COD-Heavy Last Mile in Southeast Asia in 2026: Why the Cash Ceiling Caps Your Route Before the Clock Does

Cash on delivery means the rider collects payment at the door, which makes that rider a mobile cash holder for the length of the shift. Every operation running COD sets a ceiling on how much cash one rider may carry, driven by security exposure, insurance terms and how often cash can be banked. That ceiling is usually written as a policy and enforced by a supervisor, but it behaves as a routing constraint: it caps how many drops a route can hold, independently of vehicle capacity or shift length. In markets where COD still carries a large share of e-commerce value, it is frequently the binding constraint on route size, and almost no dispatch system models it as one. Locus, the world’s first Decision-Intelligent, Agentic TMS, solves cash exposure alongside time, capacity and geography inside the same plan, across more than 250 real-world operating constraints.

Why the Cash Ceiling Matters: The Business Case

COD is declining across Southeast Asia and is still large enough to set operating design. Worldpay’s Global Payments Report 2026 puts cash on delivery at 23% of e-commerce value in the Philippines and 16% of e-commerce spending in Vietnam, with Thailand’s share among the highest globally. A payment method carrying a fifth of transaction value is not a legacy edge case, and the operational obligations it creates fall entirely on the last mile.

The volume underneath those percentages keeps rising. The e-Conomy SEA report from Google, Temasek and Bain tracks a regional digital economy that continues to expand across its major markets, which means COD’s falling share is applied to a growing base. Absolute cash volumes in the field have not fallen as fast as the percentage suggests, and the ceiling is set in absolute currency, not in share.

Route density is where the cost lands. McKinsey’s work on delivery economics finds that raising drops per stop from one to five cuts labour and vehicle cost by more than 50%, and the same logic applies to drops per route: anything that caps route size raises unit cost directly. The World Bank’s Logistics Performance Index continues to score several Southeast Asian economies weakest on timeliness and tracking, the two dimensions most sensitive to routes being cut short and re-planned mid-shift.

Also Read: Real-Time Visibility Ends at Handover: Tracking the COD Cash Chain in Southeast Asia

How the Cash Ceiling Caps a COD Route

The figures below come from a model of a single rider’s shift. Order values are drawn from a right-skewed distribution with a mean of 12 currency units and a coefficient of variation of 1.1, which is typical of e-commerce baskets. The rider can physically serve 60 drops in a shift. Inputs are illustrative; substitute your own and the shape holds.

1 The float is a capacity constraint, not a policy

A cash ceiling is normally owned by finance or risk and communicated to operations as a rule. Dispatch systems treat it the same way, as a check applied after the route is built. But a constraint that determines how many stops fit on a route is a capacity constraint by definition, in the same class as vehicle volume and shift length. Modelling it anywhere other than inside the plan guarantees that plans are built and then broken.

2 The ceiling binds before the clock does

For any given average order value there is a ceiling below which cash, not time, decides route size.

Cash ceiling	Drops at the average order value	Binding constraint	Time capacity lost
300	25	Cash	58%
400	33	Cash	44%
500	42	Cash	31%
600	50	Cash	17%
720	60	Time	0%
900	75	Time	0%

Below 720 the rider goes home with hours of usable shift left. The loss does not appear in any utilisation report, because the route was planned short and then completed on time.

3 Planning from an average order value breaches half your routes

The most common way to set a COD stop count is to divide the ceiling by a typical order value. At a 500 ceiling and an average of 12, that gives 41 drops. The problem is arithmetic: the expected cash on a 41-drop route is almost exactly the ceiling, which places the route at the median of the cash distribution.

Cash ceiling	Stops if planned to the average	Odds that route breaches	Stops at 5% breach	Stops at 1% breach	Margin cost
400	33	48%	24	21	27%
500	41	46%	31	28	24%
600	50	50%	38	34	24%
720	60	50%	47	43	22%
900	75	50%	60	55	20%

Half of all routes crossing a compliance line is not a tolerable operating state, so operations discover this and pull the number down. Holding breach at 5% costs roughly a quarter of the planned stops. That margin is the real cost of the cash ceiling, and it is paid every day on every route.

It is also the part that is recoverable, because the margin exists only to cover uncertainty about order values. Those values are known at order capture. A system that plans against the actual cash on the actual manifest, rather than against a distribution, does not need most of that buffer.

There is a second recovery most operations miss entirely. A refused COD delivery collects no cash, so refusals fill the ceiling more slowly than the manifest suggests. Sizing routes on gross order value assumes every drop pays.

COD refusal rate	Expected cash on a 41-drop route	Odds of breach	Stops at 5% breach
0%	492	46%	31
5%	467	35%	32
10%	443	25%	34
15%	418	16%	36
20%	394	10%	38

At a 20% refusal rate the same ceiling supports 38 stops rather than 31, a 23% gain against a plan built on gross value. The failure mode that costs money on the parcel side quietly buys capacity on the cash side, and an operation that already measures its refusal rate is holding the number it needs to claim it.

Also Read: Route Optimisation for Southeast Asia: Why Address Quality Caps Your Routing Gains in 2026

4 Smarter cash assignment does not help, and clever packing makes it worse

The intuitive fix is to assign drops across routes so that each route’s cash total lands just under the ceiling, pairing high-value orders with low-value ones. Tested against 3,000 drops, it fails.

Cash ceiling	Routes, cut when cash trips	Routes, best-fit packing	Theoretical minimum	Gain from packing
500	72	82	70	Worse by 14%
600	60	73	58	Worse by 22%
720	54	68	52	Worse by 26%
900	51	62	50	Worse by 22%

Two results here, and both matter. First, simply cutting the route when cash trips lands within 3% to 4% of the theoretical floor, so there is almost no headroom to win. Second, first-fit-decreasing packing, the standard near-optimal heuristic for this class of problem, is substantially worse, because filling each route to the cash ceiling efficiently means filling it with fewer, larger orders and stranding the stop capacity.

This cuts against the usual pitch. If a vendor offers optimisation of cash allocation across routes, the honest answer is that the naive rule is already close to optimal and the sophisticated approach is likely to lose you routes. The cash constraint is real and expensive, and assignment is not where it is paid.

5 Remittance scheduling is the lever, and it saturates immediately

The ceiling applies to cash held at one time, not cash collected over a shift. A mid-route handover, a deposit, an agent collection or a digital settlement, resets it. Against an 18-minute handover cost, a 480-minute shift and 7 minutes per drop:

Handovers in the shift	Capacity from cash	Capacity from time	Drops actually served
0	47	69	47
1	94	66	60
2	141	63	60
3	188	61	60
4	235	58	58

One handover moves the rider from 47 drops to the full 60, a 28% gain. The second buys nothing, because cash has stopped binding and time has taken over. The third and fourth actively cost drops. This is the whole optimisation, and it is a single decision per route about whether and when to insert one stop.

That decision is not static. It depends on the cash profile of the specific manifest, where the route passes a collection point, and how the detour interacts with delivery windows. A fixed policy of “bank at midday” is wrong on any route whose high-value drops fall in the morning.

6 The constraint has to be solved with time and geography, not after them

The remittance decision is where cash, sequence and geography meet. Inserting the handover at the wrong point costs more than the ceiling does, because the detour is charged against the same shift the cash constraint is trying to protect. Solving cash first and routing second gives a feasible plan that is unnecessarily short. Routing first and checking cash second gives a plan that gets cut mid-shift. Both are the same error in opposite order, and both are what happens when the ceiling lives outside the planner.

Cash-Constrained vs Time-Constrained Routes: Key Differences

Dimension	Time-constrained route	Cash-constrained route
What caps the route	Shift length, drive time, service time	Value collected before banking
Visible in utilisation reporting	Yes, as hours used	No, route completes early and on time
Effect of adding riders	Proportional capacity gain	Proportional, and the ceiling still binds each one
Effect of better sequencing	Direct gain	Almost none
Effect of a mid-route reset	None available	Restores capacity to the time limit
Sensitivity to order mix	Low	High, a few large orders end the route
Where the loss is recorded	Overtime, missed windows	Nowhere

Also Read: Why Rules-Based TMS Logic Breaks at Retail Scale: Three Shifts and Five Symptoms in 2026

What to Look for in Dispatch Software for COD Markets

Cash exposure as a first-class constraint. The planner should accept a per-rider cash ceiling as an input alongside volume, weight and shift length, and produce plans that respect it rather than plans that are checked against it. Ask to see a route rejected at build time for cash, not flagged afterwards.

Planning against actual order values. Order values are known when the order is captured. A system that sizes routes from a distribution rather than from the manifest in front of it is paying a variance margin it does not need to pay.

Remittance as a planned stop. The handover should be inserted by the optimiser, positioned against the route’s actual cash accumulation and its geography, not fixed by policy at a set hour. Ask how the system decides whether a route needs one handover, or none.

Per-rider ceilings rather than a fleet default. Exposure limits legitimately vary by tenure, insurance class and fleet type, and captive riders, contracted fleets and gig riders rarely carry the same limit. A single fleet-wide number forces every rider down to the most conservative one.

Honest treatment of what optimisation cannot do. Assignment of cash across routes is close to a solved problem with a simple rule. A vendor claiming large gains there is either measuring against a straw man or has not run the comparison.

COD Route Capacity in Action: Real-World Results

A global FMCG manufacturer operating across 10 Asian countries ran distribution through more than 1,000 distributors and over 5,000 riders, a structure where cash handling, route sizing and settlement all sit with partners rather than with the brand. In the logistics automation deployment, the change was moving planning into a single decisioning layer across that distributor network, so route construction accounted for the real constraints each rider operated under instead of applying a uniform assumption. The programme optimised more than $4B in orders, reached over 1.8 million retail outlets, saved more than 12,000 trips per month and returned 3X ROI. Trips saved is the number that maps to this argument: those are routes that no longer needed to exist once route size was set against real constraints rather than conservative defaults.

A Fortune 50 enterprise running a 4,500-strong driver pool across captive and third-party fleets found the same pattern in a different form. In its centralised dispatch deployment, the operation was leaving capacity unused because plans were built to assumptions rather than to measured limits. Weekly execution moved from 75% to 92%, and the programme surfaced more than $14M in annualised unused capacity, $565K at a single site before scaling across 25. Capacity that no report showed as missing is exactly what a cash margin looks like when nobody models it.

Common COD Dispatch Mistakes to Avoid

Setting stop counts from an average order value. This places the route at the median of the cash distribution and breaches on roughly half of all routes. Size against the manifest, or against a quantile if the manifest is not yet known.

Treating the cash ceiling as a compliance check rather than a capacity input. A check applied after planning produces plans that are broken mid-shift. The ceiling belongs in the objective function.

Fixing the remittance time by policy. The right handover point depends on where the route’s value accumulates and where a collection point sits. A fixed hour is wrong on most manifests and inserts an 18-minute detour where it is not needed.

Buying optimisation for cash assignment. Cutting the route when cash trips is already within 4% of optimal, and sophisticated packing performs worse. The gains are in the variance margin and the remittance decision, not in allocation.

Also Read: How Route Optimization Reduces Cost and Improves Last-Mile Delivery in Southeast Asia

How Locus Handles the Cash Constraint

Locus, the world’s first Decision-Intelligent, Agentic TMS, treats cash exposure as one of the constraints the plan is solved against rather than a rule applied to a finished plan. Route construction in the route planning system works from the actual order values on the manifest, so the route is sized against the cash it will really carry instead of against a distribution, which removes most of the margin an average-based stop count forces an operation to hold. Where a handover is worth making, the Dispatch and Settlement agents position it against the route’s own cash accumulation and geography rather than at a fixed hour, and per-rider ceilings are respected individually across captive, contracted and gig fleets rather than flattened to a single conservative default.

Locus is recognized 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 in the 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.

In COD-heavy Southeast Asian markets the rider’s cash ceiling is a capacity constraint that frequently binds before shift length does, and the loss is invisible because the route completes early and on time. The recoverable cost is not the ceiling itself but the margin operations hold against uncertain order values, roughly a quarter of planned stops, plus the mid-route handover that is usually set by policy instead of by the plan. Assignment of cash across routes is not where the gains are, and modelling shows sophisticated packing performs worse than a simple rule. Locus solves cash exposure jointly with time, sequence and geography in one decision layer, which is what turns a constraint that is currently absorbed as lost capacity into route size that can actually be used. Request a Locus COD route capacity assessment to see where the ceiling binds in your own network.

Frequently Asked Questions

What is a cash ceiling in COD delivery?

A cash ceiling is the maximum amount of collected cash a single rider may hold at one time, set by security exposure, insurance terms and banking frequency. It is usually written as a risk policy, but because it caps how many paid drops fit on a route, it functions as a capacity constraint on route size.

Does cash on delivery still matter in Southeast Asia?

Yes, though its share is falling. Worldpay’s Global Payments Report 2026 puts cash on delivery at 23% of e-commerce value in the Philippines and 16% of e-commerce spending in Vietnam, with Thailand among the highest globally. Because ceilings are set in absolute currency against a growing volume base, the operational constraint has not eased as fast as the percentage has.

How do I size a COD route correctly?

Not by dividing the cash ceiling by an average order value, which places the route at the median of the cash distribution and breaches on roughly half of all routes. Size against the actual order values on the manifest where they are known, and against a 5% breach quantile where they are not, accepting that the quantile approach costs about a quarter of the stops.

Can better route assignment solve the cash constraint?

Largely no. In this model, cutting the route when cash trips landed within 4% of the theoretical minimum number of routes, and first-fit-decreasing packing needed 14% to 26% more routes because it fills each route to the cash limit with fewer, larger orders. The gains sit in the variance margin and in remittance timing instead.

How many mid-route cash handovers should a rider make?

Usually one, and often none. In this model a single handover lifted capacity from 47 to 60 drops, the full time limit, while a second added nothing and further handovers cost drops. The correct number depends on the specific manifest’s cash profile and on where a collection point sits relative to the route.

Why does the cost of the cash ceiling not show up in reporting?

Because a cash-capped route is planned short and then completed on time, so it appears as a compliant, successful route. Shift hours go unused rather than overrun, and no exception is raised. The loss is only visible when route size is compared against what the time constraint alone would have allowed.