Staffing supplier scorecards suppliers cannot game

A supplier scorecard that only ever goes up is telling you something. It may not be about the suppliers.

This post is for anyone who measures staffing suppliers: a program owner running a contingent workforce program directly, or a managed service provider (MSP) running one on a client's behalf. MSP here means the staffing kind that runs a contractor program, not an IT managed service provider. The argument is simple. Suppliers respond to whatever you measure, so every metric on the scorecard needs a fixed definition, a data source the supplier does not control, and a named way it can be gamed.

Below is the metric-definition table I would start from. It gives definitions only, with no benchmark values, because a target borrowed from someone else's program says nothing about yours.

A supplier scorecard is only as fair as its denominators

Most scorecard arguments are really arguments about the bottom half of a fraction. Fill rate sounds objective until two suppliers compute it differently: one divides by the requisitions it was sent, the other by the requisitions it chose to work on. The second number will always look better, and it tells you less.

So the first rule of a vendor scorecard is that the client or the MSP sets every denominator, from records the supplier cannot edit. "Requisitions released to the supplier" is a fact in the system. "Requisitions the supplier accepted" is a choice the supplier made.

The second rule follows from an old observation about measurement, usually phrased as "when a measure becomes a target, it ceases to be a good measure". The best-known phrasing appears in a 1997 paper on audit in British universities (European Review, "Improving ratings": audit in the British University system, read 19 October 2026). A scorecard is a target the moment it decides who sees the next requisition. Design it on the assumption that every metric will be pushed as hard as it can be.

The supplier scorecard metric definitions

Each row gives the formula, where the data comes from, how the metric gets gamed, and the counter-measure. All counts are for one supplier over one scoring period. "Released" means requisitions the program sent to that supplier in the period, whether or not the supplier responded.

Metric	Formula	Data source	How it gets gamed	Counter-measure
Requisition coverage	Requisitions with at least one submission ÷ requisitions released	Requisition release and submission records	One token candidate per requisition to keep coverage high	Pair with submission quality; a token submission hurts that score
Time to first submission	Median hours from release to the first submission the hiring manager accepts for review	Release and submission timestamps	Fast, unqualified submissions to start the clock early	Count only submissions not rejected as unqualified
Submission quality	Interviews ÷ submissions	Submission and interview records	Submitting only on easy requisitions	Pair with coverage, which uses all released requisitions
Fill rate	Requisitions filled by the supplier ÷ requisitions released	Requisition outcomes	Ignoring hard requisitions so the denominator shrinks	Denominator is released, never "accepted"
Time to start	Median days from release to the worker's first day of approved time	Release date and the first approved timesheet entry	Reporting an agreed start date rather than the real one	Use the first approved timesheet entry, not a date the supplier enters
Ready on day one	Workers with every required onboarding step done before their first day ÷ workers started	Onboarding checklist completion dates	Asking staff to override missing steps	Count every override as a miss, with its logged reason
Early ends	Assignments ended early for supplier-side reasons ÷ assignments ended	Assignment end records and end reasons	Coding supplier-side ends as client-side	The client sets the end reason, not the supplier
Rate compliance	Invoice lines at or below the rate card ÷ all invoice lines	Invoice lines against the rate card for that role and site	Moving a worker to a higher-paid job code	The client sets the job code on the assignment
Invoice accuracy	Invoices that match approved time with no differences, on first submission ÷ invoices submitted	Invoice-to-approved-time comparison for the period	Withholding an invoice until it can be fixed quietly	Score the first submission; a credit note counts as a difference
Hiring manager rating	Average of a short, fixed-scale rating per completed assignment	A rating form sent when the assignment ends	Lobbying managers before the form goes out	Weight it lower than the record-based metrics, and show the response count next to it

Nine of the ten rows come from records. The tenth is an opinion, and I keep it on the scorecard because managers notice things the records miss. It should not carry the same weight as the rest.

Five rules that make a vendor scorecard hard to game

These are the rules I would write into the program policy before the first scorecard goes out. They are my framework, not an industry standard.

1. The program owns every denominator. Released, started and submitted are counted from system records, never from a supplier's own report.
2. Every metric has a partner. Speed is paired with quality, coverage with quality, fill rate with released volume. A supplier can push one number up only by letting its partner fall, and that shows.
3. Compare like with like. Score suppliers within a role category, not across the whole program. A supplier that fills hard engineering roles should not be measured against one that fills general administration.
4. Show "not enough data" below a minimum count. A supplier sent two requisitions should not get a fill rate. Set a threshold, such as ten released requisitions in the period (an illustrative figure), and show nothing below it.
5. Give suppliers their scores and a way to dispute them. A supplier that can see its numbers and challenge a wrong record helps you clean the data.

The fifth rule has a public precedent. The Federal Acquisition Regulation (FAR), the rulebook for US federal purchasing, sets out how agencies evaluate contractor performance in section 42.1503 (FAR 42.1503, Procedures, read 19 October 2026). It says evaluations should be "based on objective facts supported by program and contract or order performance data", lists minimum factors including quality, cost control, schedule and business relations, and gives the contractor up to 14 calendar days to submit comments, rebutting statements or additional information. Disagreements go to a review above the contracting officer.

A staffing program does not need federal procedure. It can borrow the shape: evidence-based scores, a short window for the supplier to respond, and someone other than the scorer to settle disputes.

Where the scorecard data should come from

The counter-measures in the table all depend on one thing: the record behind each metric exists independently of the supplier. In my view that is the real job of the system a program runs on. The scorecard is a query over those records.

Several of those records are ones VendOrca, the AI-native vendor management system (VMS), keeps. Each worker goes through an onboarding checklist built from the program's template before they become active (overview).

An assignment binds one onboarded worker to a rate card for a date range, and amending, extending or ending it adds a new version with the date it takes effect (key concepts). Program Admins and Hiring Managers approve or reject timesheets, and a rejection carries a reason (time and expense approval). The tie-out compares what was invoiced for a period with what was approved and lists every difference by kind (invoicing and tie-out).

Staff can ask for figures in plain language in Reports and get a figure, a chart and a short summary, with no report builder to learn (reports and Ask VendOrca). Suppliers sign in to a read-only portal that shows only their own invoices (supplier portal).

Frequently asked questions

What should a staffing supplier scorecard measure?

Measure what the supplier controls, from records it cannot edit: coverage, speed, quality of submissions, fill, readiness on day one, early ends, rate compliance and invoice accuracy. Add a manager rating if you want, but weight it below the record-based metrics.

How often should supplier scorecards be reviewed?

My preference is a monthly scorecard and a quarterly review with each supplier, so a single slow month does not decide anything. Whatever the cadence, use the same period for every supplier.

Should supplier scorecards be shared with suppliers?

Yes. A supplier that sees its own scores can fix its behaviour and can point out a wrong record. Give each supplier only its own numbers, and a set window to dispute them.

Is a vendor scorecard the same as a supplier scorecard?

I use the two terms interchangeably here: a periodic report that measures each staffing supplier against defined metrics. Outside staffing, the same names are also used for scorecards on suppliers of goods and services, which measure different things.

A scorecard does not make suppliers better on its own. Fixed definitions, denominators the program owns, a partner for every metric and a way for suppliers to answer back are what make the numbers worth acting on. If you want the records behind the scorecard in one place, you can Start free.

Sources

• European Review, "Improving ratings": audit in the British University system (1997), read 19 October 2026.
• Federal Acquisition Regulation, 42.1503 Procedures, read 19 October 2026.