- 01
Market
Saudi Arabia
- 02
Written for
COO, Head of CX
- 03
Languages
EN · AR
The gap
The minutes between the pass and a courier accepting
Every late delivery has a biography, and most of it is written before the courier moves. These are the stages an order crosses after the kitchen finishes, and how each one stalls.
A flow from kitchen-ready through assignment, offer, acceptance and collection, with the stall causes at each stage and a branch to a human dispatcher for stalled orders.
- 01
Kitchen marks ready
Or does not — a pass with no ready event leaves every later stage guessing, and the wait it causes is invisible in every report.
- 02
The order enters assignment
Batching logic may hold it, waiting to pair it with a second order that never comes — efficient for the fleet, slow for this guest.
- 03
A courier is offered the job
Zone boundaries decide who is eligible; a branch sitting at the edge of its zone can be minutes from couriers it is never offered.
- 04
The courier accepts
Offers expire on couriers whose recorded state is stale — marked free while finishing another drop, marked nearby on a position minutes old.
- 05
Travel and collection
The only stage the road controls — and often the shortest. The stages before it are where the shelf time was manufactured.
A human who can see the queue
An order stalled past a threshold surfaces to a dispatcher with the context to act — re-offer wider, call the branch, warn the guest honestly — because an escalation nobody sees is not an escalation.
The cost
The shelf is where a good kitchen loses its reputation
Hot food does not wait; it degrades in the exact qualities the guest ordered it for. Fried items soften in their own steam, grilled meat keeps cooking past where the chef stopped it, sauces split, cold items warm. The guest cannot see the shelf, so the verdict lands on the kitchen and the brand: the review says the food arrived tired, not that assignment stalled for eleven minutes. This is what makes the waiting shelf uniquely expensive — it converts good cooking into bad reviews at a rate no recipe change can compensate for.
The deeper fix is to stop treating the kitchen and the fleet as independent clocks. A kitchen that fires every order the moment it lands will always manufacture shelf time when couriers are scarce; a kitchen that fires on assignment makes the courier wait instead, which rented fleets punish. The workable middle is firing tuned to each branch's actual pace: the order is fired so that its predicted ready moment meets the predicted courier arrival, using the branch's own history for both predictions rather than a chain-wide constant. Getting that pairing roughly right, hour by hour, removes more perceived lateness than any speed gained on the road.
None of this is visible without honest measurement, and the honest metric is ready-to-pickup: the timestamped gap between the kitchen marking an order ready and a courier collecting it, kept separately from travel time. Operators who track one blended delivery time invariably conclude that the roads are slow; operators who split the metric usually find the road was never the biggest term.
Two fleets
What a rented fleet decides for you and what your own fleet obeys
Assignment behaves differently depending on whose couriers carry the order, and so does what an operator can fix. The comparison is about control, not about which is better.
| Option | Who controls the assignment moment | What you see when it stalls | How it tunes to kitchen pace | Where the fix lives | Best for |
|---|---|---|---|---|---|
| Aggregator couriers | The platform's engine, optimising across every merchant in the zone — your order is one entry in its queue. | A status field and a support channel; the queue itself is not yours to inspect. | Indirectly — accurate prep times and a disciplined ready signal are the levers the platform actually reads. | In what you feed the platform, and in measuring its pickup performance well enough to escalate with evidence. | Reach and demand you could not recruit alone, carried at the platform's rules. |
| An own fleet | Your dispatch rules — zones, batching, offer order — written for your branches rather than a whole city. | The whole queue, every courier's state, and every stalled order — visible because the system is yours. | Directly — firing and assignment share one clock, tuned per branch and per hour. | In your own rules and your own dispatcher — with the cost of running both now on your side of the ledger. | Dense delivery zones around your own branches, where control repays the operating cost. |
| Both, deliberately | Split by rule — own fleet for direct orders in home zones, platform couriers for the long tail. | One board reading both, so a stall is seen the same way whichever fleet owns it. | Own zones tune tightly; platform zones tune through prep-time honesty. | In the routing rule that decides which fleet gets which order — the decision most hybrids never write down. | Operators whose direct channel has grown enough to justify owning its best zones. |
Our role
Assignment machinery we tune and the promise the brand keeps
- 01
Honest instrumentation first
We timestamp every stage from ready to collected, per branch and hour, split from travel time — because until the wait is measured where it happens, every fix is aimed by rumour.
- 02
Firing and assignment on one clock
We tune when orders fire against when couriers actually arrive, per branch, so meals and couriers reach the pass together — on own fleets through the rules, on rented ones through the signals the platforms read.
- 03
A dispatcher view that earns its screen
For own-fleet operations we build the board that shows stalled assignments while there is still time to act, ranked by how late the guest is about to be rather than by order number.
What delivery time the brand promises, and what it costs to keep, are commercial choices; we make the minutes visible and the machinery tunable, so the promise is set from evidence rather than from hope.
Common questions
What buyers ask
Why does food sit ready while no courier is assigned?+
Usually for one of four reasons, and rarely the one first blamed. Batching logic may be holding the order to pair it with another from the same area; the branch may sit at the edge of a delivery zone whose eligible couriers are drawn from the wrong side; the courier state the engine assigns from may be stale, so offers go to riders who are marked free but are not; or the kitchen never emitted a ready signal, so the engine does not know the food is waiting at all. Each cause leaves a different fingerprint in the timestamps, which is why the first step is instrumenting ready-to-pickup per branch and hour — the pattern usually identifies the culprit within a week of honest data.
Can kitchen firing be timed so the courier and the food arrive together?+
Yes, and it is a tuning problem rather than a software purchase. Each branch has a real preparation pace per daypart, and each fleet has a real arrival pattern; firing so the predicted ready moment meets the predicted courier arrival is arithmetic once both are measured. On an own fleet the pairing is enforced directly in the dispatch rules. With aggregator couriers the lever is indirect but real: the prep times you declare and the discipline of your ready signal are what their engines schedule from, so a branch that declares honest, per-daypart prep times gets couriers that arrive when the food does far more often than a branch running one optimistic constant. What cannot be tuned away is a genuine courier shortage at peak — that shows up in the data as offers expiring, and it is a capacity conversation, not a timing one.
How should ready-to-pickup time be measured so nobody can argue with it?+
From events, separately from travel, and against the branch's own baseline. The clock starts when the kitchen marks the order ready — which means the ready action has to be a real, one-tap habit at the pass, not a step staff backfill later — and stops when the courier scans or confirms collection. Keep it apart from travel time, because blending them lets each side blame the other; report it per branch, per hour and per fleet, because an average across all three hides exactly the pattern you need; and compare each branch to its own history rather than to a sister branch with a different zone and a different menu. Measured that way, the number stops being an accusation and becomes a shared instrument — the kitchen, the fleet and the dispatcher are all reading the same clock.
In this cluster
Instrument the wait with us
A review of your ready-to-pickup numbers — or the instrumentation to produce them — that names where the waiting minutes actually accumulate.
Start the conversation
