A few months ago we sat in on a quarterly business review at a contract manufacturer in Michigan. The procurement director walked through a 22-slide deck. PO compliance: 99.2%. Spend under management: 87%. Savings reported: $4.1M. Supplier consolidation: down to 412 from 580 the prior year. Green dots everywhere.

Then the COO interrupted. "Why was the GE order four weeks late?"

Nobody had an answer on the slide. Because nothing on the slide actually measured whether procurement was doing its job. The deck was full of procurement KPIs. None of them predicted the outcome the rest of the business cared about.

This happens constantly. Procurement teams chase metrics that look like performance, score well on them, and then get blindsided when delivery slips or quality tanks. The standard "top 25 procurement KPIs" lists you'll find on every consulting blog are mostly junk. They confuse activity with outcomes and inputs with results.

Here's the short version. Roughly seven procurement KPIs actually correlate to real performance: supplier on-time delivery, PO acknowledgment cycle time, quote turnaround time, supplier date adherence, cost variance from PO to invoice, first-pass yield on supplier quality, and procurement cycle time. Everything else is either a vanity metric, a category descriptor pretending to be a performance metric, or a number so easy to game that tracking it actively makes you worse.

The rest of this is the longer version: what each one is, how to measure it, what movement means, and why the popular ones you've been reporting on for years are getting you nowhere.

Why Do Most Procurement KPIs Fail?

Most procurement KPIs measure procurement's view of itself, not procurement's effect on the business. PO compliance tells you whether buyers followed a process. It says nothing about whether parts arrived. Savings reported tells you what a buyer claims they negotiated, not what actually hit the P&L. Supplier count reduction tells you the directory got smaller, not that the supply base got stronger.

The test for any KPI: if it moves, does the business notice? If it gets better and your COO can't feel it, you're measuring the wrong thing.

The Seven That Matter

Supplier On-Time Delivery (OTD)

Definition: Percentage of POs (or line items) delivered on or before the supplier-confirmed ship date, measured at the receiving dock.

This is the only procurement KPI your customer actually cares about. If you can only track one thing, track this one. Best-in-class manufacturers run between 95% and 98% supplier OTD on direct materials. Anything under 85% means production is rebuilding the schedule every week and your planners are basically a manual escalation engine.

Two things people get wrong. First, they measure against the original PO date instead of the confirmed date. That conflates "supplier missed their commitment" with "buyer issued a PO with an unrealistic date." Track both separately. Second, they measure OTD at the supplier ship event instead of at receiving. A part that left China on time but is sitting in customs for nine days is not on time.

If OTD is dropping, the cause is almost never random. Sort by supplier and you'll find three or four offenders driving 70% of the misses. That's where the work is.

PO Acknowledgment Cycle Time

Definition: Hours or business days between PO issuance and supplier confirmation of the order (price, quantity, ship date).

This one is wildly underrated. Most teams don't track it at all, then wonder why orders silently fall through the cracks. We've seen median acknowledgment times of 6 days at manufacturers who think their suppliers are responsive. Six days. On a 30-day lead time part. That's 20% of your runway gone before anyone has confirmed the order exists.

Good benchmark: 80% of POs acknowledged within 48 hours. Excellent: 90% within 24 hours.

The reason this matters more than people realize is that an unacknowledged PO is a probabilistic delivery. The supplier might be working on it. They might have set it aside. The PDF might be in someone's spam folder. You won't know until the date slips. By then it's too late. The teams that fix PO tracking properly instrument this first, because every other downstream metric depends on suppliers actually confirming the order.

Quote Turnaround Time

Definition: Hours or business days between RFQ being sent and a complete, comparable quote being received.

If you're sourcing custom parts, this is your sourcing tempo. It controls how many bid cycles you can run on a part before tooling commits, how quickly you can react to a cost spike, and how much room you have to push back when an engineering revision drops mid-source.

Most procurement teams have no idea what their actual quote turnaround time is, because the data lives in 400 Outlook threads. When we've helped teams measure it properly, the numbers are usually worse than they assumed. One team thought their average was 4 days. The actual number was 11. The difference was the long tail of suppliers who needed three follow-ups before they responded, which got mentally written off as "well, that one's always slow."

Track it per supplier and per part category. If you ever do a proper RFQ time study, this is the first number you'll want.

Supplier Date Adherence (Schedule Attainment)

Definition: For multi-line POs or scheduled releases, the percentage of individual line items shipped on the committed date, weighted by either quantity or unit value.

Different from OTD on a single dimension that matters. OTD measures whether a PO arrived. Date adherence measures whether the schedule is holding line by line. If a supplier has a PO for 12 line items and ships 11 on time but the most expensive part is two weeks late, OTD might still register as a partial win. Date adherence catches the actual problem.

This is the metric that matters for anyone running a planned production schedule. If your MRP is regenerating weekly because suppliers can't hit individual line dates, no amount of PO compliance reporting will help.

Cost Variance: PO Price vs Invoice

Definition: Difference between the unit price on the PO and the unit price on the matched invoice, aggregated as a percentage of total spend.

This one is boring and unglamorous and it surfaces more real money than any "savings reported" deck ever has. Suppliers raise prices. They add freight charges. They invoice for a slightly different part number. They bill for the expedite even when the part wasn't expedited. If you're not actively matching invoice to PO at the line level, this is probably running 1-3% against you and nobody is catching it.

Best-in-class is under 0.5%. Most teams who measure it for the first time find they're at 2% or higher. On $50M of direct spend, that's a million dollars a year leaking. Real money, not negotiated-savings-on-a-slide money.

The reason this gets missed is that AP and procurement live in different systems. AP processes the invoice, three-way match passes within tolerance, payment goes out. Nobody flags the variance because the tolerance was set too wide. Tighten the tolerance to 1% and you'll find half a dozen suppliers who've been raising prices quietly for a year.

First-Pass Yield on Supplier Quality

Definition: Percentage of supplier-delivered parts that pass incoming inspection (or production use) on the first attempt, without rework, RTV, or deviation.

Quality teams measure this. Procurement often doesn't, which is bizarre because procurement chose the supplier. A supplier with 99% OTD and 80% first-pass yield is not a good supplier. They're a fast bleeder.

The honest version of this metric folds in everything that disrupts production: visible defects, dimensional out-of-tolerance, missing certs, paperwork errors that hold up receiving. Anything that means production can't use the part without intervention.

Track this per supplier per category. The relationship between price and first-pass yield is the actual story of supplier performance, and it's almost never on the dashboard. The lowest-bid supplier is rarely the cheapest supplier once you build in the cost of the 8% of parts you reject.

Procurement Cycle Time (Request to PO)

Definition: Calendar days from purchase requisition being submitted to PO being issued to the supplier.

This is the internal one. How long does it take your team to actually buy something once someone asks for it? On indirect, 3-5 days is reasonable. On direct, 5-10 days is normal but suspect; under 3 is good. Anything over two weeks and engineering will start ordering parts on personal credit cards, which they're already doing anyway.

The reason this matters: long internal cycle times push lead time pressure onto suppliers. If it took you 12 days to issue the PO and the part has a 30-day lead time, you've effectively halved your runway and now you need expedite fees. Most "expedite cost" budgets are really "slow internal procurement" budgets in disguise.

The Vanity KPIs to Stop Reporting

Now the fun part.

PO Compliance %

The percentage of spend that flows through proper POs instead of credit cards, expense reports, or after-the-fact paperwork. This is the metric every procurement deck leads with. It's also one of the easiest to game.

The dirty secret: PO compliance goes up when you make the system easier to use, but it also goes up when buyers retroactively cut POs for things people already bought. Both look identical on the dashboard. We've seen teams hit 99% PO compliance by having one buyer spend two days a month cleaning up after-the-fact requisitions. That's not procurement. That's filing.

PO compliance is fine as a hygiene check. It's not a performance metric. It does not predict whether supplies arrived, whether you paid the right price, or whether quality was acceptable. Stop putting it on slide 2.

Supplier Count Reduction

A favorite of consultants and CPOs who need to show transformation. "We reduced our supply base from 1,400 to 800 in 18 months."

OK, but what did it cost you? Supplier consolidation is sometimes the right move and often a disaster. The strategic suppliers you keep get pricing power as soon as they realize you have no alternatives. The long-tail suppliers you cut were often the ones with the flexibility to expedite a part nobody else stocks. Six months later you're paying premium rates to the consolidated supplier and re-onboarding two of the suppliers you cut.

Counting suppliers isn't a strategy. The right question is whether your supply base has the capability and resilience to deliver. That's a qualitative judgment, not a number.

Savings Reported

The single most fictitious number in procurement. Buyer negotiates a 7% reduction off a quote that the supplier inflated by 15% to give the buyer the negotiation room. Buyer reports the 7% as savings. Net effect on the P&L: zero. Or worse.

If your CFO can't reconcile reported savings to actual P&L impact (and they almost never can), the number is meaningless. The teams who report this honestly use cost-avoidance vs cost-reduction categories and reconcile a sample to invoice prices quarterly. Most teams don't.

If you want a real cost metric, use cost variance (PO vs invoice) and unit price trend over time on top-spend SKUs. Both are auditable. Savings reported is theater.

Spend Under Management

The percentage of total company spend that flows through procurement. Useful as a category descriptor. Useless as a performance metric. A team can have 95% spend under management and still be slow, expensive, and missing dates. The number tells you scope, not effectiveness.

Number of Sourcing Events Run

Activity dressed up as outcome. Running more events isn't better. Running the right events on the right categories with measurable cost or capability outcomes is better. We've seen teams who ran 47 sourcing events in a year and saved nothing measurable. We've seen teams who ran 6 and shifted $4M of spend to better suppliers. Count the impact, not the meetings.

How Do You Actually Instrument These?

Here's where it gets uncomfortable. Most procurement teams can't measure half the metrics above with any precision, because the data is scattered across the ERP, email, PDF attachments, supplier portals, and spreadsheets on individual buyers' laptops.

Take PO acknowledgment cycle time. To measure it you need: the PO issue timestamp (in your ERP), the supplier acknowledgment timestamp (in email), the parsed acknowledgment content to confirm price/qty/date (in the email body or PDF attachment), and a join key between them (often manually entered into a NetSuite custom field, if at all). If your buyer is forwarding acknowledgments to themselves and pasting confirmed dates into a spreadsheet, the data doesn't exist as data. It exists as 1,200 disconnected events.

This is the part of procurement most KPI dashboards quietly skip. They report the metrics that are easy to extract from the ERP (PO count, spend, supplier count) and ignore the metrics that actually matter because they require pulling structured signal out of unstructured email.

A few practical things that help:

• Stop trying to extract metrics from your ERP. Your ERP wasn't built for procurement, and especially not for measuring the operational tempo of supplier interactions. Use it as the system of record, not the analytics layer.

• Treat email as the actual source of truth for supplier communication, because it is. The acknowledgment, the date change, the price update, the expedite request, the apology: all in email. Until you can parse and timestamp those events, OTD and acknowledgment cycle time will both be guesses.

• Build a single source of truth for "supplier-committed date" separate from "supplier-promised date" and "PO date." Most teams collapse these into one field and lose the ability to attribute lateness correctly.

• Score suppliers on a proper performance scorecard that combines OTD, date adherence, and first-pass yield, weighted by spend. Single-metric scorecards always undersell the worst suppliers.

If you can't measure something, you can't manage it. If you measure the wrong things, you'll manage them, and they'll keep getting better while the business keeps getting worse. Which is exactly what was happening on that 22-slide deck in Michigan.

Where to Start

If you've been reporting on PO compliance, savings, and supplier count and want to switch, don't try to flip the whole dashboard at once. Pick three: supplier OTD measured at receiving, PO acknowledgment cycle time, and PO-to-invoice cost variance. Run them for a quarter alongside your existing reports. Compare what each one predicts. The vanity metrics will keep being green. The real ones will reveal which suppliers are actually performing, where time is actually being lost, and where money is actually leaking.

Then drop the vanity ones from the deck and watch what happens to the conversation. The QBR question stops being "why are all the dots green but delivery is late?" and starts being "we know exactly which two suppliers are driving 80% of the late receipts and here's what we're doing about it." That's a procurement team that has its hands on the right levers.

Instrumenting these KPIs is mostly a data problem, not a process problem. The numbers exist; they're just trapped in email threads, PDFs, and ERP fields that nobody updates consistently. Lumari reads your supplier email, parses acknowledgments and date changes, and writes the events back to your ERP so OTD, acknowledgment cycle time, and date adherence become actual dashboards instead of quarterly archaeology projects. If you want to know what your supply base looks like in real numbers, that's a faster path than another KPI workshop.