B2B Last Mile Delivery: The Order is Not Final Until the Shopkeeper Agrees

In consumer delivery the commercial transaction closes at checkout and the last mile executes a settled agreement. In business-to-business last mile through general trade, it does not. The outlet owner can look at what arrived, take five cases instead of ten, refuse a slow-moving line, question a price, or ask for the balance next week. The order is a proposal until the shopkeeper agrees to it.

That single difference makes B2B last mile the only delivery model where quantity is still commercially open when the vehicle arrives. It is not an exception workflow and it is not a service failure. It is the normal closing step of a trade relationship, and almost no delivery system is built to hold it, which is why one two-minute conversation at a shop counter generates work in three different departments for the rest of the week.

Why the transaction stays open in general trade

The channel structure explains the behaviour. McKinsey’s work on emerging-market grocery records small proprietors accounting for roughly 98% of the market in India, 97% in Nigeria and 85% in Indonesia, served through multilayered distribution networks with high logistics costs. Those proprietors are owner-operators buying on their own account with their own working capital, and they hold the same right any buyer holds: to inspect on arrival and take less than was offered.

Bain’s work on route to market in fragmented trade treats that route as a durable source of competitive advantage, which is the right framing and also the reason the behaviour persists. A distributor that refuses all renegotiation loses the outlet to a competitor who will accommodate it. The flexibility is a commercial feature, not a process defect.

The volume flowing through this channel is large and growing. The Google, Temasek and Bain e-Conomy SEA report put Southeast Asia’s digital economy on track to pass 305 billion dollars in gross merchandise value in 2025, growing around 15% year on year, and traditional trade still carries the majority of physical consumer goods across most of those markets.

The cost visibility is poor for a reason that is now familiar. McKinsey surveyed 35 senior leaders at 28 North American consumer packaged goods companies and found only 17% believe they recover more than 75% of the true cost to serve. Door-level renegotiation is precisely the kind of cost that never lands in a cost-to-serve model, because it is absorbed in three places by three teams who each see a fragment.

And it works against the thing the brand is buying. Research summarised by ECR Retail Loss puts the global out-of-stock rate at 8.3%, with store stocking and store forecasting the largest causes. A shopkeeper who takes half the order is choosing a shorter cover period, which raises the probability of the very stockout the brand’s coverage strategy exists to prevent.

Also Read: FMCG Logistics Trends 2026: Top Challenges and Orchestration Fixes

How to bound and capture the door-level decision

1. Count the decisions before designing anything

Renegotiation is usually discussed anecdotally. Sized, it stops being anecdotal.

Network	Stops per day	Renegotiation rate	Decisions per day	Driver-hours per day
200 distributors, 40 outlets	8,000	10%	800	40
1,000 distributors, 40 outlets	40,000	15%	6,000	300
1,000 distributors, 60 outlets	60,000	20%	12,000	600

Six thousand commercial decisions a day, each taken at a shop counter by someone employed to drive a vehicle, with no pricing authority and no visibility of the outlet’s credit position. The scale is the argument for bounding it.

2. Map the three records that break together

This is the mechanism that makes renegotiation expensive out of proportion to its size. One conversation invalidates three separate records, held in three systems, owned by three functions.

Record	What breaks	Who owns it
Delivery record in the TMS	Quantity delivered does not equal quantity dispatched	Operations
Financial record in invoicing	Invoice does not equal collection	Finance
Inventory record at the distributor	Stock returned does not equal stock shipped	Supply chain

At 6,000 renegotiations a day that is 18,000 record adjustments. The automation level decides what that costs.

Share auto-reconciled	Manual adjustments per day	Hours at 2 minutes each
0%	18,000	600
50%	9,000	300
90%	1,800	60

The reason this work is invisible is that no single team sees more than a third of it. Operations sees delivery variances, finance sees collection variances, supply chain sees unexplained returns, and nobody sees that they are the same 6,000 events.

3. Accept that the reconciliation control cannot work as designed

Most cash-market operations run a variance check: compare collection to manifest and investigate the gap. The arithmetic makes that useless.

Renegotiation rate	Average reduction	Collection as share of manifest
10%	25%	97.5%
15%	30%	95.5%
20%	40%	92.0%

Every route comes back short, and the shortfall is legitimate. So a control designed to detect cash shrinkage is looking at a signal dominated by lawful renegotiation, and legitimate reduction and theft become indistinguishable. The only way to separate them is to capture the renegotiation as an authorised event at the moment it happens, which converts the expected shortfall into a documented one and leaves any remaining variance meaningful.

The related point is that exact reconciliation is not a realistic baseline. A 30-stop route at a 10% renegotiation rate reconciles with zero changes about 4.2% of the time, and at 20% it is effectively never. Designing the process around the clean case means designing for the case that does not occur.

Also Read: Failed Deliveries Do Not Have to Mean Lost Customers

4. Separate commercial authority from dispatch discretion

These are different ladders and they are routinely conflated. A driver who can sensibly re-sequence two stops is not thereby qualified to vary a price. The test is reversibility and whose money is at risk.

Decision at the door	Reversible	Should the driver hold it
Accept a reduction within a set share of order value	Yes	Yes
Accept a reduction beyond that share	Yes	Escalate
Refuse the reduction and return the full order	Yes	Yes
Offer a price concession	No	Never
Vary or extend credit terms	No	Never
Accept a return of previously delivered stock	Partly	Escalate

Two of those six should never sit with a driver, and in practice both often do, because the shopkeeper asks and the alternative is an argument at the counter and a lost sale. A bounded authority with a fast escalation path is what removes the pressure to improvise, and the boundary has to be a system rule rather than a policy document.

Also Read: How to Optimize Beat in Sales for FMCG Distribution Success

5. Price the returned fragments, and remember they carry a date

Partial acceptance does not return orders, it returns fragments, and in FMCG those fragments are dated.

Renegotiation rate	Average reduction	Returned share of dispatched volume
10%	25%	2.5%
15%	30%	4.5%
20%	40%	8.0%

Between two and eight percent of dispatched volume comes back as part-picked lines that have to be re-inspected, re-entered to inventory and re-sold inside their remaining shelf life, or written off. That is a different problem from a whole-order failed delivery, which returns intact and can be re-attempted.

6. Plan the return capacity, even though it is small

Returned fragments accumulate across the route while the vehicle is emptying, so they rarely create a capacity crisis. The exception is where returnable crates or cases come back with them, since those occupy volume that the outbound plan did not reserve. It is a small volume on any single route and a systematic one across a fleet, and it belongs in the plan rather than in the driver’s judgement about whether there is room.

7. Give the shopkeeper the decision before the vehicle arrives

The cheapest renegotiation is the one that happens before dispatch. If the outlet can confirm or amend the order the evening before, the amendment costs a data change rather than a truck visit, three record breaks and a returned fragment. Pre-delivery confirmation is usually built as a customer service feature and is in fact the highest-leverage intervention available here, for the same reason pre-arrival notification matters in consumer delivery: it moves the decision to where it is cheap.

Where B2B last mile differs from consumer last mile

Dimension	Consumer last mile	B2B general trade last mile
When the transaction closes	At checkout	At the counter, on arrival
Who can change the quantity	Nobody, after dispatch	The buyer, on arrival
Nature of a partial delivery	Service failure	Commercial negotiation
What the driver decides	Where to leave it	Quantity, sometimes price and credit
Payment	Prepaid, usually	Often cash, amount set at the door
Records affected by a change	One, the delivery record	Three: delivery, invoice, inventory
Right response	Recover the delivery	Bound the authority and capture the event

The middle column is what every delivery platform is built for. The right-hand column is where most of the world’s consumer goods volume actually moves.

Also Read: Enterprise FMCG Logistics: Market Overview, Trends and Strategies

Five criteria for evaluating B2B last mile capability

1. Can a partial acceptance be captured as a structured event at the door? Not as a note or a photograph. Quantity accepted per line, reason code, and the resulting collection amount, recorded before the driver leaves.

2. Does that single event update delivery, invoice and inventory records? If it writes to one system and leaves the other two to a reconciliation process, the cost stays.

3. Are authority limits enforced by the system rather than by policy? The driver should be unable to accept a reduction beyond the permitted share without an escalation, rather than instructed not to.

4. Is the escalation fast enough to use at a counter? An approval path that takes ten minutes will be bypassed. Ask what the round trip is and what happens when there is no signal.

5. Can pre-delivery confirmation be offered per outlet? The ability to let a specific outlet confirm or amend before dispatch, targeted at the outlets that renegotiate most, is the intervention with the best return.

What this looks like in enterprise deployments

A global FMCG manufacturer distributing across ten Asian countries through more than 1,000 distributors and 5,000 riders reached 3X ROI while saving more than 12,000 trips a month, optimising over 4 billion dollars of orders and reaching 1.8 million retail outlets. At 1.8 million outlets the door-level decision is not a marginal case, it is a daily volume in the tens of thousands, and the value of holding it in a system rather than in conversation scales with that count.

Settlement is where the same discipline already exists, pointed the other way. A leading paint manufacturer processing more than 1,500 carrier invoices a month across 160 depots used Locus Settlement, Carrier and Orchestrator agents to automate freight reconciliation, catching 5% to 6% variance above contracted rates and compressing payment cycles from 30 to 45 days to 7 to 10. That is money flowing outward to a carrier, matched across contract, shipment, proof of delivery and invoice. Door-level renegotiation is the same matching problem with the money flowing inward from an outlet, and it is the direction that usually has no equivalent machinery.

Four mistakes in B2B last mile delivery

Treating partial acceptance as a delivery failure. It is a commercial negotiation. Coded as a failure it distorts service metrics and hides the real cost.

Running a cash variance control in a renegotiation market. Every route comes back short for legitimate reasons, so the control cannot distinguish lawful reduction from shrinkage until the reduction is captured as an authorised event.

Giving drivers price and credit discretion by omission. Nobody grants it. The shopkeeper asks, the alternative is a lost sale, and the driver decides. A system limit removes the pressure.

Building pre-delivery confirmation as a service feature. It is the cheapest place to absorb a change, and pointing it at the outlets that renegotiate most is the highest-return action available.

Also Read: Freight Audit and Settlement Software Buyer’s Guide 2026

How Locus handles the decision taken at the counter

Locus, the world’s first Decision-Intelligent, Agentic TMS, is built for operations where the execution record and the commercial record have to agree, which is the requirement this problem creates. The Driver Companion App carries the delivery workflow including electronic proof of delivery with signature, scan and chain of custody, plus partial cancellation and return-to-origin handling, which is the mechanism for capturing an amended quantity as a structured event at the point it is agreed rather than reconstructing it later from a paper note.

The settlement machinery already exists for the outbound direction. The Settlement Agent performs multi-way matching across contract terms, shipment data, proof of delivery and invoice with continuous audit, and posts actuals to ERP accounts. Applying the same matching to an inbound collection, where the agreed quantity at the door is the authoritative input, is what turns three broken records into one reconciled event. Order management holds the rule engine and enriched manifests that carry per-outlet terms into execution, and the route planning system accounts for the returning volume so the remaining sequence is planned rather than improvised.

Authority is the governance question, and it is set rather than assumed. Autonomy Levels run per agent and per domain, L1 where a human approves, L2 where the agent acts within guardrails and L3 where it acts autonomously, which is how a reduction inside a permitted share can be accepted at the counter while a price or credit variation routes to someone who owns that decision. Explainability and Traceability record the trigger, context, reasoning, action and outcome, so a disputed collection three weeks later has a retrievable answer rather than a recollection.

One boundary is worth stating. Locus is not the system of record for pricing, trade terms or credit limits. Those sit in the ERP or the distributor management system, and Locus consumes them as constraints rather than owning them. A deployment that wants door-level authority enforced needs that integration named as a requirement, because an authority limit is only as good as the credit and pricing data behind it.

Locus supports more than 360 enterprise customers across 30-plus countries, with over 1.5 billion deliveries optimised, more than 320 million dollars in documented client logistics savings and 99.99% uptime. It has been recognised by Gartner for seven consecutive years, featured in the 2026 Hype Cycle for Supply Chain Execution and Logistics Technologies, named a Leader in TMS by QKS Group (SPARK Matrix), and ranked #1 in Route Planning on G2’s 2026 Best Software Awards.

In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Locus continues to operate independently.

So what changes when the order is not final until the shopkeeper agrees? The delivery stops being an execution task and becomes the closing step of a commercial negotiation, and the systems have to hold both. At 1,000 distributors serving 40 outlets each with a 15% renegotiation rate, that is roughly 6,000 door-level commercial decisions a day, each breaking a delivery record, an invoice and an inventory position held by three different teams, for 18,000 adjustments that no one team can see. In cash markets it also renders the variance control useless, because legitimate reduction and shrinkage look the same until the reduction is captured as an authorised event. The workable design is narrow: bound the driver’s commercial authority in the system rather than in policy, capture the amended quantity once and write it to all three records, plan for the returned fragments and their shelf-life clock, and push the decision to pre-delivery confirmation wherever the outlet will use it. Locus captures the amended quantity through its driver workflow and ePOD, applies settlement-grade matching to the collection, holds per-outlet terms in the order rules, and sets the authority boundary through per-domain autonomy levels. Request a Locus assessment to size your own door-level decision volume.

FAQs

Why do B2B delivery orders change at the point of delivery? Because in general trade the buyer is an owner-operator purchasing on their own account and inspecting on arrival, so the order is a proposal until accepted. Reducing quantity, refusing a slow-moving line or asking for the balance later are ordinary commercial acts, and a distributor that refuses all of them loses the outlet to one that will accommodate them.

Is a partial acceptance the same as a failed delivery? No, and coding it as one causes problems. A failed delivery is a service breakdown to be recovered. A partial acceptance is a completed negotiation with a different outcome than planned. Recording it as failure distorts service metrics and hides the reconciliation cost it actually creates.

Why is door-level renegotiation so expensive relative to its size? Because one conversation breaks three records at once: quantity delivered against dispatched, invoice against collection, and stock returned against shipped. Those sit in three systems owned by operations, finance and supply chain, so at 6,000 renegotiations a day the network generates about 18,000 record adjustments that no single team sees in full.

How should cash reconciliation work in a renegotiation market? Not as a simple variance check. At a 15% renegotiation rate with a 30% average reduction, collection runs about 95.5% of manifest as a matter of course, so lawful reduction and cash shrinkage are indistinguishable in the gap. Capturing each reduction as an authorised event at the door converts the expected shortfall into a documented one and makes any residual variance meaningful.

What commercial authority should a delivery driver have? Enough to close an ordinary reduction and no more. Accepting a reduction within a set share of order value and refusing a reduction outright are both reversible and reasonable to delegate. Offering a price concession or varying credit terms are not reversible and should never sit with the driver, which means the limit has to be enforced by the system rather than written in a policy.

What reduces door-level renegotiation most cheaply? Moving the decision earlier. If an outlet can confirm or amend the order before dispatch, the change costs a data update rather than a vehicle visit, three record breaks and a returned fragment with a shelf-life clock. Targeting pre-delivery confirmation at the outlets that renegotiate most often is the highest-return intervention available.