Supplier Performance Management · Bangladesh Buyer Desk

A scorecard with no consequence attached is only a report.

Most buyers already have supplier numbers somewhere. What they usually lack is a person in Dhaka who collects those numbers from a source the factory does not control, puts them in front of the factory on a fixed date, and is still there the following week to see whether anything actually changed.

SUPPLIER SCORECARD
Q3 · 1,240 lines
OTIF
91%
Right first time
86%
Corrective action closure
74%
Responsiveness
95%
Illustrative card. The metrics, targets, weights and the exact measurement point are agreed with you before a single period is scored.
01 · THE DESK
Coordination
Local presence, supplier scorecards, order tracking, an issue register and one consolidated report, run by a named account manager.
02 · THE CLAIM
Remedies
Evidence assembled and discussions supported where rework, replacement, credits, shortages or other agreed remedies are in question.
03 · THE RECORD
Documentation
Communications and settlement commitments written down as they happen, not reconstructed months later from memory.
04 · THE FIX
Root cause
Causes of a failed audit or inspection diagnosed, and the permanent process fix verified before you pay for a re-inspection.
01 · The Scorecard

Five Numbers, Measured The Same Way Every Month

Four or five metrics is enough. A card with fifteen on it gets read once, argued about, and then quietly ignored — which is the same outcome as having no card at all.

OTIF — on time, in full
(on-time and in-full lines ÷ lines due) × 100
The headline reliability measure, because on-time delivery alone hides a short shipment. The detail that decides whether the number means anything is the measurement point: ex-factory date or arrival date, and what quantity tolerance counts as in full. A line that misses either condition is a miss. We fix both definitions in writing before the first period, so the figure cannot be renegotiated after a bad month.
EXAMPLE RESULT91% vs 95% target
Published benchmarks put world-class performance in the 95–98% band, with around 98% treated as the modern standard.
Right first time
(lots accepted at first pass ÷ total lots) × 100
The share of shipments that clear inspection without a re-offer. This is the metric factories feel most directly, because a failed first attempt costs them the re-inspection fee and the delay. It is also the one most worth tracking by style rather than only in total — a factory can be excellent on basics and consistently poor on anything with a lining.
EXAMPLE RESULT86%
Tracked against AQL first-attempt outcomes so the figure traces to specific inspection reports, not to a general impression.
Defect rate
(defective units ÷ units inspected) × 1,000,000
Expressed in parts per million rather than as a percentage, because at volume a percentage loses resolution — 0.4% and 0.9% look similar on a page and are more than twice apart in practice. Recorded by defect type as well as in total, since the total tells you there is a problem and the breakdown tells you where it starts.
EXAMPLE RESULT6,200 PPM
Bar shown against an illustrative 10,000 PPM ceiling. Your acceptable level depends on the product and what your buyer has written.
Corrective action closure
(findings closed and verified ÷ findings raised) × 100
Not findings the factory says are closed — findings someone went back and confirmed. This is usually the weakest number on a new supplier’s card, and the most predictive one, because a factory that does not close findings will present you with the same failure again next season under a different purchase order.
EXAMPLE RESULT74%
Counted only where closure evidence exists and has been checked on site. An unverified closure is left open on the card.
Responsiveness
median hours to a substantive reply
Measured as a median rather than an average, so one forgotten email over a holiday does not distort the month. A substantive reply means one that answers the question — an acknowledgement that the message was received does not stop the clock, because it does not move the issue.
EXAMPLE RESULT36 hours
Bar shown against a 48-hour target. Slow replies rarely cause a problem on their own; they reliably make an existing one later and larger.
The part that makes a card work is not the metrics. It is the three things around them: a fixed review cadence, weights agreed in advance, and a stated consequence when a threshold is breached — a corrective action request, a hold on new orders, a shift of volume. Without the third one, a scorecard is an email attachment. We will build the card either way, but we will say plainly that a card with nothing attached to it has never changed a factory’s behaviour.
02 · The Desk

Two Documents That Run The Week

The scorecard is monthly. Between scorecards, the desk runs on two live records that anyone on your side can open without asking a factory for an update.

Order tracker
UPDATED FROM SOURCE, NOT FROM A FACTORY EMAIL
PER ORDER
Style, quantity, agreed ex-factory date, and the current stage — fabric in, cutting, sewing, finishing, packed.
MILESTONES
Fabric in-house, PP sample approved, first bulk cut, inspection booked, inspection result, ex-factory. Each with planned and actual.
SLIPPAGE
Days late against the plan, shown per milestone rather than only at the end, so a delay is visible while it is still two days rather than three weeks.
SOURCE
Our own site visits, inspection reports and gate records, with the factory’s figure shown alongside where the two disagree.
Where our count and the factory’s count differ, both appear. We do not quietly pick one — a disagreement is itself information, and the pattern of who turns out right over a season tells you a great deal.
Issue register
ONE ROW PER ISSUE, OPEN UNTIL VERIFIED
RAISED
Date, order, what the issue is, who raised it, and what it puts at risk — quantity, date, quality or cost.
OWNER
A named person at the factory, not a department. Issues assigned to a department are the ones that sit unmoved for a month.
AGREED ACTION
What was committed, by when, and in which message or meeting it was committed — referenced, so it can be produced later.
STATUS
Open, actioned, or closed and verified. Only the third state removes it from the register.
Both records feed one consolidated report on a fixed date each month: scorecard, order status, open issues, and anything that needs a decision from you rather than from us.
03 · Remedies

When Something Has To Be Put Right

Rework, replacement, a credit, a shortage made good. These conversations are won or lost on what was written down at the time, which is months before anyone realises there is going to be a conversation.

STAGE 01

Assemble the evidence

Before anything is raised, the file is built: what was ordered, what arrived, what was found, when, and by whom. A claim that arrives as an assertion gets negotiated. A claim that arrives with a file gets answered.

  • Purchase order and approved sample
  • Inspection reports with dates and sample sizes
  • Photographs identified by order and carton
  • Counts, packing list and goods-received figures
  • The messages where the issue was first raised
→
STAGE 02

Support the discussion

We sit in the conversation with the factory, in the same time zone and in the same language, and keep it on the file rather than on who said what. Our role is to hold the record steady, not to decide your commercial position for you.

  • Meeting held in person where it matters
  • The file presented, point by point
  • Factory response recorded as given
  • Options put to you with what each costs
  • Your decision carried back, unaltered
→
STAGE 03

Document the settlement

Whatever is agreed is written down the same day, in terms specific enough to act on: the remedy, the quantity, the date, and who does what. Vague settlements are the ones that reappear as a second dispute.

  • Remedy stated with quantity and value
  • Deadline and the party responsible
  • How completion will be verified
  • Confirmed back in writing by the factory
  • Filed against the order and the register
What we do not do is decide your claim or guarantee its outcome. We are not your lawyers and we do not hold your money. A remedy is agreed between you and the factory, under your contract; our part is to make sure the discussion happens on evidence rather than on recollection, and that whatever is agreed exists in writing afterwards. Where a matter is genuinely contractual or heading towards legal action, we will say so and recommend you take advice rather than keep negotiating through us.
04 · Root Cause

Why The Audit Failed, And Whether It Is Actually Fixed

A failed audit produces two things: a finding and a deadline. What it rarely produces on its own is a cause — and a correction applied to a symptom will hold exactly until the next audit.

01
Contain
Stop the immediate exposure first — quarantine affected stock, suspend the operation, isolate the batch. Containment is not a fix and is never recorded as one.
02
Describe
State the failure precisely: which clause or criterion, which line, which shift, which operation, and when it started. Most causes are found in the gap between a vague description and a precise one.
03
Trace the cause
Work back through the process until the answer stops being a person and starts being a system — a missing check, an unwritten instruction, a machine setting nobody owns.
04
Change the process
The corrective action has to alter how the work is done, not add a reminder to be careful. Re-training on its own is the most common action and the least durable one.
05
Verify on site
Go back after the change and check it is in place and being followed under normal pressure, not on the day it was installed. Only then is the finding closed.
This is also why we verify before you book a paid re-inspection. A re-inspection is charged whether it passes or fails, so a factory that has corrected the symptom and not the cause costs you the fee twice. Checking the fix on site first is cheaper than a second failed inspection, and it is the one place in this service where doing the work properly reduces what you spend rather than adding to it.
05 · Questions

Asked Before Signing A Retainer

Including the two about cost and about whose side the desk is on.

01
How is this different from hiring an agent?
A buying agent is usually paid a commission on the order value, which means the agent earns more when you buy more and earns nothing when an order is cancelled for quality. A retained desk is paid a fixed fee for the work, so recommending that you hold a shipment costs us nothing. That is the structural difference, and it is worth checking in any arrangement you are offered, ours included — ask how the person advising you is paid before you weigh the advice.
02
What does a desk cost, and when is it not worth it?
It is a monthly retainer scaled to the number of suppliers, order volume and how much site time is included, quoted after we see your supplier list. It is not worth it below roughly a handful of active suppliers or a few orders a year — at that scale per-order inspection plus your own follow-up is cheaper and does the same job. It earns its keep when you have enough suppliers that nobody on your side can hold the whole picture, or when one factory keeps producing the same problem and nobody has had time to find out why.
03
Where do the scorecard numbers come from?
From your own records and ours — purchase orders, goods-received figures, inspection reports, our site visits — rather than from the factory’s self-reported performance. This matters more than it sounds. A scorecard built from supplier-supplied data measures the supplier’s reporting, not the supplier. Where we have no independent source for a metric, the card says so instead of showing a figure we cannot stand behind.
04
Will the factory cooperate with a desk measuring them?
Usually yes, and more readily than buyers expect, because a published card with agreed definitions also protects a good factory from being blamed for things outside its control. Resistance, when it comes, is almost always about the definitions rather than the measuring — which is why they are agreed in writing first. A factory that refuses any form of measurement has told you something useful at no cost.
05
Can you represent us in a dispute?
We can assemble the evidence, attend the meeting, and document what is agreed. We cannot act as your legal representative, sign on your behalf, or hold funds. Where a matter moves beyond a commercial conversation into a contractual claim, the right step is a lawyer who practises in the relevant jurisdiction, and we will tell you when we think you have reached that point rather than carrying on because the retainer is running.
06
How long before any of this shows up in the numbers?
Realistically two to three scoring periods. The first card establishes a baseline and is usually uncomfortable reading. The second shows whether anything moved. The third is where a pattern becomes reliable enough to act on commercially. Anyone promising improvement in the first month is describing the reporting, not the factory — the underlying processes take a season to shift, and the honest version of this service says so at the start.
START HERE

Send your supplier list and one season of inspection results.

Those two are enough to build a first scorecard and to tell you whether a retained desk is worth the fee for your volume. If it is not, we will say so — a retainer that cannot pay for itself is a bad arrangement for both sides, and it ends badly in month four rather than quietly.