Grocery fulfilment: worked out examples

A useful interim synthesis is short:

“Picking accuracy improved, but 28 of the 100 routes still leave late. I would keep the accuracy improvement and trace those delays from order readiness through loading to departure.”

That tells the interviewer what you have learned and why the next question follows. It also avoids two common leaps: declaring fulfilment fixed because one metric improved, or calling the picking intervention useless because another metric did not move.

If the case asks whether to expand the picking intervention, its cost and the value of fewer errors still matter. But those economics are separate from diagnosing the remaining departure delay.

To find where those 28 routes lose time, ask for the dispatch records:

“For the 100 scheduled routes, could we compare when their orders, vehicles and drivers were ready with their loading times and departure deadlines? How many routes can the depot fully load in each departure window?”

These timestamps let you distinguish between three possible delays:

• Orders are not ready when loading should begin.
• Orders are ready, but trucks, drivers or loading resources are unavailable.
• Loading finishes, but another release step delays departure.

Compare routes with similar departure deadlines. Knowing that the depot has spare capacity later in the day will not explain whether it can load an early route in time.

In response, the interviewer shares Exhibit 1: loading demand and capacity for that day’s 100 routes, split into early, middle and late departure windows. They confirm that each route’s orders, vehicle and driver were ready at the start of its assigned window. Loading must finish by the end of that window for the route to depart on time; assume there is no further delay after loading.

Each route needs one vehicle to be fully loaded. The capacity column shows how many vehicles the depot can finish loading within each window, using the bays, staff and equipment available then.

Compare the routes needing to leave in each window with the loading capacity available. Where would routes have to wait?

In the early window, 40 routes need to leave but only 24 can be loaded in time, leaving 16 late. The middle window has a shortfall of 12, with capacity for 28 of its 40 routes. The late window has capacity for all 20 routes due then.

Even if loading is scheduled to maximise on-time departures, only 24 + 28 + 20 = 72 routes can leave on time. That matches the 72% result in the problem statement.

The late window has 28 spare loading slots. They can help clear the backlog, but cannot make an earlier departure on time. Total daily capacity of 100 therefore does not mean the depot can dispatch all 100 routes on schedule.

The arithmetic suggests moving 16 slots to the early window and 12 to the middle, leaving 20 late. That would match 40, 40 and 20 routes respectively.

It is a scheduling proposal to validate. Ask what creates a loading slot: bay availability, trained staff, equipment, vehicle access or a combination. If the late slots depend on a shared loading crew that is committed elsewhere in the morning, changing the schedule on paper will not create usable early capacity.

Also check whether orders can actually be ready under the revised plan. The table’s readiness assumption must hold when the operation changes.

A practical response would be:

“The capacity mismatch could explain all 28 late departures in this example. I would test whether the late loading resource can be reassigned to the first two windows before adding more daily capacity.”

Suppose a scheduling change is feasible and costs £300 a day in shift premiums. It prevents 28 late departures, each associated with an expected £20 avoidable cost from redelivery, service credits and additional handling.

The daily benefit is 28 × £20 = £560. After the £300 premium, the expected improvement is £260 a day on the stated assumptions.

The change needs to prevent at least 15 late departures to cover its £300 cost at £20 each. If the trial only prevents ten, it saves £200 and costs £300, so the economics need revisiting.

Keep service benefits that are not included in the £20 visible, but do not count them twice. Equally, include any new errors or overtime created by the compressed loading plan.

Monitor departure performance by window, loading queues, picking errors and the extra operating cost. If on-time departures improve while error rates worsen, the team may have transferred pressure upstream.

Use comparable route mixes and volumes when judging the trial. The question is whether the changed schedule delivers more accurate orders on time at an acceptable cost.

Then update the synthesis:

“The picking change addressed accuracy. The dispatch analysis identified a timing mismatch: 28 required slots were missing before early deadlines despite spare late capacity. I would trial the feasible shift change and expand it if it prevents more than 15 late departures a day at the stated cost while preserving picking quality.”

Try Grocery Delivery Suburban Losses. Before requesting another exhibit, say what the current evidence resolves and which remaining question matters. The cold-chain capacity example shows how to structure a related timing and handoff problem.