--- title: "The Last Mile: The Cost That Disappears Into the Average" description: "The last mile is the one segment where economies of scale invert: every upstream step aggregates volume while the terminal step disperses it, making unit cost a function of density rather than load. The institutional cost is that a network-average figure conceals the true cost-to-serve distribution across customers or connections, and pricing is then built on that average." url: https://www.beirek.com/en/blog/last-mile-problem canonical: https://www.beirek.com/en/blog/last-mile-problem published: 2026-02-02 modified: 2026-02-02 category: "Operations & Supply Chain" category_url: https://www.beirek.com/en/blog/category/operations-supply-chain language: en-US reading_time_minutes: 7 publisher: BEIREK LLC publisher_url: https://www.beirek.com license: "© BEIREK LLC — citation with attribution and link permitted" keywords: ["last-mile problem","cost-to-serve","mechanical completion versus commercial operation","gross margin dispersion","delivery risk allocation"] topics: ["Operations and supply chain economics","Project completion and schedule risk","Cost measurement and management reporting architecture","Valuation and due diligence readiness"] alternate_language_url: https://www.beirek.com/tr/blog/last-mile-problem --- # The Last Mile: The Cost That Disappears Into the Average > **In short:** The last mile is the one segment where economies of scale invert: every upstream step aggregates volume while the terminal step disperses it, making unit cost a function of density rather than load. The institutional cost is that a network-average figure conceals the true cost-to-serve distribution across customers or connections, and pricing is then built on that average. *The final leg of a distribution network, and the final leg of a project schedule, obey an economics unlike everything upstream of them; scale runs in reverse at that point. A cost architecture that does not measure the distinction reports profitability at the network mean rather than at the customer, and prices accordingly.* --- In an operations review, distribution cost is almost invariably presented as a single network-wide average: cost per ton-kilometre, cost per shipment, or logistics expense as a percentage of revenue. Because the figure is comparable across quarters, it reads comfortably at the board table, and it is precisely that comfort which keeps it from being opened. Yet within the same network, the leg running from plant to central distribution centre and the leg running from regional depot to the door are governed by two structurally different economics, the first of which lowers unit cost with every incremental unit of volume while the second pushes unit cost upward with every incremental address. A single average, aggregating two curves that move against each other, renders neither of them visible. In capital projects the same pattern surfaces on the schedule rather than in the cost report. Progress reporting confirms that the overwhelming majority of physical scope has been executed, while the date on which revenue begins slips one quarter at a time; the outstanding items list does not shorten but lengthens with each site visit. Each remaining item is individually trivial — an approval letter, a commissioning test, a meter connection, an acceptance protocol — and yet none can be substituted for another, and most depend on the calendar of a party outside the direct control of the team executing the project. The divergence between percentage of completion and the commencement of cash flow is not an incidental delay; it is the intrinsic behaviour of that segment. This pattern is known as the **last-mile problem** — the tendency of the final leg of distribution or implementation to generate cost and coordination burden of an order comparable to the sum of every leg preceding it. The term originates in parcel distribution, but its mechanics are in no way confined to trucks: the distance between the final pole and the meter in a grid interconnection, the energisation of white space in a data centre, end-user acceptance in an ERP deployment, the final circulation of security documents before first drawdown under a credit facility, all occupy the same structural position. The shared attribute is that the last leg is the segment in which every consolidation gain achieved upstream is unwound. The core of the mechanism is the reversal of the direction of scale. In line-haul transport, unit cost falls as vehicle fill rate rises; in the terminal leg, the variable that determines cost is not load but the number of stops and the density between them. Each stop carries a fixed time cost — parking, access, handover, signature — and that fixed cost is indifferent to consignment size, which means it grows disproportionately per unit on small consignments. Once delivery-window constraints, failed delivery attempts, address quality, urban access regulation, and the second leg generated in reverse by returns flow are added, the terminal segment ends up governed by a cost function wholly distinct from that of the rest of the network. A contractual asymmetry accompanies this. Every counterparty on the upstream legs of the chain — the carrier, the terminal operator, the line-haul provider, the EPC contractor — is bound by an instrument whose performance is measurable and whose delay is priceable. In the final leg, by contrast, the counterparty is frequently subject neither to a negotiated obligation nor to any mechanism through which delay converts into compensation: the consignee who is not at home, the utility that will not schedule an appointment, the municipal department that has queued the file, the operator who does not attend training. Delay risk in the terminal segment therefore has no natural contractual owner, and unowned risk, by definition, remains on the sponsor's balance sheet. Reporting on the average is entirely functional under a particular set of conditions: where the network is homogeneous, customer density broadly comparable, and pricing driven by a single list, the average delivers both comparability and a lower cost of measurement. The difficulty lies not in the shortcut itself but in the shortcut persisting after the conditions have changed. As the customer base extends toward rural geographies, smaller order sizes, narrower delivery windows, or categories with elevated return rates, the average ceases to represent the same reality, and because the reporting format has not changed, that extension transmits no signal whatsoever to the management table. The counterpart of this in the income statement appears not in a line item bearing the name of the final leg, but in the dispersion of gross margin across customer cohorts. So long as the aggregate margin target holds, the fact that one group of customers is internally subsidising another goes unremarked; and because the commercial team meets its growth objective through whichever segment grows most easily, the growing segment is frequently the subsidised one. On the working capital side, the same tendency accumulates in goods in transit and undelivered stock, in the cash tied up by the returns flow, and, where collection is conditioned on proof of delivery, in receivable days. No one of these items generates an alarm on its own; read together, they point to the same segment. On the project finance side, the cost settles somewhere sharper. The contract architecture typically defines mechanical completion as a threshold and stops the liquidated damages clock there, whereas revenue, debt service, and the DSCR calculation run from the commercial operation date. The interval between the two thresholds falls, by construction, outside the contractor's LD cap, and because most of what remains in that interval is contingent on third-party approvals, it cannot be accelerated either. The contingency budget, by the time this stage is reached, has largely been consumed, since the visible surprises of the construction period drew on it earlier; the consequence is that the least buffer remains at the moment of least flexibility. At the valuation desk this rarely presents as an argument about cost; it presents as an argument about measurability. The question posed in diligence is generally not whether the final leg is expensive but whether the company measures it at the customer or connection level, and where no such measurement exists, the buyer cannot test the durability of the margin and will price the uncertainty it was unable to test. The practical expression of that is sometimes a direct adjustment to the multiple, but more often a shift of consideration into earn-out or escrow and an extension of the representations and warranties package to encompass delivery performance. A cost that is not measured returns, at the moment of sale, as a risk that cannot be measured. The mechanism that neutralises this tendency is not individual attention but an architecture of measurement and authority, and it separates into four components. The first is that cost be measured at the smallest unit of service — the stop, the delivery, the connection, the acceptance — and that this measure enter management reporting in place of the network average rather than alongside it. The second is that the definition of completion be anchored to the revenue-generating event rather than to physical scope, with contractual thresholds and incentive structures recalibrated against that definition. The third is that the terminal segment have a named owner, since it has no natural contractual owner and any segment left unowned devolves by default to the sponsor. The fourth is that the review cadence change within this segment: as the count of outstanding items rises and the individual value of each falls, a monthly rhythm no longer offers sufficient resolution. BEIREK establishes this segment as a separate register before the contract is signed. Every work item falling in the final leg is enumerated individually, together with its own counterparty, its own triggering document, and its own calendar dependency; that register is not a sub-breakdown of the construction programme but a closing discipline operating independently of it. Completion thresholds are redefined against the revenue trigger, LD and incentive structures are attached to that definition, and items contingent on third-party approvals are governed by a monitoring cadence built on an obligation to pursue rather than on an expectation to wait. On the operations side, the same logic is applied to cost-to-serve measurement: unit cost is computed at the customer or connection level, and price lists and service level commitments are revisited against that distribution. The final leg is expensive not because it is the weak link in a project or a network, but because it is the point at which control changes hands; and control that has changed hands remains invisible in everyone's budget for as long as it goes unmeasured. The maturity of an organisation can be read with considerable reliability by asking whether it dissolves this segment into the average. ## Key Points - Economies of scale reverse in the final leg, because upstream steps consolidate volume while the terminal step disperses it, tying unit cost to stop density rather than to load factor. - The last mile is the only segment in which the counterparty is typically not a contractually bound professional, so delay risk finds no contractual owner and settles by default on the sponsor's balance sheet. - Distribution cost reported as a network average conceals the dispersion of gross margin across customer cohorts and systematically miscalibrates pricing at the segment margin. - On the project side, the interval between mechanical completion and commercial operation opens precisely where the liquidated damages cap has expired and the contingency budget has already been drawn down. - The final leg becomes manageable only when cost is measured per terminal unit and the definition of completion is anchored to the revenue-triggering event rather than to physical scope. ## Questions ### Why is the last-mile problem not confined to logistics? The mechanism operates in any terminal segment where control passes from a contractually bound party to one that is not. The final step of a grid interconnection, end-user acceptance in a software deployment, and the commissioning and administrative acceptance sequence of a facility all share the same structure: what remains is small in value, large in count, and non-substitutable, and because most of it depends on a third-party calendar, it cannot be accelerated by the sponsor. ### Where does last-mile cost appear in the financial statements? It does not appear in a line item bearing that name. It typically accumulates in the dispersion of gross margin across customer cohorts, in returns and reverse logistics expense, in the working capital tied up by undelivered stock, and in receivable days where collection is conditioned on proof of delivery. Because no single item triggers an alarm, the pattern becomes visible only when these items are read together against the same customer or geographic breakdown. ### Why does the interval between mechanical completion and commercial operation create a risky gap? The liquidated damages clock generally stops at mechanical completion, while revenue and debt service run from the commercial operation date. The intervening period falls outside the contractor's LD cap, and because most remaining work depends on third-party approvals, the sponsor cannot compress it. The contingency budget has usually been substantially consumed by that stage, with the result that the least buffer is available at precisely the moment of least flexibility. ### How is a company's valuation affected in the absence of cost-to-serve measurement? A buyer cannot test which customer segment generates the margin or how that margin will behave as volume grows, and it prices the uncertainty it could not test. The practical expression is sometimes a direct adjustment to the multiple, but more frequently a shift of consideration into earn-out or escrow and an extension of the representations and warranties package to cover delivery performance. Cost left unmeasured returns at exit as risk that cannot be measured. --- Source: https://www.beirek.com/en/blog/last-mile-problem Publisher: BEIREK LLC — https://www.beirek.com