Covenant Compliance Monitoring: How a Direct Lender Tests Each Quarter’s Certificates Against Its Own Deal Definitions

Last updated: November 12, 2026

By Khurram Tehseen, FSA, Founder and Principal at Operating Alpha

Two pages from one borrower’s file, laid side by side.

Example only. The borrower, the sections and the figures are invented.

Circle the add-back line. The borrower took $4.6M, and the cap on the second page allows $2.76M. Run the quarter again with the cap in and leverage comes out at 4.29x, where the certificate says 3.86x and the test level is 4.25x. Same borrower. Same quarter. A different conversation with credit.

Everything on the certificate ties to the financials. The arithmetic holds. Where it parts from the agreement is a cap, and a cap rarely shows on the certificate itself, so you see it only when somebody sets the two pages next to each other.

Somebody has roughly three weeks to do that. Certificates and financials show up together after quarter end, and the analyst who knows which deal caps what is often the one closing the books. Dozens of pages. One reader. A calendar that does not move, and a credit team waiting on the answer, because a ratio that is wrong in the middle of a quarter tends to stay wrong until someone checks it.

What follows is the testing method I use, in the order I would hand it to a new analyst. KORE1 places contract analysts through its accounting and finance staffing practice, so one section below is about seats and says so plainly. The rest works with people from your own bench. If you have wondered whether a tool can read the agreement for you, I ran that question in a test of AI reading a credit agreement. Reading it is half of the work. Applying it is the other half.

What Does Covenant Compliance Monitoring Test Each Quarter?

Covenant compliance monitoring is the quarterly work of rebuilding each borrower’s financial covenant ratios from its compliance certificate, using the deal’s own defined terms, and recording every place the result differs from what the borrower reported.

Most direct lending deals carry some mix of these tests.

• Total leverage, which is debt against Consolidated EBITDA, usually with a maximum that steps down over time.
• Fixed charge coverage, a minimum, where “fixed charges” is another defined term with its own list.
• Interest coverage. Simpler. Still not simple enough to skip.
• A liquidity floor on some deals, which can be tested on a different date from the rest.

Levels and dates matter as much as the ratios. A leverage test that reads 4.25x this quarter may read 4.00x two quarters on, and the step-down can sit in an amendment instead of the original agreement. The analyst needs the test level for this quarter, for this deal, as amended. Read the record. Agreement first. Then every amendment after it, in order. Skip the amendments and you will test a borrower against a level that stopped applying two quarters ago, which produces a clean result that nobody on the credit team should rely on once the amendment surfaces.

Why Can’t One Formula Test Every Deal?

Because every credit agreement defines Consolidated EBITDA its own way. Same two words, different arithmetic.

Monitoring analysts put the problem plainly. Our credit agreements define EBITDA add-backs differently deal by deal. They do. One deal caps add-backs at a share of EBITDA before the add-backs, and the next says after. One lets restructuring charges through without a limit and caps only run-rate savings, while a third counts a pro forma adjustment for an acquisition over twelve months where another uses six. Unrestricted cash netted against debt is capped on some deals and uncapped on others.

A formula built once in a workbook tab and copied down thirty rows is wrong on most of those rows. It is right on the ones somebody checked. The tool you already bought probably has one EBITDA field, and your agreements hold as many definitions as you have deals. That gap is where the days actually go, because every exception turns into an email thread about what the agreement meant, written after the quarter when it should have been settled before it.

What Goes on the Per-Deal Definition Sheet?

Each borrower gets one page. On it go the defined terms the tests use, every one with its section reference, along with the add-back caps, the test levels by quarter and their step-downs, the cure rights, the reporting deadline, and the amendment that last touched any of it. Whoever tests the certificates fills it in before the first one arrives, working from the agreement and every amendment after it. Then the person who knows the deal reads it for mistakes.

Here is part of one, with invented values.

Look hardest at the last column. Say the leverage level stepped down in the second amendment. Whoever tests Borrower 14 in the second quarter of 2027 needs that level, and the original agreement gives the wrong one. So tie every term to its source. Where a term has no clear source, write “unknown” and ask whoever holds the executed documents, which is a habit I describe in how to tag what your data cannot support.

In most shops this page already exists. Nobody has written it down. One person is the database, and everyone knows their name.

How Do You Recalculate a Certificate Instead of Retyping It?

Retyping copies the borrower’s ratio into your tracker and checks that it sits under the test level. Recalculating rebuilds the ratio from its parts. The work has three steps, and the middle one needs a person.

Start by tying the inputs, with debt balance going to your own records of the facility and EBITDA components going to the borrower financials that came with the certificate. A certificate that fails to tie to its own financials already has an exception, and no ratio has been computed yet. Next comes the middle step. Compute each ratio from the definition sheet, applying the cap, the pro forma rule, the cash netting and whatever else this deal’s wording says. Last, put your figure beside the borrower’s. A variance goes in the log whether or not it crosses the test level, since one that fits inside this quarter’s headroom can fall outside it next quarter.

Extraction tools can capture the figures from the certificate and the financials, and I would take that saving. Applying the definition is a different job, and it belongs to someone who has read the agreement. A tool can read $4.6M off page two. It has no idea your cap sits at 20% of the pre-add-back figure. If a tool does feed the process, somebody owns the queue behind it, a subject I took up in staffing the exception queue behind an extraction tool.

Before you trust any setup, replay a closed quarter. Run last quarter’s certificates through the same steps and compare your recalculated ratios with what the team reported. A match means you have a method. A miss hands you the next three definition-sheet corrections. Either result is useful.

What Goes in the Exception Log, and Who Sees It?

The exception log records every difference between the borrower’s figure and yours, with enough on each row that someone who was not there can pick it up. Eight fields do it. Borrower, tranche, definition version, borrower figure, recalculated figure, variance, cure date, owner.

Example only. Borrower 14 and every value are invented.

The definition version field looks fussy until the first amendment lands mid-year. After that, it explains why two quarters of the same deal were tested to different levels. The owner field holds a name. Not a team.

The log is also what the fund keeps when a contractor rolls off. Their recalculations are in it, with the version they used. That is the handover. Sign-off, waivers and anything an LP will read stay with the fund, and each log row has an employee’s name on it for that reason.

What Breaks When Testing Runs Late?

Two things, neither dramatic. Credit hears late. Then credit hears everything at once.

A borrower with thin headroom, or a recalculated miss, reaches the credit team after the cure window has already started to run, and the watchlist meeting that would have taken it up has come and gone. The quarterly reporting pack waits on a test result still sitting in an analyst’s queue. An analyst who finds the variance in week four is retesting while the borrower, busy with the next month’s close, answers questions slowly. That is the operating cost. Nothing more. What a missed test means for a given deal belongs to the fund’s counsel and its credit committee.

How Many Extra Hands Does a Quarter-End Peak Need?

This section is the one place staffing comes up, and it carries the disclosure from the top of the page. Discount it as you see fit.

Size the seats by certificates per analyst per week, measured at the busiest week and not the average. Count the certificates that land in the three weeks after quarter end. Time a full retest on your own deals, meaning tie-out, recalculation and the log entry. Then divide by the testing hours one analyst really has once close, meetings and vacations come out of the week.

An illustration. Sixty certificates in three weeks, three hours per retest, and thirty testing hours per analyst per week. That is 180 hours of work against 90 hours per analyst across the window, so two analysts. Those inputs are mine, chosen for the arithmetic, and they are not a benchmark. I have not seen a certificates-per-analyst figure published that I would trust, and the numbers that matter come from your own log.

Contract analysts fit this shape because the peak is a quarter-end shape. KORE1 partner Devin Hornick said the appeal in one line, that “you can get them quick… flexible… to scale up, to scale down as needed.” They bill hourly and you manage them, which is where the definition sheet and the log pay off, since they are what let you manage someone who started on Monday. For scale, across 1,887 ended KORE1 contract assignments the median ran 116 days and a quarter ended inside 44 days (KORE1 placement records, n=1,887). About one quarter, then. The bench is smaller by the next one. If the analysts you need sit closer to credit than to operations, contract credit analysts are the seat. A scope of work for a contract analyst is a useful thing to have in hand before the first call, and the wider private credit data contractor practice covers the data seats behind the testing cycle. KORE1’s contract staffing page explains how those engagements run.

Questions Before the Next Certificate Date

So how long do borrowers have to deliver a certificate?

Each agreement’s reporting covenant sets its own deadline, so put it on the definition sheet and do not assume one. Take a credit agreement filed on SEC EDGAR. It requires financial statements not later than 45 days after each of the first three fiscal quarters and 120 days after the fourth, with a covenant compliance certificate. One deal, one term. Another loan agreement filed on SEC EDGAR runs several clocks at once, and which one applies turns on whether the borrower has drawn. Your agreements will say something different again.

Can software apply each deal’s covenant definitions automatically?

Software can capture the figures from a certificate and run the arithmetic, but a person who has read the agreement has to say which definition applies. Capture, yes. Apply, once the sheet exists. When the sheet is filled and checked, a good share of the rules can run as code, and the first test of that code is a replay of a quarter you already closed. A match with what your team reported gives you something to trust. A mismatch is the work.

Which definition applies after an amendment or waiver?

The definition in force on the test date, as amended. That is why the sheet names the amendment that last changed each term and each log row names the version it was tested against. A waiver covers what it names for the period it names and nothing beyond that, which is why the log row should cite the waiver by date next to the definition version it modified. Have counsel read the instrument itself. I am describing a working method here and I’m not giving legal advice.

Who actually signs off a covenant test?

A named employee of the fund, usually the head of portfolio operations or the credit officer, signs off the test. Contract analysts prepare the recalculation and the log. They do not sign. Waivers and anything an LP will read stay with the fund too.

Fill In Three Sheets First

Before the next certificate date, fill in the definition sheet for your three largest deals. Use the agreement and every amendment, and mark anything you can’t source as unknown. Then test one certificate against each sheet. Count the variances.

If one of them comes back strange, I would like to see it. Connect with Khurram on LinkedIn.