Bring four things to a delivery partner review: a count of dispatch requests and their outcomes, a breakdown of delays and failed dispatches, a reconciliation of charges against your written terms, and a short list of decisions you need from the partner. Use the same dates and order identifiers on both sides so you can resolve differences in the meeting.
The aim is to improve an existing service, not renegotiate every part of the relationship at once. A recurring monthly review may suit a busy restaurant; a quieter operation might choose a different interval. Either way, send the summary ahead of time and keep individual order records available for investigation.
Reconcile dispatches and completed handoffs
Start with the restaurant's order log and the delivery partner's dispatch report for the same period. Ask each side to supply order or dispatch IDs, timestamps, status changes, cancellations, and charges. Your manager can add notes about food readiness and handoff problems. The partner should identify who can explain courier assignment and delivery outcomes.
Agree on what each status means. A dispatch request is not necessarily an accepted job, and a courier picking up food is not the same as the guest receiving it. If the partner reports a completed delivery, decide what record supports that status and how the restaurant can investigate a disputed one.
Match records by ID rather than by dollar amount or guest name. Check for duplicate requests, restaurant cancellations, partner cancellations, and orders that appear on only one report. Keep unresolved records in their own category until both sides agree on the outcome. Avoid putting guest addresses or payment details in the meeting summary; use order IDs to look up the details when needed.
Hypothetical service example: A restaurant sends 100 dispatch requests in a month. The matched reports show 90 completed deliveries, six canceled requests, and four unresolved records. The completion rate is 90% of requests. The six cancellations still need a reason and owner; the four unresolved records are not automatically failures or completed deliveries.
Use a small sheet like the one below at each review. The figures belong to this example, not to a provider benchmark. The owner column identifies who should bring or explain the evidence, not necessarily who caused a problem.
| Measure | Example result | Evidence or question | Review owner |
|---|---|---|---|
| Dispatch outcomes | 90 completed, 6 canceled, 4 unresolved | Match all 100 requests by ID | Restaurant and partner |
| Late completed deliveries | 18 of 90, or 20% | Compare actual arrival with the agreed promised time | Partner, with restaurant records |
| Cancellation reasons | Six require classification | Was the request canceled by the restaurant, partner, or guest? | Both sides |
| Charge exceptions | Two charges to investigate | Match invoice lines to order IDs and written terms | Restaurant manager and partner billing contact |
| Open actions | Four unresolved records | Assign an owner and a date for each answer | Both sides |
Record the denominator beside every rate. Here, late deliveries are measured against completed deliveries, while completion is measured against dispatch requests. Those figures answer different questions and should not be compared as if they cover the same set of orders.
Review delays and failed dispatch reasons
Before discussing whether service was late, agree on the promised time being measured. Was it the time shown to the guest, a pickup target, or a delivery estimate updated after the order was placed? Save the relevant timestamp when possible. A changing estimate can make a final report look better while a guest has waited longer than expected.
Break the trip into stages: order accepted, food ready, courier assigned, courier arrival, pickup, and guest handoff. Ask the partner which timestamps it can provide and how each is recorded. Your staff may need to record when food is ready separately. If either side lacks a timestamp, mark that stage as unknown rather than assigning blame.
In the example, 18 of 90 completed deliveries arrived after the agreed promised time. That rate alone does not say why. Pick a few late orders from different days and compare their timelines. Was the food ready before a courier arrived? Did assignment take longer during a particular shift? Was the address difficult to reach? A pattern across several orders is more useful than arguing from the worst single case.
Classify failed or canceled requests separately from late completed deliveries. Useful reason groups include no courier assigned, restaurant cancellation, guest cancellation, address problem, and reason unknown. Ask who recorded the reason and whether it can be corrected when the records disagree. For the six cancellations in the example, the first action is to classify all six, not assume that the partner caused them.
Discuss trade-offs before choosing a fix. Earlier dispatch might reduce courier waiting after food is ready, but it could also bring couriers to the counter before the kitchen can hand over an order. A wider delivery area may reach more guests while making arrival times harder to predict. Agree on a practical change, the shifts it covers, and the measure that will show whether it helped.
If staff need to practice a new workflow, use a simulated order that cannot trigger a real payment, guest message, or courier dispatch. A live service pilot is different: agree on its scope and monitoring with the partner before using real orders.
Compare charges with the written contract
Bring the current agreement, amendments, rate schedule, and invoice for the same period as the dispatch report. Match invoice lines to order IDs where possible. Check the agreed base or minimum charge, per-dispatch fees, cancellation fees, adjustments, credits, and any charges tied to distance or other conditions in your terms.
Keep the service review focused on operational quote totals and actual invoice lines. A quoted delivery fee is not necessarily the all-in cost of taking an order: software, payment processing, packaging, and other expenses may sit elsewhere. Ask a qualified adviser about contract or tax treatment when that is relevant to a dispute.
Separate hypothetical charge example: Assume written terms call for a $100 monthly fixed charge and $6 for each of 90 completed deliveries. Also assume two valid exception charges of $10 each. The total is $100 + $540 + $20 = $660. This includes the fixed charge, the stated variable delivery fees, and those two exception costs. It excludes payment processing, packaging, tax, and any other charges not listed in the example.
If guests paid $180 in delivery charges, the restaurant's delivery cost after that offset would be $480, assuming the full $180 is retained by the restaurant. The $180 remains a customer cost; it is not a reduction in the partner's $660 invoice. Guest charges, restaurant sales, and payment-processing fees are different measures. Do not count the same guest charge as both an offset and additional savings.
For a disputed line, note the order ID, billed amount, contract term, and the question to answer. Ask when the partner will respond and whether an approved credit will appear on the current or a later invoice. If the agreement does not clearly cover an exception, request a written explanation rather than assuming the charge is valid or invalid.
Agree on actions and owners
Send the metrics sheet and a short agenda before the meeting. That gives the partner time to find order records and bring someone who can answer billing or dispatch questions. Keep the meeting to the decisions that need both sides.
- Confirm the period and counts. Resolve mismatched IDs and agree which records remain open.
- Review service patterns. Compare late deliveries and cancellations by shift and reason, using a few order timelines.
- Resolve invoice questions. Review exceptions against the written terms and assign follow-up for disputed lines.
- Choose changes. State what will change, who will do it, when it starts, and how it will be measured.
- Set the next review. Agree on the reports and definitions both sides will use next time.
Turn each decision into a specific action. For example: the partner investigates the four unresolved dispatch records by the following review; the restaurant records food-ready times during the next two Friday dinner shifts; both sides then compare those timestamps with courier arrival times. This is more useful than an action item that simply says “improve delivery speed.”
Decide what result would lead to another change. If the shared records show long waits before courier assignment, ask the partner for a response focused on assignment. If food is routinely not ready when couriers arrive, review kitchen timing and dispatch timing together. If records are incomplete, fix the measurement first. Keep the decision tied to evidence rather than to a promised percentage improvement.
Direct orders can make it especially important to know which part of the guest experience the restaurant controls and which part belongs to the delivery partner. Review your restaurant website and ordering flow alongside the dispatch handoff, and use the online ordering buying guide if you need to revisit how orders enter the business. The guide to comparing direct and marketplace order contribution can help separate channel economics from courier service performance.
For your next meeting, complete the sheet for one recent period and attach three representative order IDs: a normal delivery, a late delivery, and an unresolved or disputed order. Use that checklist to evaluate your current setup. If you are considering changes to direct ordering, ask Fleksa about restaurant online ordering and verified availability in your US location.
Frequently asked questions
How often should a restaurant review its delivery partner?
Choose a cadence that gives you enough orders to spot patterns and enough time to act. Monthly may work for a busy operation; a lower-volume restaurant can use a longer period. Review serious recurring problems sooner.
Which delivery metric should an owner check first?
Start by reconciling dispatch requests with completed, canceled, and unresolved outcomes. Once the counts match, examine late deliveries and reasons. A rate is less useful when the two reports cover different orders.
What if the partner and restaurant disagree about a delay?
Compare the order ID, promised time, food-ready record, courier timestamps, and status history. Mark missing evidence as unknown, assign someone to investigate, and agree on a response date rather than assigning fault without a shared timeline.
