Last updated: October 26, 2026
Staff certificate review with contract analysts while verification is automated, then shrink the bench as each field goes live. A two-seat bridge, one mid-level financial analyst and one mid-level data analyst, models at about $92,000 for three months at national medians (KORE1 Contract Rate Index 2026-Q4, as of October 2026).
Picture a redacted borrowing base certificate on your desk. One signature. A request to draw. Twenty-two lines of arithmetic that belong to the borrower, and an aging behind it that belongs to nobody yet.
Three fields on that aging will be misread by any machine you buy this year. The sign on a credit, the account debtor’s name, and the date behind an aging bucket. I put dollars on each of them in the three fields a machine misreads. This piece starts one step later, at the question a head of portfolio operations asks me next. Who reads those fields while the machine is learning to?
Most desks answer with a requisition or a vendor demo. Both are guesses. What holds up is a short, priced bridge with an exit condition written down before anyone starts, because the bridge exists to be shortened and the contract should say how. KORE1 staffs these seats through its accounting and finance staffing practice, so I’m a guest with something to sell on a page about whether to buy it. The reasoning below works if you hire the analysts yourself.
Is Your Borrowing Base Review a Data Entry Job or a Fraud Control?
Both. That is the problem.
Two jobs share one desk, and they age differently, which is why a review that looks like one workload keeps turning out to be two when you try to staff it or automate it. The first is retyping, which means summing the column, comparing it with the total the borrower printed, tying the aging to gross receivables on the certificate, and tying this month’s opening balance to last month’s close. All arithmetic. A script does that in seconds, and a person doing it by hand is spending a salary on subtraction.
The second job is reading, and it means deciding whether this customer is an affiliate of the borrower under a different spelling, whether that invoice got younger between months, and whether the credit memo sitting in the current column is really a past-due credit that someone moved. The borrowing-base review is a fraud control disguised as data entry, and the retyping is the disguise.
The regulator is blunt about where the weight sits. The Comptroller’s Handbook on asset-based lending (version 1.1, January 2017) says, “Fraud is a frequent cause of loss in ABL. Regular collateral monitoring and timely field audits are the best deterrents to fraud-related losses.” Read that twice. Monitoring is the control. Typing is not. The handbook is published by the OCC, and the same reasoning carries over to a factoring desk, where a schedule of accounts arrives in place of a certificate and the debtor-name and invoice-date problems arrive with it.
So the useful question about any line of the review is which job it is. Automate the first job first. Keep a person on the second until the system has earned it, field by field.
| Review step | Which job | When a system can own it | Who owns it until then |
|---|---|---|---|
| Column totals against the printed total | Retyping | Day one | Analyst spot-checks |
| Aging total to gross receivables on the certificate | Retyping | Day one | Analyst spot-checks |
| Opening balance to last month’s close | Retyping | Day one, once last month’s file is stored | Analyst |
| Sign on each credit | Reading | After the column-sum test catches a doubled credit | Analyst clears every tag |
| Invoice date and re-aging | Reading | After two months of carried-forward ledger | Analyst |
| Account debtor name across spellings | Reading | After a debtor master exists and has been checked by hand | Analyst, the longest |
| Cross-aging, contra accounts, concentration limits | Judgment | Rules yes, exceptions no | Analyst, for good |
When Do Contract Analysts Beat Automation, and When Does Automation Win?
Analysts win where the rule isn’t written down yet. Automation wins where it is.
That sounds like a slogan, so here is the test I use on each field, which is whether you could write the check as a formula that a new hire could run without asking anyone a question. The doubled credit passes. A column sum that misses by exactly twice one value is a formula, and the system should own it. “Is this the same customer?” fails, because the answer lives in three years of the borrower’s habits, and the first machine to guess at it will guess with confidence.
Analysts win in a handful of situations, and they are easy to spot in advance.
- A new borrower’s first three certificates, when nobody yet knows which system produced the aging or how it prints a credit.
- Any month the debtor master changes. A changed master can move a concentration limit without a single invoice moving.
- The weeks after a borrower swaps its ERP or hires a new controller, when formats shift without notice and the tie-outs start failing for reasons that look like fraud and are not.
- Every tag that needs a reason a credit committee can read.
Automation wins the rest, mostly on volume and consistency, because a person on the four-hundredth line of the day makes a worse tie-out than a script that checks all four thousand lines the same way every single time.
One desk needs neither. A shop with a handful of borrowers and one analyst who knows every aging by heart should hire nobody, buy nothing, and write the three checks into a spreadsheet. I would tell a client that to their face, and KORE1 would not get the seat.
For desks past that point, the buying format matters, and contract staffing is hourly by design. KORE1’s Devin Hornick described it on an internal call this month. “Hourly, which we prefer. It’s really more of a staffing model to do the hourly billing and you manage the person.” You manage the person. That fits this seat well, because the analyst works inside your credit policy, under your reason codes, on your borrowers’ files, and you supervise the work directly, which is a different arrangement from handing a project to someone and waiting for a deliverable. The two roles to look at are contract financial analysts for the reading and contract data analysts for the ledger. KORE1’s credit analyst staffing page covers the neighboring credit seat.
How Do You Size the Bridge Team by Certificates per Month?
Start with a log, not a ratio.
Pull one month, and for every certificate record the borrower, the time it arrived, the time it was cleared, the minutes an analyst actually spent on it, and how many lines got questioned. Five columns. If nobody logged it, one person can rebuild most of it from email timestamps in about a week, and the gaps in that rebuild are themselves a finding.
Then do the sum. Hours of review per month equal certificates times average minutes touched, divided by sixty, and dividing that by the productive hours one analyst really has once meetings, month-end and vacations come out gives you your seats. Size to the busiest week and not the average, because certificates cluster around month-end and around the weekly borrowers, and an average hides exactly the week that breaks the desk.
An illustration, with invented inputs. Say the log shows 80 certificates, a median of 75 minutes touched on each, and 120 productive hours per analyst per month. Eighty times seventy-five, divided by sixty, is 100 hours. A hundred divided by 120 is 0.83 of a seat. Those inputs are mine, picked for the arithmetic. They are not a benchmark. I haven’t seen a certificates-per-analyst figure published anywhere that I would trust, and I would treat any number that did not come out of your own logs as modeled.
The second seat is easy to skip, and skipping it is how bridges fail. The analyst who clears the tags cannot also build the ledger that makes the tags go away. Someone has to hold the debtor master, the alias list, last month’s aging carried forward and the tie-out rules, in a form a machine can run. In the first quarter that is data analyst work. It hands to a data engineer if the build outgrows a spreadsheet, and the build itself is described in borrowing base verification automation.
From KORE1: Contract Rate Index 2026-Q4, the cost of the bridge
A mid-level contract financial analyst bills $72 to $94 an hour nationally, median $82 (KORE1 Contract Rate Index 2026-Q4, modeled). A two-analyst borrowing base bridge team models at about $92,000 for three months at national medians.
| Seat | Bill rate range | Median | 3-month cost | 6-month cost |
|---|---|---|---|---|
| Financial analyst, mid, United States | $72 to $94 | $82 | $42,400 | $84,800 |
| Financial analyst, senior, United States | $94 to $114 | $106 | $55,000 | $110,000 |
| Data analyst, mid, United States | $82 to $111 | $95 | $49,600 | $99,200 |
| Data engineer, mid, United States | $98 to $122 | $110 | $57,200 | $114,500 |
| Financial analyst, mid, New York | $92 to $121 | $102 | $53,200 | $106,400 |
| Financial analyst, mid, Chicago | $72 to $92 | $81 | $42,000 | $83,900 |
| Financial analyst, mid, Dallas-Fort Worth | $71 to $91 | $80 | $41,400 | $82,800 |
| Financial analyst, mid, Atlanta | $72 to $92 | $80 | $41,600 | $83,200 |
| Bridge team: financial analyst plus data analyst, mid, United States | Combined | $177 an hour | $92,000 | $184,000 |
New York runs about $20 an hour above Dallas-Fort Worth, Chicago and Atlanta for the same mid-level financial analyst seat. The data engineer row is the build seat for the verification layer and is not part of the bridge total. Other roles and metros are on the contract staffing rate card for IT, ERP and finance roles.
Price a borrowing base bridge team. Request a bill rate quote with your certificates per month, the number of seats, a start date, the length of the bridge and your metro. KORE1 will send back a range, and the log from the section above is the best attachment you can give it.
What Breaks If the Bridge Team Leaves Before Verification Is Automated?
Three things, in the order they tend to happen.
Tags pile up first. An open tag has an owner or it has a default, and without the analyst the default decides, so either the tagged amount stays out of availability and the borrower phones the relationship manager on Friday afternoon, or somebody clears the tags in bulk so the draw can fund. The second path is how a control stops existing without anyone deciding it should.
The reason codes walk out next, meaning which borrower’s system prints credits with a trailing minus and which debtor is really the borrower’s cousin under a different spelling. One person is the database, and a contractor becomes that person in about six weeks, which is the whole value of the contractor and also the exposure. I wrote about counting what would leave with that person. The fix is dull. Put the codes and the alias list into the ledger before the last day, as data, and skip the handover memo.
Last, nobody can say whether the system is right. A system that replaced the analyst has nobody left to disagree with it. This is where I run a replay. Take last month’s certificates, run them through the system, and compare its tags with the list the analysts actually cleared, scoring the hits and the misses as you go. A seat comes off the schedule only after the replay agrees on the fields that seat was carrying, and a replay on your own files is the same method used in the diagnostic that replays your last close.
Write the exit condition down before the start date. Mine reads like this: the analyst comes off a field after the replay matches the cleared list for two consecutive month-ends, with the first replay run at the end of month 2 so month 3 has room to correct what it finds. Pick your own number now, because the week the build slips is the worst week to negotiate one. Hourly contracts stretch a week at a time, and a few extra weeks of one analyst cost far less than a control with nobody sitting on it.
How Does the Bench Shrink as Fields Go Live?
Roughly the order of that first table. Tie-outs go live in week one, since they are three formulas, namely the column sums, the aging against the gross receivables line, and the opening balance against last month’s closing line. The sign check comes second. Once the column-sum test catches a doubled credit like a $61,480 memo, the analyst stops eyeballing every credit and touches only the tagged ones. Re-aging drags, because a machine needs two month-end agings from the same borrower before it can say an invoice got younger. Debtor name is last. It needs the debtor master, the master needs the alias list, and the alias list lives in one analyst’s head, spelled four different ways, until somebody makes them write it down.
Last to go live, and the first to tempt anyone into a shortcut. The usual one is a fuzzy-match setting turned up until the debtor tags go quiet. On a dashboard that looks like progress. It is a control switched off.
A bench that follows that sequence looks like this.
From KORE1: Contract Rate Index 2026-Q4, an illustrative six-month bench
| Phase | Fields live | Contract seats | Months | Cost |
|---|---|---|---|---|
| Build the ledger, clear every tag | Tie-outs | Financial analyst and data analyst | 1 to 3 | $92,000 |
| Sign and invoice date proven on replay | Tie-outs, sign, invoice date | Financial analyst | 4 to 6 | $42,400 |
| Debtor master proven on replay | All three | None, an in-house owner clears the tags | 7 on | $0 contract |
| Six-month total on this schedule | $134,400, against $184,000 for both seats held |
The difference, $49,600, is what dropping the second seat at month three is worth when the replay says the ledger is ready. Move the phase lengths and the saving moves with them. The tags still need a human at the end, so the third row ends in an employee and not a vendor, and if that employee does not exist yet, hiring one during the bridge is the real work of the six months.
What Portfolio Operations Asks Before Approving the Seats
Realistically, what does a mid-level contract financial analyst cost for borrowing base review?
A mid-level contract financial analyst bills $72 to $94 an hour nationally, median $82, or about $42,400 for a full-time quarter and $84,800 for six months. That is modeled from BLS OEWS May 2025 wages for SOC 13-2051, a 24% employer burden and KORE1’s observed bill bands, as of October 16, 2026. Metro moves it. The New York median is $102 against $80 to $81 in Dallas-Fort Worth, Chicago and Atlanta.
Does a review bridge really need a data analyst too?
Usually yes, for the first quarter, because the analyst clearing the tags cannot also build the debtor master and the carried-forward ledger that make the tags disappear. A mid-level contract data analyst bills $82 to $111 an hour nationally, median $95, or about $49,600 for three months (KORE1 Contract Rate Index 2026-Q4, modeled). Release the seat when the replay agrees with the cleared list.
How many certificates can one analyst review a month?
Measure it on your own review logs, because I have not seen a published figure I would trust. Log arrival time, cleared time, minutes touched and lines questioned for one month. Then divide hours of review by the productive hours one analyst has. The sizing section above works an example with invented inputs.
Should an ABL lender hire analysts or automate borrowing base review?
Both, in sequence. Hire contract analysts to clear the review queue now, automate the checks that are formulas first, and release each analyst seat as the system proves a field against the cleared list. A desk with a handful of borrowers and one analyst who knows every aging can skip both and keep the three checks in a spreadsheet.
How do lenders track borrowing base compliance without re-checking every collateral schedule by hand?
They let rules re-check the arithmetic and let people read only what the rules flag. Column sums and the aging-to-certificate and opening-to-closing tie-outs run on every file. The sign, the debtor name and the invoice date get a tag with a reason when no rule can settle them, and an analyst reads the tags and not the schedule. That is the regular monitoring the OCC points to.
Start With the Log
Pull one month of certificates and give each one five columns, borrower, arrived, cleared, minutes touched and lines questioned. If the minutes column is empty because nobody tracks it, that is already your answer to the staffing question, and it is the first thing to fix before anyone signs a requisition or a purchase order.
The data behind the cost table is public in outline. The wage basis is the BLS Occupational Employment and Wages release for May 2025, and the metro and national tables sit at the BLS OEWS tables page. Check my arithmetic against them if you like, since the cost table is only as good as the wage data under it and the 24% burden on top.
If your log comes back with something strange in it, I’d like to see it. Connect with me on LinkedIn. If the log says you need a seat filled, you can ask KORE1 about a contract placement.
