Key Takeaways
- Network modelling should test whether a big shed lowers cost-to-serve and improves resilience, not just whether a site has enough square footage.
- UK big-box decisions in 2026 sit in a normalising market, with Savills forecasting baseline logistics vacancy trending towards 7% and rental growth around 2.7% for 2026.
- Road freight assumptions matter: Department for Transport data shows GB-registered HGVs lifted 1.53 billion tonnes of goods in 2025 and travelled 19.0 billion kilometres.
- Model labour, power, planning, port access, intermodal options and customer promise alongside rent, rates and transport cost.
- Board papers should show scenarios, trigger points and implementation sequencing, not a single “best site” with fragile assumptions.
Big sheds need a network case before a property case
A big shed is usually a 100,000 sq ft plus logistics unit, but the important question is not the building size. The question is whether the building changes the network in a way that customers, transport planners and finance teams can defend. A cheap rent can still be expensive if it adds trunking miles, weakens cut-off times or leaves the operation short of labour during peak.
That matters in 2026 because the UK big-box market is active without the exceptional urgency seen during the pandemic years. Savills describes the current market as a normalisation phase after a two-year correction, with occupational rental growth moving back towards pre-pandemic levels. Its modelling for units over 100,000 sq ft is built around future supply, future demand, availability and historic supply-demand relationships.
The modelling task should therefore start before a shortlist of sites becomes emotionally fixed. A property search may ask where space is available. A network model asks where inventory, orders, returns, inbound flows and labour should sit if the business wants a lower-risk operating footprint for the next five to 10 years.
For many operators, the answer will not be “lease the newest shed on the best-looking motorway junction”. It might be to consolidate two older warehouses into one regional campus, add a cross-dock rather than a full fulfilment centre, split inventory between north and south, or hold overflow with a 3PL until demand justifies a longer lease. The model is there to expose those choices before the heads of terms narrow them.
Start with the service promise and demand map
The first input is the promise you are trying to keep. Next-day parcel delivery, store replenishment, construction-site delivery, wholesale pallet networks and returns processing all produce different network shapes. A location that is excellent for trunking full pallets to the M1 corridor may be poor for late parcel injection or timed urban delivery.
Build the baseline from actual demand, not sales averages alone. Plot orders, pallets, cube, returns and failed deliveries by postcode area, customer type and delivery service. If seasonality is sharp, model peak weeks separately rather than smoothing them into an annual average.
Inbound flows need the same treatment. A network serving Asian container imports will weight Felixstowe, Southampton, London Gateway, railheads and empty-container management differently from one fed by UK factories or European groupage. If your existing inbound profile is changing because of nearshoring or supplier diversification, model the future mix as a scenario rather than a note in the margin.
The current site network is the control case. Document rent, rates, utilities, labour agency use, transport spend, overtime, storage density, service performance and known capacity pinch points. Without a disciplined control case, every candidate big shed risks being compared with a vague memory of the present operation rather than a measurable baseline.
This is also where existing internal planning work should be connected. A business already reviewing nearshoring and supply chain options may need a warehouse network that can absorb more European road freight, shorter inbound lead times or smaller production batches. A business reworking its 3PL model may need to compare owned space with outsourced overflow before committing to a long lease.
Add the 2026 property-market constraints
The property layer turns the ideal network into a practical set of options. It should include rent, rates, service charge, fit-out, lease length, incentives, power availability, yard depth, dock doors, eaves height, planning constraints and the timing of vacant possession. These variables can change the preferred answer even when two sites look similar on a map.
Savills’ 2026 rental-growth work is useful because it shows that national averages can hide local risk. It expects UK-level logistics rental growth of 2.3% to 3.5% per annum until 2029, with a 2.7% baseline estimate for 2026. It also expects vacancy to trend down to 7% in its baseline scenario and 5.5% in its upside scenario, while regional positions vary sharply.
Those regional differences should be in the model. Savills cites East Midlands vacancy around 10.1% to 10.5% with a 1.75 million sq ft confirmed development pipeline, while the South East may remain around 9% over the forecast period. JLL’s UK Big-Box Market Dynamics for H1 2026 reports 12.9 million sq ft of take-up, up 3.0% year on year, and 10.9 million sq ft under offer at the end of the half.
Do not treat these figures as a forecast for your own rent line. They are market context for scenario design. If the model only works when rent remains flat, incentives stay generous and Grade A supply is easy to secure, the board should see that sensitivity before approving the property route.
Power deserves its own line rather than a footnote. Automation, battery charging, cold chain, data-heavy operations and on-site renewable plans all make capacity and connection timing operational constraints. A lower-rent site with weak power may be less attractive than a more expensive building that can support automation, electric MHE charging and future vehicle-charging options.
Model transport flows, not just postcode distance
Transport cost is where many big-shed models become too neat. Straight-line distance to customers is useful for a first screen, but it does not capture road speed, delivery windows, driver availability, port dwell, pallet-network cut-offs, backhaul quality or the cost of serving remote demand from a centralised site.
Use official road-freight data to sanity-check the scale of assumptions. The Department for Transport says GB-registered HGVs operating in the UK lifted 1.53 billion tonnes of goods in 2025, down 3% from 2024, moved 162 billion tonne kilometres, down 4%, and travelled 19.0 billion kilometres, down 2%. Those numbers do not tell you where to put a shed, but they show why small percentage changes in loaded miles, empty running and delivery density matter.
Commodity mix also matters. DfT data says groupage was the most common commodity type by weight in 2025, at 383 million tonnes or 25% of goods lifted. If your network depends on groupage carriers, pallet networks or parcel injection, the model should include depot cut-offs and carrier service maps rather than simply calculating linehaul distance from a warehouse postcode.
Intermodal assumptions need the same discipline. DfT reports that 7.3 million of 163 million HGV journeys in 2025 involved at least one intermodal element, with 130 million tonnes, or 9% of goods lifted, using HGVs plus at least one other transport mode. Of those intermodal journeys, 66% began or ended at a shipping dock, 24% at a rail siding or terminal, and 12% at an airport.
For port-led supply chains, that means the model should test dock, rail and inland-terminal options explicitly. A warehouse close to a motorway may still be poorly placed if it adds container restitution cost, weakens rail access or increases risk when a port is disrupted. The same logic applies when modelling resilience against UK port delays or using bonded space in an ecommerce network design.
Build scenarios around decisions, not spreadsheets
The best network models are built around decisions the business might actually take. Each scenario should describe an operating choice: stay as-is and add overflow, consolidate into one national distribution centre, add a northern or southern regional node, use a 3PL for peak, move returns away from the main fulfilment centre, or build a port-centric inbound hub.
For each scenario, calculate at least six outputs. These are annual property cost, transport cost, labour cost, implementation cost, service coverage and operational risk. Add carbon or mileage proxy where the business reports emissions, but keep the method clear so the board understands whether it is a directional estimate or a formal carbon calculation.
Sensitivity testing is where weak recommendations usually appear. Test lower-than-expected demand, faster ecommerce growth, a lost customer, a new customer cluster, delayed automation, higher agency labour cost, rent inflation, fuel movement and port disruption. Savills notes that its own logistics rental model can run lower take-up or higher speculative-delivery scenarios; an operator’s network model should be equally honest about uncertainty.
Labour deserves more than a wage-rate input. Model commute catchments, competition from nearby warehouses, shift patterns, agency availability, public transport, unemployment rates where reliable, and the operational impact of churn. A site that looks strong on transport can fail if it cannot recruit for nights, peak or specialist roles.
The output should not be a ranking table alone. It should explain why one network shape is preferable, what would have to change for another option to overtake it, and which assumptions are most exposed. That is the difference between a property justification and a decision model.
Turn the model into a board-ready recommendation
A board paper should start with the recommended network shape and the conditions under which it remains valid. The headline is not “Site B scores 84 out of 100”. It is “a Midlands-led two-node network protects next-day service at lower transport risk, but only if labour availability and power connection dates are secured before lease commitment”.
Include a clear comparison table. Keep it short enough for directors to use, but detailed enough to show the trade-offs.
| Decision output | What to show | Why it matters |
|---|---|---|
| Cost-to-serve | Cost per order, pallet, case or tonne | Lets finance compare scenarios on operating economics |
| Service coverage | Cut-off times and delivery promise by region | Shows whether the network supports customer commitments |
| Property exposure | Rent, rates, lease length, incentives and break options | Prevents transport savings being bought with rigid property risk |
| Labour risk | Availability, wage pressure, agency reliance and commute catchment | Tests whether the building can be staffed reliably |
| Resilience | Port, carrier, supplier and demand-shift stress tests | Shows whether the network can absorb disruption |
| Implementation | Fit-out, systems, stock move, launch phasing and dual-running cost | Makes transition risk visible before approval |
Implementation phasing is often as important as the final state. A large warehouse move can require dual running, temporary stock buffers, WMS changes, transport re-tendering, carrier re-labelling, customer cut-off changes and new health-and-safety procedures. If the model assumes a clean switch from one network to another, the business may approve the right endpoint with the wrong transition budget.
Set trigger points for review. Examples include vacancy rising or falling beyond a defined range, demand moving 15% above or below plan, a port mix shift, a lost customer, a new automation case, or a labour market change in the chosen catchment. The board should know when the decision needs to be reopened rather than waiting until the lease event forces the question.
Checklist before leasing a big shed
Use this checklist before a site moves from preferred option to commitment:
- The baseline network has been measured from actual orders, pallets, cube, returns and service failures.
- Candidate sites have been tested against customer promise, inbound flows, port access, labour catchments and carrier cut-offs.
- Property assumptions include rent, rates, incentives, fit-out, power, lease length, planning and vacant-possession timing.
- Transport modelling includes loaded miles, empty running, trunking, final-mile service, intermodal options and disruption cases.
- At least three scenarios have been tested against demand, rent, labour, fuel and implementation sensitivities.
- The recommendation shows cost-to-serve, service impact, property risk, resilience and implementation sequencing.
- The decision paper states the triggers that would make the business defer, lease, consolidate or use 3PL capacity instead.
If the evidence is weak on two or more of these points, the answer may be to slow the property process rather than force a decision. In a normalising market, speed still matters, but poor commitment speed can lock in higher operating cost for years.
Frequently Asked Questions
What is network modelling in big-shed logistics? Network modelling is the process of testing where warehouses, cross-docks, transport routes and inventory should sit to meet service and cost targets. For big sheds, it links property decisions with customer demand, inbound flows, labour, transport and resilience. It should compare several network shapes rather than justify a site already chosen.
How large is a big shed in UK logistics? Property-market research commonly treats big-box or big-shed logistics units as buildings over 100,000 sq ft. Savills’ 2026 rental-growth modelling uses that threshold for UK logistics units. The threshold is useful for market comparison, but the operational test is whether the building supports the network role you need.
Should transport cost or rent drive the site decision? Neither should be modelled alone. Rent can be outweighed by trunking miles, failed service levels, driver cost, carrier constraints or weak labour availability. A good model shows the total cost-to-serve and the service risk of each scenario.
How many scenarios should a warehouse network model include? Most strategic decisions need at least three scenarios: keep and optimise the current network, consolidate into a larger node, and add or shift to a regional node. Add 3PL overflow, port-centric or returns-specific scenarios when those are genuine options. The goal is to show trade-offs, not create a long list of theoretical sites.
When should a business defer a big-shed lease? Deferral is sensible when demand is uncertain, labour risk is unresolved, the power or fit-out case is weak, or the model only works under optimistic rent and service assumptions. It can also be the right answer when short-term overflow through a 3PL protects service while the network evidence improves. The board should see deferral as an active option, not a failure to decide.