Where every site should be, and what it costs to get there.

Hubshape models a distribution network from its factories to every customer: the ground each site serves, the lanes between them, the cost to serve and the peak, then sets each option against today on cost, service, miles and carbon.

Talk to Libra about Hubshape Watch a hub find its centre

Run as a gated studyBuilt on your own loads and costsThe model is handed over

A Libra Europe product

Network · demo demandBase case

Factories: 2
Hubs: 3
Outbases: 4
Customers: 122
Lanes: 10

FactoryHubOutbaseLane, width by pallets

Every site tested against where its customers really are.

Centre of gravity · East hubToday's network

Road miles a week

87,126

0% on today

Average to a customer

68 mi

0 mi on today

Volume within 2½ hours

83.4%

+0.0 pts on today

Customers changing hands

0

of 122

18% fewer road miles, service held

East hubOther hubsCustomer, by volumeCandidate site

Each customer belongs to its nearest hub.

Hubshape draws the ground each site serves from modelled road miles, with every customer weighted by the pallets it takes.

Weighted by volume, the East hub's centre sits far inland.

Its customers pull the centre of gravity inland, towards London. Then the sites you could really take are each costed in full.

The hub moves 110 miles, and 17 customers change hands.

The best site is 34 miles from the centre. As the hub travels the catchments re-form and every lane is redrawn.

Fewer road miles, with service held.

The same customers on the same demand, 18% fewer road miles a week, and more of the volume within two and a half hours.

Cost to serve, worked out for every customer from the flow its goods take.

Cost to serve · flow pathsLanded £ a pallet

£22£99 a pallet

Flow path · Scotland hub to C-215

£99.19

↑ site average £28.64

First mileOrigin to hub, full loads£6.40
StorageStorage days × rate per pallet£4.10
PickItem pick, touches × rate£7.90
Last mile108 mi each way, 1 pallet a drop£79.19
Peak spaceReserve space, peak weeks only£1.60

3.5× the site average, and 80% of it is last mile on a small drop.

Invented customers. Figures are illustrative.Next dearest: C-221 £93.25 · C-208 £78.10

Every customer, shaded by what it costs to serve.

Landed cost per pallet runs from £22 to £99, built from the segments its goods really pass through.

The dearest is lit, and its flow path traced.

First mile, storage, pick, last mile and peak space, each traced to the driver behind it.

3.5 times the site average, and 80% of it is last mile.

Now you know what to fix: price it, consolidate its drops or serve it from somewhere else.

Peak space sized for the weeks that need it.

Seasonality and capacity · hub storage52 weeks, indexed to the mean

Peak week

Wk 49 · 1.33×

Weeks over ceiling

5 of 52

Design point

P90 · 1.05×

Space bought

5 of 52 weeks

An invented demand profile. Figures are illustrative.

Every option set against the base case, on the measures a board will ask about.

Configurations against base caseBase case signed

Same demand, same rates

Base

Base case

Option A

Close the South West hub

Option B

Add a Midlands cross-dock

Option C

Move the East hub inland

Cost to run

£m a year

0.00

0.00+3.07

0.00−0.45

0.00−1.71

Cost of change

£m once

–

Within 2½ hours

share of volume

0%

0%−26 pts

0%+6 pts

0%+5 pts

Road miles

m a year

0.00

0.00+50%

0.00−11%

0.00−18%

Carbon

t CO₂e a year

0

0+50%

0−11%

0−18%

Cost against carbon

Crosshair on the base case. Bubble size is volume within 2½ hours.

Invented network. Figures are illustrative.Option C: £1.71m a year less to run, for £1.8m once

The base case is today's network, costed.

It is signed off as a fair picture of last year before any option runs, so every option is measured against the same thing.

Each configuration runs on the same demand.

Close a hub, add a cross-dock, move a hub: each is a configuration of one model, run on the signed baseline with the same rates.

Cost to run, cost of change, service, miles and carbon.

Each measure sits against the base case, and the scatter puts cost against carbon, so cheaper and cleaner is bottom left.

The option to take forward, with what it costs to get there.

Option C costs £1.71m a year less to run for £1.8m once, with 18% less carbon and service held.

Run as a study with four gates, each closed by a decision you sign.

No stage starts until the one before has met its exit criteria and been signed Go, No-go or Reshape, with the evidence in a printable gate pack.

Gate 1 of 4 · Mobilise and data: confirm the measures and the candidate sites

M1 · stop point
Mobilise and data
- Data request returnedOpenMet
- Validation checks run, treatments agreedOpenMet
- Candidate sites listedOpenMet
Decision
Confirm the measures and the candidate sites
GoNo-goReshapeSigned
M2 · stop point
Baseline
- Reconciles to last year, line by lineOpenMet
- Service constraints agreedOpenMet
- Every assumption bound to the modelOpenMet
Decision
Sign off the baseline as a fair picture
GoNo-goReshapeSigned
M3 · stop point
Scenarios
- Each configuration run on the signed baselineOpenMet
- Both centre of gravity methods testedOpenMet
- Rejected options written up, with reasonsOpenMet
Decision
Choose the options to take forward
GoNo-goReshapeSigned
M4 · stop point
Recommend
- Recommendation with its robustness rangeOpenMet
- Sequence of changes and board reportOpenMet
- Model handed over to your teamOpenMet
Decision
Take the recommendation to the board
GoNo-goReshapeSigned

At M4 the board gets a decision it can take, with the evidence for every figure in the gate pack, and your team keeps the model to re-run when the business changes.

What operations and finance directors ask before a study starts.

Is Hubshape software we buy, or a study Libra runs?

Both. Libra runs the study with your team through four gates, using Hubshape as the model. At the end the model is handed over, so your team can re-run it when volumes, sites or rates change.

What data does it need?

A coded data request covers delivery points, loads, the site master and the cost ledger, each item marked essential or desirable. Uploads are checked as they arrive, every gap gets a treatment you agree, and estimated rows stay flagged in every output.

How do we know the baseline is right?

It has to reconcile to last year's actual costs, line by line and within a tolerance you agree, before anyone signs it at the second gate. Any option run before then is labelled indicative.

How are road miles and service worked out?

Miles are straight-line distances corrected by a road circuity factor. Service is tested on modelled drive time, with trunk and urban speeds, so a site that looks close on the map but sits behind slow roads shows up.

Where does the carbon figure come from?

From the miles each option runs, at emission factors agreed for your fleet, with building energy as a separate layer. It sits beside cost, so an option that saves money by running more miles is plain to see.

Does it just pick the cheapest option?

No. Each option is shown against the base case on cost to run, cost of change, service, miles and carbon, tested against failures such as losing the main hub in peak, and every rejected option is written up with its reason. The decision stays with your board.

Can it hold our own sites' limits?

Yes. Site visits feed a constraint register, and each constraint is a rule the model tests: a site's space, a delivery window, a road restriction. An option that breaks one is either changed or carries it as a stated trade-off.

Start with a year of your own loads.

Talk to Libra about a Hubshape study. We will scope the data request, the measures and the first gate with you.

Talk to Libra about Hubshape

Hubshape is built and run by Libra Europe, and the model stays with your team at the end.