Supplier performance scorecards that procurement actually uses: KPIs, SLAs and service credits

Why most supplier scorecards are ignored

Most scorecards fail in the same way. Someone builds a spreadsheet with twenty measures. Half of them have no reliable data source, the supplier disputes the rest, and by the third quarter nobody updates it. When the contract comes up for renewal, the decision is made on the relationship and the last escalation anyone remembers, not on the scorecard.

A scorecard gets used when three things are true. It measures only what someone will act on. Every number comes from a source both sides accept. And the result changes something the supplier cares about: money through service credits, volume through allocation, or the contract itself at renewal. The rest of this guide covers those three points in order.

How many KPIs should a supplier scorecard have?

Five to seven per supplier category is a workable ceiling. Fewer than five usually leaves out a dimension that matters, such as responsiveness or compliance. More than seven spreads attention thin, and the extra measures tend to be ones nobody owns. The test for each KPI is a single question: if this number turns red, who does what? If there is no answer, drop the KPI.

Pick across dimensions rather than stacking measures of the same thing. A logistics supplier scored on on-time pickup, on-time delivery and transit time is being measured three times for punctuality and never for damage or billing accuracy.

• Delivery or availability: on-time-in-full for goods, uptime or turnaround time for services.
• Quality: defect or rejection rate, first-time-right, rework.
• Responsiveness: time to acknowledge and time to resolve issues, by severity.
• Commercial accuracy: invoices that match the PO and the contract rate the first time.
• Compliance: documents, certifications and statutory registrations current on the review date.
• Relationship or improvement: a short structured stakeholder rating, or progress on agreed improvement actions.

What does every KPI need before it goes on the scorecard?

Write a one-line definition card for each measure before the first review, and share it with the supplier. Most scorecard disputes are about definitions, not performance: whether a partial delivery counts as on time, whether a ticket's clock stops while waiting on the customer, whether a public holiday counts as a business day.

• Formula: numerator, denominator and what is excluded.
• Data source: the ERP, ticketing tool or goods-receipt record the number is taken from.
• Measurement period and the date the score is fixed.
• Target and threshold: what counts as green, amber and red.
• Owner on each side, and the escalation route when the score is red.

What is the difference between an SLA and a KPI?

A KPI is a measure you track to manage the relationship. An SLA, a service level agreement or service level, is a measure the supplier has contractually committed to, with a defined consequence if it is missed. Every SLA is a KPI, but most KPIs should not be SLAs.

The distinction matters because turning a KPI into an SLA changes how it behaves. Once money depends on a number, the supplier will contest the definition, the data source and every exclusion, so the measure has to be precise enough to survive that. Keep SLAs to the two to four outcomes that would actually hurt the business if missed, such as system availability, critical-incident resolution time, or on-time delivery to a production line. Leave the rest as KPIs that are reviewed and discussed, not priced.

A useful drafting pattern is to put SLAs, their measurement rules and their consequences in a schedule to the contract, and keep the wider KPI set in an operating document that can be revised at a quarterly review without amending the contract.

Service credits or penalties: what does Indian law allow?

In India the starting point is section 74 of the Indian Contract Act, 1872. Where a contract names a sum to be paid on breach, or contains any other stipulation by way of penalty, the party complaining of the breach is entitled, "whether or not actual damage or loss is proved to have been caused thereby", to reasonable compensation "not exceeding the amount so named" or the penalty stipulated. Unlike English law, the section does not draw a hard line between a valid liquidated damages clause and an unenforceable penalty. Either way, the named amount is a ceiling, and the court awards reasonable compensation up to it.

The Supreme Court restated how this works in Kailash Nath Associates v. Delhi Development Authority, decided on 9 January 2015 (Civil Appeal No. 193 of 2015, reported at (2015) 4 SCC 136). Building on earlier decisions including the five-judge bench in Fateh Chand v. Balkishan Das, the Court said that a named liquidated amount can be awarded as reasonable compensation only if it is a genuine pre-estimate of damages fixed by both parties and found to be such by the court. Otherwise only reasonable compensation is awarded, with the named amount as the upper limit. The Court also held that damage or loss caused is a sine qua non for the section to apply, and that the words "whether or not actual damage or loss is proved" do not dispense with proof where loss can be proved. Only where loss is difficult or impossible to prove can a liquidated amount that is a genuine pre-estimate be awarded without it.

Reasonable compensation is fixed on the principles in section 73, which allows compensation for loss that naturally arose in the usual course of things from the breach, or that the parties knew when contracting was likely to result, and excludes remote and indirect loss.

How to draft service credits that hold up

If a service credit is an amount the supplier pays, or forgoes, because it failed a contractual commitment, it is likely to be read as a sum named for breach and tested under section 74. Some contracts frame credits as a price adjustment for a lower level of service rather than as compensation. Whether that framing changes the analysis depends on the drafting, so take advice on your own clause rather than relying on the label.

The practical lessons from section 74 and Kailash Nath point the same way. Size the credit by reference to what a miss actually costs you, and record that reasoning when the contract is signed. A credit of a few per cent of the monthly service charge, scaled to the depth of the miss, with an overall cap, is easier to defend than a flat amount that bears no relation to loss.

• Attach credits only to SLAs, never to the wider KPI set.
• Write down, at signing, how each credit relates to the expected cost of a miss: lost production hours, extra freight, staff time, or customer credits you pay out.
• Scale credits to severity and duration rather than a single flat sum.
• Cap total credits per period, and say whether credits are the sole remedy for a miss or sit alongside termination and damages for serious or repeated failures.
• Say how credits are applied, for example as a deduction from the next invoice, and that the supplier's monthly report is the basis unless disputed within a set number of days.
• Add a termination right for persistent failure, such as the same SLA missed in a set number of months within a rolling period, because a credit alone rarely fixes a chronic problem.

How should scorecards drive renewal decisions?

A scorecard that never touches the contract is a report, not a control. Link the two explicitly. Put the renewal or notice date on the supplier record, and schedule the final scorecard review far enough before it that a poor score can still lead to non-renewal, renegotiation or a competitive tender. Many renewals are lost to the calendar, not the score: the notice window closes before anyone looks at the numbers.

Agree the rules in advance with your own business owners. One workable pattern is three bands. Suppliers consistently green are eligible to renew on similar terms. Suppliers with amber trends renew only with an improvement plan and tighter SLAs. Suppliers red on an SLA for consecutive periods go to tender or exit. Publish the bands to suppliers at the start of the term, so the renewal conversation is about data both sides have seen, not a surprise.

Keep the history. Service credits paid, disputes raised, and improvement actions agreed and closed are better evidence at renewal than an annual satisfaction rating, and they protect the decision if the supplier challenges it.

Running the review cadence

Match the cadence to materiality. Critical suppliers warrant a monthly operational review of SLAs and a quarterly business review of the full scorecard. Lower-tier suppliers can be scored quarterly or half-yearly. Send the scorecard to the supplier before the meeting, with the data behind it, so the meeting is spent on causes and actions rather than arguing about the numbers.

Record actions with owners and due dates, and carry open actions into the next review. A scorecard that shows the same red KPI three quarters running with no recorded action is a sign the process has stopped working, whatever the supplier's score.

Where software helps, and where it does not

Spreadsheets fail at scorecards for the same reason they fail at onboarding: the data is copied in by hand, the definitions drift, and the history is lost when the file owner moves on. Vendor management software helps by holding SLAs, scorecards and issue logs against each supplier and linking the supplier to its contracts and renewal dates. IntelloVendor tracks performance this way across the supplier base, with each vendor connected to its contracts, renewals and obligations.

Software cannot choose your KPIs or decide what a miss is worth. Agree those with the business owner and the supplier first, then configure the tool to match.