Own Warehouse vs 3PL: A Ten-Year Cost Comparison

The board pack shows two bars. One is the 3PL proposal at $3.3 million a year; the other is the company-run option at $3.27 million, built from a lease quote and a headcount plan. The cheaper bar wins, the lease is signed, and three years later the controller is explaining racking capex, an overflow contract every November and a lift-truck fleet that needs replacing earlier than anyone modelled.

The fault was not the arithmetic inside either bar. It was comparing one year of operating cost when the decision commits cash for a decade. A ten-year discounted model, built on the same volume profile for both options, shows which choice is cheaper in present-value terms, how sensitive that answer is to the discount rate and the 3PL escalator, and which tax and accounting rules move the result in the US and in India.

• Model structure: a common volume baseline, a company-run cost build, a 3PL rate card turned into a ten-year cash stream, and one discounting formula applied to both.
• Market detail: a worked US ($) and India (₹) comparison, depreciation and GST credit effects, a peak-overflow failure trace, and a checklist finance can use before approval.

Why one-year cost comparisons pick the wrong warehouse option

A single-year view hides the three things that decide this choice: up-front capital, different escalation paths, and the value left at the end. Company-run sites spend heavily in year zero and recover it slowly; 3PL contracts spend little at the start and compound their escalator every year. Only a multi-year cash view puts both shapes on one scale.

Three distortions recur in the spreadsheets finance teams inherit. The lease rent is compared with the 3PL storage charge alone, leaving out labour, equipment and systems. Year-one numbers are used for all ten years, so a 3.5% labour escalator and a 6% contract escalator look identical. And the own-site model is sized for average stock while the 3PL quote bills actual usage, which flatters the company-run option every peak season.

The comparison also needs a defined owner for each input. Demand planning owns volume, procurement owns the rate card, real estate owns the lease, and treasury owns the discount rate. When one analyst invents all of them, the model reflects that analyst's optimism rather than commitments anyone will be held to.

Setting one volume and service baseline for both options

Both options must be priced on the same physical workload, stated in the units each one bills or staffs against. If the 3PL is quoted on average pallets and orders while the company-run site is staffed for a different service level, the comparison measures two different operations, not two ways of running one.

Start from the WMS, not from a sales forecast. Pull daily pallet-on-hand snapshots for two years, inbound receipts in pallets, outbound shipments, orders and order lines, and any value-added work such as labelling or kitting. The ratio of peak to average stock is the single most important number: a site you own must hold the peak, while most shared-warehouse contracts bill what you actually occupy.

• Storage driver: average pallets on hand for the 3PL bill; peak pallets plus a 5-10% honeycombing allowance for the company-run racking count.
• Handling driver: pallets received and pallets shipped per year, kept separate because many 3PLs price inbound and outbound differently.
• Order driver: orders and lines per order, which set picking labour on your side and per-order fees on theirs.
• Service level: same cut-off time, same dispatch-day target and same accuracy requirement written into both cost builds.

The worked example in this article uses a US site holding 8,000 pallets on average and about 9,000 at peak, receiving and shipping 60,000 pallets a year and dispatching 240,000 orders. The India site holds 6,000 pallets on average, peaking near 7,500, moving 45,000 pallets and 3,00,000 orders a year. The company-run racking counts of 9,600 and 8,200 positions cover those peaks with allowance. Our warehouse capacity planning guide explains how to measure positions in the units that actually constrain a site.

Building the company-run cost stack line by line

A company-run site carries five cost families: year-zero capital, occupancy, people, running costs and periodic replacement. Each needs its own escalation rate, because rent, wages and power do not rise together, and a single blended inflation figure hides the line most likely to break the business case.

Capital comes first. In the US case, 9,600 selective racking positions at $95 each cost $912,000; reach trucks, counterbalance trucks and pallet jacks cost $540,000; the WMS licence and implementation cost $450,000; dock levellers, lighting, battery charging and security fit-out cost $380,000; and recruitment, training and parallel running during start-up cost $260,000. That totals $2,542,000 before the first pallet is received.

Operating cost in year one is $3,273,000: a 120,000 square-foot lease with service charges at $9.75 per square foot ($1,170,000, rising 3% a year), 22 associates at a loaded $64,000 ($1,408,000, rising 3.5%), three supervisors and a site manager budgeted at $330,000, utilities and maintenance at $210,000, WMS support at $95,000 and contents and liability insurance at $60,000.

The India company-run build follows the same structure in rupees. Year-zero spend is ₹11,54,60,000: racking for 8,200 positions at ₹6,500 each (₹5.33 crore), lift trucks (₹2.40 crore), WMS (₹1.10 crore), fit-out including docks, fire hydrant and sprinkler work, lighting and CCTV (₹1.90 crore), start-up (₹60 lakh) and ₹21.6 lakh of GST that cannot be credited, explained in the tax section. Year-one operating cost is ₹6,34,20,000, led by a 90,000 square-foot lease at ₹24 per square foot per month, rising 5% a year.

Turning a 3PL rate card into a ten-year cash stream

A 3PL proposal is a price list, not a cost forecast. To compare it with ownership, multiply each rate by the matching volume driver, add the fixed fees, apply the contract escalator year by year and add the one-off onboarding and transition costs that the proposal often prices separately or leaves out.

The US proposal in the example charges $18 per pallet per month for storage on 8,000 average pallets ($1,728,000), $5.50 per pallet in and $5.50 per pallet out on 60,000 pallets ($660,000), $3.20 per order for picking and packing on 240,000 orders ($768,000), and a $180,000 account management and systems fee. Year one totals $3,336,000, escalating 3.5% a year under a CPI-linked clause with a cap.

• Minimum charges: a monthly minimum or a guaranteed pallet volume turns part of the variable fee into a fixed cost; model it as a floor under the storage line.
• Escalator wording: "CPI plus 1%" and "the higher of CPI or 3%" compound very differently over ten years; use the contract text, not the sales summary.
• Onboarding and transition: stock transfer freight, relabelling, dual running and integration build; the example uses $150,000 for the US and ₹45 lakh for India.
• Exit terms: notice period, handback condition and any fee for removing stock at term end; include them if the plan assumes a re-tender in year five.

How the storage line is calculated matters as much as the rate. Whether the 3PL bills on a month-end snapshot, the monthly peak, or a daily pallet-day count can move the storage line by 10% or more for seasonal stock. Our article on pallet-day storage billing explains those clocks for an invoice that has already been raised; here the question is which clock to assume when you forecast ten years of invoices before signing.

The India proposal charges ₹650 per pallet per month on 6,000 pallets (₹4.68 crore), ₹180 per pallet in and ₹180 out on 45,000 pallets (₹1.62 crore), ₹38 per order on 3,00,000 orders (₹1.14 crore) and a ₹36 lakh management fee, for ₹7,80,00,000 in year one before GST, escalating 6% a year. GST on the invoice is excluded because the depositor recovers it as input tax credit, provided the Section 16(2) conditions are met.

The discounting formula that puts both options on one scale

Discount every cash outflow to today at the company's hurdle rate and compare the two present values. The option with the lower present value of cost is cheaper; the equivalent annual cost converts that present value back into a level yearly figure that operations managers can compare with a budget line.

Keep the model pre-tax unless the tax adviser supplies after-tax cash flows for both options; mixing a tax-adjusted company-run case with a gross 3PL case is a common way to manufacture a winner. Residual value at year ten enters as a negative cost, because racking and trucks can be resold or redeployed. The illustrative model credits 25% of racking cost and 10% of truck cost.

Ten-year present value of cost, run once per option

PV  = sum over t = 0..10 of  (K_t + O_t + M_t - R_t) / (1 + r)^t

O_t = sum over cost lines k of  B_k * (1 + g_k)^(t - 1)     for t >= 1
EAC = PV * r * (1 + r)^n / ((1 + r)^n - 1)                   with n = 10

K_t  capital spend in year t (racking, trucks, WMS, fit-out, start-up, onboarding)
B_k  year-one cost of line k (rent, labour, storage fee, order fee)
g_k  escalation rate of line k (lease clause, wage forecast, 3PL escalator)
M_t  one-off replacement or overhaul in year t
R_t  residual value recovered in year t (zero except t = 10)
r    pre-tax hurdle rate approved by treasury (example: 9% US, 12% India)
EAC  equivalent annual cost: the level yearly charge with the same PV

Decision rule: choose the lower PV only if the gap exceeds 3% of the smaller PV;
below that, the choice turns on flexibility and risk, not cost.

A ten-year own-warehouse versus 3PL model kept in the FP&A planning workbook should discount every outflow at the approved hurdle rate and treat a leased building under ASC 842 or Ind AS 116 as cash rent, not accounting expense; at 9%, the illustrative US case favours the 3PL by $1.75 million.

Use cash rent rather than lease accounting entries in this model. A building lease puts a right-of-use asset and a lease liability on the balance sheet under both ASC 842 and Ind AS 116, but those entries do not change the cash paid to the landlord. Accounting presentation matters for covenants and EBITDA, which the scope note later in this article addresses, but it should not decide a cash comparison.

Worked ten-year comparison for a US site and an Indian site

On the illustrative inputs, the 3PL is cheaper in present-value terms in both markets: clearly in the US, and only narrowly in India, where the undiscounted totals point the other way. That reversal is exactly why the discounting step cannot be skipped. All figures below are illustrative and exclude recoverable sales taxes.

In the US, the company-run site has the lower year-one operating cost by $63,000, yet it loses by $1.23 million undiscounted and $1.75 million in present value, because $2.5 million of year-zero capital is never recovered by such a thin annual saving. In India, the company-run site saves ₹9.45 crore undiscounted, but at 12% the 3PL is ₹45.3 lakh cheaper: the saving arrives late, the capital leaves early.

A gap of ₹45 lakh on a present value of ₹55 crore is under 1%, well inside the 3% threshold in the decision rule. The India result is therefore a tie on cost, and the decision should move to the sensitivity tests and to non-cost factors such as control of service, flexibility and exit risk.

Depreciation, GST credit and lease accounting effects by market

Tax rules do not change the cash paid to suppliers, but they change when the tax saving arrives and, in India, whether GST becomes a cost at all. Those effects favour different options in each market, so they belong in an after-tax version of the model prepared with the tax adviser.

In the US, IRS Publication 946 lists a warehouse as nonresidential real property with a 39-year recovery period, so an owned building yields a slow tax shield. Racking and lift trucks fall in shorter MACRS classes; property with a recovery period of 20 years or less acquired after 19 January 2025 qualifies for the 100% special depreciation allowance, which pulls the tax saving on $1.45 million of equipment into year one.

For a US company-run distribution centre, the fixed-asset register should depreciate an owned warehouse building over 39 years as nonresidential real property, while racking and lift trucks with recovery periods of 20 years or less qualify for the 100% special allowance when acquired after 19 January 2025, per IRS Publication 946.

Publication 946 also describes a separate allowance for qualified production property, but it covers buildings used as an integral part of manufacturing, production or refining that substantially transforms a product. A finished-goods distribution warehouse should not be modelled with that allowance unless the tax adviser confirms the activity qualifies.

• India GST on construction: CGST Section 17(5)(c) and (d) block input tax credit on works contract services and on goods or services used to construct immovable property on your own account, other than plant and machinery.
• Plant and machinery boundary: the Explanation to Section 17 covers apparatus, equipment and machinery fixed to earth by foundation or structural support and excludes buildings and civil structures; the Finance Act 2025 wording applies from 1 July 2017.
• India GST on 3PL fees: tax charged on the 3PL invoice is normally creditable, so the 3PL option carries no comparable trapped tax.
• India income-tax depreciation: Appendix I to the Income-tax Rules allows written-down-value depreciation by block, with a general rate of 15% for plant and machinery.

In an Indian own-warehouse plan, the GST credit ledger cannot absorb tax on works contract or other inputs used to construct the building, because CGST Section 17(5)(c) and (d) block that credit outside plant and machinery; the illustrative ₹21.6 lakh blocked on ₹1.2 crore of civil fit-out therefore sits in capex.

Peak overflow, ramp-up and exit: the costs models forget

The most common post-approval surprise at a company-run site is overflow: stock that exceeds the racking you built and has to go somewhere expensive at short notice. A 3PL absorbs that peak inside its network; your own site pushes it into spot storage contracts priced at the worst moment of the year.

The trace below comes from an illustrative US site in its third November, where the model sized racking at 9,600 positions against a forecast peak of 9,000 pallets and did not include an overflow line.

1. 06:40, 3 November: the WMS location utilisation report shows 9,410 of 9,600 positions occupied (98.0%); twelve inbound trailers are booked for the day.
2. 08:15, 3 November: ASN 450-118204 for 28 pallets is received at door 6; directed putaway returns "no eligible location" and the pallets are staged on the dock floor.
3. 10:30, 3 November: the site manager raises purchase order 4500087121 to a public 3PL for 600 overflow pallets at $24 per pallet per month and $14 per pallet in and out, plus shuttle trips at $420 each.
4. 28 November: month-end accrual for overflow posts $40,440: $14,400 storage, $16,800 handling and $9,240 for 22 shuttle trips, none of it in the approved model.
5. 15 January: the post-peak review adds an overflow line, priced at the probability of exceeding 9,600 positions times the spot cost, to model version 4 for all future site cases.

Ramp-up is the second gap. New sites rarely reach engineered productivity in month one; budgeting 70% in the first quarter and 85% in the second is a reasonable illustrative assumption. Exit is the third: a ten-year lease signed for a five-year volume forecast leaves stranded rent if a channel shifts, while a 3PL contract with a 12-month notice period caps that exposure.

Which model steps are fixed rules and where software assists

The arithmetic of the comparison should be locked formulas that nobody overrides; the judgement-heavy inputs are where tools, including AI extraction and forecasting, can speed up a person's work without replacing their sign-off. Keeping the two apart makes the model auditable when the board asks how a number was reached.

For a unit view of the same costs, our guide to warehouse cost per order builds the company-run cost from activities, which is a useful cross-check on the labour line here. If the decision goes to a 3PL, testing its billing engine against your hardest contract terms is covered in our 3PL billing software evaluation.

Approval checklist before committing to either option

Approve the decision only when every input has an owner, the model has been run at three discount rates and two escalator cases, and the result survives the overflow and exit tests. If the present-value gap is under 3%, record that the choice was made on control and flexibility, not cost.

More practical guides on warehouse cost, contracts and billing are collected in our logistics and supply chain hub. If you want help building this ten-year model from your own WMS history and 3PL proposals, or need the volume and billing data it depends on pulled together across systems, you can send us a request for proposal and we will scope the work with your finance and operations leads.