Skip to content

Vendor management best practices for retail

Vendor management best practices begin by ending re-discovery

The characteristic failure of retail vendor management is not conflict. It is amnesia: Harbor Provisions sat under 92% on-time in-full for eleven of thirteen weeks at our 62-store operator before anyone raised it. A supplier's service degrades in February, somebody notices in April, the quarterly review in May treats it as news, the supplier promises to fix it, and the next review in August treats the identical problem as news again. Nothing in that loop is adversarial. It is just that the only time anyone looks at the number is the meeting, so every meeting starts from zero.

At our 62-store operator this was measurable. Harbor Provisions had been under 92% on-time in-full for eleven of thirteen weeks before anyone raised it, and when it was finally raised, the conversation was about whether the number was right rather than about what to do. Eleven weeks of degraded service produced zero corrective action and one methodology argument. Vendor management best practices are mostly the practices that prevent that specific loop.

Cadence: monthly numbers, quarterly decisions

The single highest-leverage change is separating when the number is seen from when decisions get made.

CadenceWhoWhat happensWhat does not
WeeklyBuyer, internalException review: any supplier crossing into red on serviceNo supplier contact
MonthlyBuyer + supplierScorecard sent, no meeting unless a line is redNo negotiation, no new items
QuarterlyBuyer + supplier + category leadFull review, targets set, commercial decisionsNothing that was not already visible monthly
AnnualCategory lead + financeRange review, terms, continue or replaceSurprises

The monthly send with no meeting attached is the part people skip and the part that does the work. It costs nothing, and it removes the supplier's ability to be surprised. When Harbor received a monthly card showing 84%, 81%, then 82%, the quarterly stopped being a debate about whether there was a problem.

The rule underneath the table: nothing should be raised at a quarterly that was not visible monthly. A quarterly review that contains new information is a reporting failure, not a supplier failure.

The escalation ladder

Escalation should be mechanical, published, and known to the supplier in advance. Discretionary escalation is slow, inconsistent between buyers, and reads as personal when it finally happens.

TriggerStepOwnerTimebox
One month below 92% OTIFFlag on the card, written ack requestedBuyer5 days
Two consecutive months below 92%Corrective action plan, named owners both sidesBuyer + supplier account lead15 days
Plan missed, or a third month below 92%Commercial review: terms, promotional support, or range at riskCategory lead30 days
Fourth month, or plan abandonedRange reduction or replacement beginsCategory lead + financeNext range review

Two design choices worth defending.

The first step asks for a written acknowledgement, not a plan. Demanding a corrective plan after a single soft month generates paperwork for what may be a one-off. Asking for acknowledgement establishes that the number was received, which is what makes step two legitimate.

The ladder has a terminal rung. A ladder that never ends in consequence teaches suppliers that the ladder is theatre. The fourth rung has to be real, and it has to be real occasionally in practice, or the first three rungs stop working across your whole vendor base.

Corrective action plans that are not wishes

Most supplier corrective action plans fail because they name an outcome rather than a mechanism. "Improve fill rate to 95%" is an outcome. It commits the supplier to nothing they can actually do on Monday.

A usable plan names the mechanism, the owner, and the checkpoint. Harbor's real problem was allocation: 61% of their missed cases were short at confirmation, meaning our order was being cut before it ever shipped, because their allocation rule ranked accounts by historical volume when supply was tight. The plan that worked did not mention fill rate:

MechanismOwnerCheckpoint
Move our account to committed-volume allocation tierHarbor supply leadWeek 2
We commit to 4-week rolling forecast, 6 core itemsOur buyerWeek 2
Order lead time extended 48 hours on the 6 core itemsBothWeek 3
Weekly short report by item, sent FridayHarborWeekly

Note that two of the four commitments are ours. A plan where the supplier carries every obligation usually means the diagnosis was lazy, because most service problems have a demand-signal component on the retailer side. Extending lead time by 48 hours cost us flexibility and bought a measurable share of the recovery.

Joint business planning, for the five percent that earn it

Strategic suppliers get one thing the rest do not: a forward plan rather than a backward review. A joint business plan is an annual agreement covering volume commitments, promotional calendar, new-item pipeline, and the service levels each side is committing to, with a mid-year checkpoint.

The reason most JBPs are worthless is that they are written as aspiration and never referenced again. Three properties separate the ones that work:

Both sides commit to something measurable. A plan where the supplier commits to growth and the retailer commits to nothing is a forecast, not a plan. Our side of the Harbor recovery included a four-week rolling forecast and 48 extra hours of order lead time, both of which were real costs and both of which appeared in the document.

The commitments map to scorecard lines. If the JBP commits to 96% OTIF and the monthly card measures OTIF the same way, the plan is self-monitoring. If the plan invents its own metrics, it becomes a separate reporting exercise that nobody maintains past March.

There is a mid-year checkpoint with the authority to change the plan. Annual plans reviewed annually are reviewed once, at the end, when nothing can be done.

The relationship-quality trap

The most common structural mistake in vendor management is letting relationship quality substitute for performance. It happens quietly. A supplier who is responsive, pleasant, and quick to answer email accumulates goodwill that functions as a buffer against their numbers, and a supplier who is difficult accumulates the opposite. Neither has anything to do with whether the shelf is full.

The countermeasure is procedural rather than attitudinal: the scorecard is reviewed before the relationship conversation, and the escalation triggers are mechanical. You cannot legislate away the bias, but you can make sure it acts after the number is on the table rather than before.

There is a real version of relationship value, and it is worth separating from the false one. A supplier who tells you early that they will be short is genuinely more valuable than one who confirms and cuts, even at identical OTIF, because early warning is actionable and a confirmation that evaporates is not. That belongs on the card as a measurable line (notice given before the promise date), not as a general impression of pleasantness.

Not every supplier gets the same management

Running all 140 suppliers on an identical cadence is how vendor management consumes a buyer's entire week and still misses the important failures. Segment the base and spend the attention where it changes an outcome.

SegmentShare of vendorsShare of $CadenceWhat you manage for
Strategic~5%~45%Monthly card, quarterly review, annual JBPGrowth and joint planning
Core~20%~40%Monthly card, quarterly reviewService consistency
Transactional~60%~14%Exception onlyNothing until they break
Specialty / local~15%~1%AnnualRange distinctiveness
% of vendors% of spendStrategic5%45%Core20%40%Transactional60%14%Specialty / local15%1%
Share of vendor count (light) against share of spend (dark). Strategic and core are a quarter of the vendors and 85% of the dollars, which is why an even review cadence buys a fifth of the coverage where it matters.

Two implications fall out of this table. Strategic and core together are a quarter of the vendor count and 85% of the dollars, which is the entire case for segmentation: the same review effort spread evenly buys you a fifth of the coverage where it matters. And transactional suppliers should be managed purely by exception, meaning the escalation ladder still applies but no scheduled review exists at all.

The trap is segmenting by dollars alone. A small supplier can be strategically important because they hold a range position nobody else fills, and a large one can be genuinely transactional if the product is a commodity with three substitutes. Segment by dollars, then adjust for substitutability, and write down why any vendor sits above their dollar rank.

When to replace, and the honest cost of switching

Replacement is expensive and chains under-count the cost, which is why bad suppliers persist. The switching cost includes the range gap while the new supplier ramps, the promotional calendar rebuild, the store-level reset labor, and roughly two quarters before the new relationship produces trustworthy performance data.

Set the bar accordingly: replace when the ladder has run its course and the mechanism is still unfixed, not when a quarter is bad. And when you do replace, authorize the new supplier narrow and deep so the read is clean, exactly as in supplier selection.

One measurement discipline that pays for itself here. Hold the category view across the switch. The question that matters is whether the category improved, not whether the new supplier hits their numbers, and those can differ: a new supplier can run 97% OTIF on a range that sells worse than what it replaced.

Doing this in Scout

Everything above depends on the number being visible between meetings, which is exactly what does not happen when the scorecard is a spreadsheet somebody rebuilds each quarter. The eleven weeks Harbor spent under target were not invisible because the data was missing. They were invisible because nobody was going to rebuild the join until the quarterly forced it.

Scout keeps the supplier card standing. The monthly send is a saved view rather than a rebuild, the weekly exception check is a filter on it, and the escalation triggers read off the same definitions the quarterly uses, so a supplier crossing 92% surfaces in week two rather than month four. The late/short decomposition that turned Harbor's vague service problem into a nameable allocation problem is a slice of that view, not a separate analysis.

Scout is the measurement and review layer. It does not hold contracts, manage supplier onboarding documents, or issue purchase orders.

What good looks like after a year

The best test of whether vendor management best practices have taken hold is not the scorecard average. It is what happens the next time a supplier degrades.

In a chain where this works, a supplier crossing 92% is flagged in week two, acknowledged in week three, and either recovering or on a corrective plan by week six. Nobody convenes a special meeting, nobody argues about the calculation, and the quarterly review records what already happened rather than discovering it. That is the entire deliverable: not better suppliers, but a shorter distance between a problem starting and somebody doing something about it. Harbor's eleven weeks became two.

Summary

  • Separate cadence from decisions: send the scorecard monthly with no meeting attached, and let quarterly reviews contain nothing that was not visible monthly.
  • Publish a mechanical escalation ladder with a terminal rung, and make the first step an acknowledgement rather than a corrective plan.
  • Write corrective plans around mechanisms, not outcomes, and expect roughly half the commitments to be yours.

Further reading: supplier performance metrics defines the lines the ladder triggers on, and building a retail supplier scorecard covers the weighting.

Want this as a Google Sheet?

Drop your email and we'll send the worked example.

Book a demo with your data