- Recomputed The average, once the schedule arrives — 1.62x
The schedule confirms the figure: 1.62x is the arithmetic mean of cover across the eight years. It is an accurate number describing something the committee does not size against, and it now sits in the file as the borrower's framing.
- The test The minimum, in year two — 1.18x
Cover in the second year of the tenor, as the plant ramps toward design capacity while full debt service has already commenced. This is the year the facility has to survive, and 1.18x against a covenant of 1.20x is a breach in the base case, before any downside has been applied.
- Redefined The minimum on the bank's definition — 1.04x
The borrower computed cash available for debt service from EBITDA. The officer reinstates tax and the maintenance capital expenditure the asset requires to hold capacity, both of which the projection omitted. The minimum year moves from 1.18x to 1.04x on the same cash flows.
- Sensitised The minimum at 90% of forecast volume — 0.94x
A ten per cent volume shortfall in the ramp years, which for a plant commissioning into an untested market is a mild rather than a severe case. Cover in year two falls below 1.0x: the facility does not service itself from its own cash flow in that year.
- Restructured Extend the moratorium by four quarters — minimum 1.31x
The finding is not that the project is uncreditworthy. It is that the amortisation profile was set against a ramp the project cannot deliver. Deferring the start of principal repayment moves the minimum year to year four, where capacity is established. The same project, the same amount, a structure that matches the cash flow.
Nothing in that sequence involved a dispute about the project. The volume assumption was not challenged, the cost estimate was not reopened and no additional security was requested. The entire movement came from asking for a schedule rather than a ratio, applying the lender's definition of cash available rather than the borrower's, and testing a modest shortfall — after which the structural fix was obvious to both sides. A borrower who had run that sequence before submitting would have proposed the longer moratorium themselves, and the file would have been assessed as competent rather than as optimistic.
A Returned File Is a Process Failure; a Rejected One Is a Credit Failure
Borrowers treat any outcome short of approval as the same outcome. A file returned for further information and one declined on credit grounds are different events, with different causes and different remedies.
Confusing them leads borrowers to fix the wrong thing: to strengthen a case that was never weak, or resubmit an unchanged file declined for a reason no additional information addresses.
A return is a documentation and evidence failure. The committee couldn't assess the proposal, so it didn't.
The commonest triggers are mundane. An assumption with no stated source. A projection that doesn't reconcile to the audited accounts it claims to build from. A statutory approval described as in process without a date. A security valuation on the wrong basis.
A delayed sanction moves the drawdown, the drawdown moves commissioning, and commissioning moves the first year of the cash flow the whole case was built on.
A rejection is a judgement. The committee understood the proposal and concluded the risk isn't one the institution wants at the price and structure offered.
Those causes are structural: cover that doesn't hold in the downside at any sensible amount, a sector the institution is reducing exposure to, a group history that can't be got past, a repayment case depending on a single counterparty.
More information changes none of them. What changes them is a different structure: a smaller facility, a longer tenor, more promoter contribution, additional security.
The distinction is worth establishing explicitly, because the language used doesn't always make it obvious and both can arrive in similarly worded correspondence.
Asking directly which occurred, and if a return, exactly which items are outstanding, is legitimate and credit officers generally answer plainly.
| Outcome | Typical trigger | What it means | What remedies it |
|---|---|---|---|
| Return | An assumption with no stated source | The reader cannot test the revenue line, so cannot assess the repayment case at all. | Cite the basis: a signed contract, an independent study, an operating history, a published tariff. Any traceable source, stated on the page. |
| Return | Projections that do not reconcile to the audited accounts | The starting point of the forecast disagrees with the filed position, so the whole projection is unverified. | A visible bridge from the last audited figure to the first projected one, with each reconciling item explained. |
| Return | A cover ratio with no year-by-year schedule | The minimum cannot be seen, so the sizing test cannot be performed. | The full schedule, with the minimum year identified before any average is quoted. |
| Return | Statutory approvals described as "in process" | An unquantified timing risk sits directly on the commissioning date the cash flow depends on. | The application reference, the date filed, the expected determination window, and what the position is if it slips. |
| Decline | Cover fails in the downside at any sensible facility size | A credit judgement. The project may be sound; this debt quantum is not supportable by it. | Restructure — less debt, longer tenor, longer moratorium, more equity. Not more documentation. |
| Decline | Single-counterparty or single-approval dependence | The repayment case has one point of failure, and the institution is not being paid for concentration. | Diversify the dependence, or secure it contractually in a form that survives the counterparty's own difficulties. |
| Decline | Sector or exposure policy | Nothing about the borrower. The institution is managing its own concentration. | A different lender class. This is not a document problem and cannot be solved by improving the document. |
A generalised description of common lending practice, not a statement about any institution's decision criteria. Outcomes turn on the facts of the proposal, the institution's policy at the time and the evidence available; nothing here is a prediction, and no cause listed produces any outcome on its own.
When Several Lenders Must Agree, the Document Has to Survive Every One of Them
A facility needing several institutions changes the problem in a way that isn't additive. The document is now read by several credit teams, under several policies, each producing its own note for its own committee.
A lead typically arranges the facility and conducts the primary appraisal. But participation is a decision each institution takes for itself.
A participant who isn't comfortable doesn't negotiate. They decline to participate, leaving the lead with an unfunded portion and the borrower with a problem that has nothing to do with the merits.
So the document must satisfy the most conservative reader in the group, not the average one.
A treatment the lead accepts and one participant doesn't is a treatment that fails, because the participant's alternative to arguing is simply to withdraw.
Sequencing compounds it. Queries arrive from different institutions at different times, and an answer given to one doesn't automatically reach the others.
Where an answer materially changes an assumption, a version problem develops. Institutions appraise different cases while believing they appraise the same one, and the discrepancy surfaces late, at documentation, when it's most expensive.
In the mandates our advisers run on consortium facilities, every substantive query response is issued to all participants and folded back into a single controlled version. The alternative is discovering at signing that two institutions sanctioned different projects.
The lead's own position is worth understanding, since borrowers mistake it for neutrality.
The lead has arranged the facility, usually taken the largest ticket, and has a reputational stake in the participants being satisfied. That makes them an ally on process and a demanding counterparty on substance.
A lead who anticipates a participant's objection raises it themselves, early and firmly. A borrower who reads that as the lead becoming difficult has misread it.
Build the Schedule and Watch the Minimum Separate From the Average
The instrument below builds a level-annuity amortisation across the tenor and computes cover in every year of it. It reports the minimum, the year the minimum falls in, and — separately and more quietly — the average, because the two are different figures and only one of them is the sizing test.
The readouts worth watching are the gap between the two and where the minimum actually falls. Because principal is deferred through the moratorium and then retired over fewer years, debt service steps up the moment the moratorium ends — so the minimum lands in the first year after it, not in year one. That is the profile the worked example above describes, and it is the reason a cover ratio quoted without a year attached tells a credit officer almost nothing.
Three structural levers move it. Lengthening the tenor lowers the instalment and lifts every year; lengthening the moratorium defers the step into a year where cash flow is stronger, but compresses amortisation into fewer years and raises the instalment when it arrives; raising the interest rate does the reverse of both. Watching which lever moves the minimum year — rather than which one moves the average — is the whole of the exercise.
- During the moratorium (years 1 to M): debt service = D × r — interest only
- After it, principal amortises over the remaining years A = N − M
- Level annual debt service = D × r ÷ (1 − (1 + r)^−A)
- CFADS in year n = CFADS in year 1 × (1 + g)^(n−1)
- DSCR in year n = CFADS in year n ÷ the debt service falling due in year n
- Minimum DSCR = the lowest DSCR across the N years — the sizing test
- Average DSCR = the mean DSCR across the N years — not the sizing test
Debt service steps up at the end of the moratorium, because the same principal is now retired over fewer years. That step is why the minimum cover commonly falls in the first year after the moratorium rather than in year one — the single most useful thing this schedule shows, and the reason a cover ratio quoted without a year attached cannot be assessed. Set the moratorium to zero for a plain level annuity across the whole tenor. At a zero interest rate the annuity reduces to straight-line amortisation, handled explicitly rather than dividing by zero. Cash available for debt service is taken as given here; in a real appraisal it is itself a contested derivation, after tax, working capital and maintenance capital expenditure.
Debt Service Cover Across the Tenor
Seeded with the illustrative facility above. All figures in millions; the currency is immaterial to the mechanism.
The minimum falls in year 3 at 1.08x against a covenant of 1.20x. The average across the tenor is 1.53x — the figure a borrower's own paper most often leads with.
The average is 0.45x above the minimum. A facility sized on the average would be approved on a year that never has to be survived.
- Year 1 — interest only 2.28x
- Year 2 — interest only 2.42x
- Year 3 — minimum 1.08x
- Year 4 1.14x
- Year 5 1.21x
- Year 6 1.28x
- Year 7 1.36x
- Year 8 1.44x
- Debt service steps from 11.4m during the moratorium to 27.2m once principal begins — which is why the minimum falls in year 3, not year one.
Illustrative only. Nothing entered here is transmitted or stored; the instrument runs entirely in your browser.
Write for the Officer Who Has to Defend the Recommendation
Everything above reduces to a single reorientation.
The reader who decides the outcome isn't the committee. It's the credit officer who writes the note the committee reads, and that officer isn't looking to be persuaded.
They're looking to establish what they can state as verified, what they must attribute to the borrower, and what they'll have to defend if the facility doesn't perform.
A document making each of those easy to determine gets recommended, because it has done the officer's work rather than asked them to do the borrower's.
In practice that means a small number of habits, none requiring additional analysis.
Source every material assumption on the page where it's used, not in an annexure.
Present cover as a schedule with the minimum named before any average appears.
Value security on the basis it's actually realised on, and say which basis that is.
Publish a downside severe enough that the reader doesn't feel the need to construct their own.
List the risks the officer will list anyway, with whatever mitigant genuinely exists, and say plainly where none does.
None of this is adversarial. A credit officer and a competent borrower want the same thing, a facility that performs, and the officer's scepticism is a professional obligation rather than a posture.
A borrower who has genuinely tested their own case finds the questions coming back are the questions they already answered. The process becomes an exchange of evidence rather than a contest of framings.
How this work is carried out
Projectzo has prepared project reports, CMA data and credit documentation read by bank credit committees since 2010, across 22 countries. Each mandate is assigned a single senior advisor, from scoping through final delivery, and is reviewed adversarially by a second senior reviewer before release — the questions a committee will ask are put to the document first.
Commission a Detailed Project Report