Lastmily · Industries · Grocery & quick commerce

Grocery & quick commerce

A two-hour window, not a day.

Supermarkets, mini markets, convenience stores, fresh food, coffee and beverages – where the promise is measured in hours, the goods do not wait, and the basket is worth less than the van.

The economics

The clock and the cold chain are the same constraint.

Grocery is the segment where a routing mistake spoils the goods, not just the mood.

Three things are true at once and they fight. The window is short, so there is no slack to absorb a bad sequence. The goods are perishable, so a late stop is not a late stop, it is a write-off. And the basket margin is thin, so the van has to be full and the round has to be dense.

A solver that handles those one after another – sequence first, then check capacity, then check temperature – will produce a plan that is wrong in a way nobody notices until the crate is opened.

Cold chain is one of the constraints this engine solves together with all the others. A stop that breaks it is not a late fix; it is a different route.

And because the day moves – a store order lands, traffic closes a street – the plan is re-solved live against the remaining work rather than defended until it fails.

≈20sto solve 700 stops – a product decision, not a benchmark
≈35%typical route adherence in B2C – industry baseline, not a Lastmily result

Next day, same day and express are modes of the same engine, not three products.

Lastmily · Grocery & quick commerce

Density, filmed

Hyperlocal is where this engine is at its most obvious.

When the stops are a few doors apart, the difference between a good round and a bad one is not kilometres of driving – it is how many times the van stops and whether the driver agrees with the order. Two of the keynote films, because they are the two that matter most at this density.

Park once 11 · 46 mparking stops · longest walk

Eleven parking stops instead of one per door

At hyperlocal density most doors are walkable from a parking spot the driver already uses. The engine plans the walk as well as the drive: eleven parking stops for the whole round, and the longest walk on it is forty-six metres.

His own order +1.5%above the best found

The sequence he was going to drive anyway

In a small dense area a driver has a route in his head before he opens the app. Thousands of sequences score within a couple of per cent of each other, so the engine picks his – and pays 1.5% for a round that actually gets driven.

Every film on this page is a solved round, not a drawing. The kilometres and the times are modelled from the geometry on screen – they are what these rounds cost, not a result measured in your operation. From the Parcel+Post Expo 2026 keynote at the Last Mile Hub, ExCeL London.

Send us one dense morning.

One ordinary round in your tightest area, re-planned both ways, with the parking stops and the walking distance on it.

Lastmily · Grocery & quick commerce

What the engine routes against

The constraints that bind a dense round.

A slice of the full catalogue – the ones that decide a grocery day. All of them are solved in the same pass, not applied to each other as filters.

Stops

  • Time windows
  • Predicted service time
  • Cold chain
  • Entrance & floor

Drivers

  • Driver skills
  • Driver profiles
  • Max walking
  • Park & walk

Vehicles

  • Vehicle capacity
  • Truck dimensions
  • Multi-depot & reloads
  • ADR

Environment & day

  • Live traffic
  • Re-solved against the remaining work

That is a slice of what the engine solves against. The whole constraint set is on the routing page, with the films that show what each one changes.

Same day and one hour

Modes of the same engine. The round is re-solved live as orders land, so a one-hour promise is planned rather than raced.

Cold chain

Solved with capacity, dimensions and windows – not applied afterwards as a filter over a route that already exists.

Tell me before I ask

A real window, a proactive notification and the chance to change it from a phone. In grocery the recipient is usually at home and impatient, which is a different problem from being out.

Execution

Re-plan, do not race.

The failure mode in quick commerce is heroics: a dispatcher on the phone, a driver told to skip ahead, and a round that unravels for the rest of the shift. The engine exists so the answer to a late order is a new plan in twenty seconds rather than an instruction shouted down a line.

Live monitoring on a same-day grocery operation
Live monitoring on a same-day grocery operation

Lastmily · Routing

What it does to the numbers

Measured where it has been measured.

There is no published grocery case in our set yet – so these are the platform figures, with the labels they were measured under, and the store-replenishment case as the closest neighbour.

97%

time windows met

store-replenishment case

−12%

vehicles on the road

store-replenishment case

−26%

CO₂ emissions

store-replenishment case

−35%

inbound calls – notifications and self-service

comparable implementations

None of these is a promise, and none of them was measured in grocery. Each one is a number we can baseline on your data before you commit to anything.

Millions

monthly shipments

platform-measured, across the customer base

≈0%

logo churn

platform-measured, to date

Top 10

Global Routing Startup

StartUs Insights · 2025

Top 5

Most Promising EU Logistics Startups

Leaders in Logistics · 2024

Awarded

Most Disruptive Company in Supply Chain, Greece

Thales the Milesian Business Awards

Give us one peak day.

Saturday morning, every order, every constraint. We will build it every way and show you what each profile costs the other.

Keep reading

Lastmily · Proof

The close

Pick the number we should be measured on.

Then let us measure it before you commit to anything.

Lastmily