In the first months of direct ordering, measure completed orders, the money left after the costs of serving them, repeat buying, and the work required to keep service reliable. A growing order count is encouraging, but it does not tell you whether the channel is paying for itself or creating problems at the counter.
Set the definitions before launch. Decide which costs belong in the calculation, who will record them, and when you will review the results. Your goal is a scorecard that helps you make a decision: continue as planned, change the setup, or investigate a problem before spending more to promote it.
Start with completed orders and contribution
Count a completed direct order when the restaurant has fulfilled it and the sale remains valid after cancellations and refunds. Separate pickup from delivery if their costs or service problems differ. Do not count an abandoned checkout, a canceled ticket, or a test order as a completed sale.
For a first-pass contribution measure, start with completed-order sales and subtract the costs you choose to associate with the channel. Include food and packaging, order-related labor, payment processing, delivery expense paid by the restaurant, and the channel's fixed charges. Also include promotional and exception costs if you want the result to support an overall channel decision. Record guest delivery charges separately so you can see both what the restaurant pays and what the guest pays.
Use a consistent sales definition. For example, decide whether your sales figure includes guest delivery charges, and exclude amounts collected for others from the contribution calculation. Keep tips separate from restaurant sales. Ask your accountant how to treat any item you are unsure about.
Hypothetical monthly example: A restaurant completes 100 direct orders with $3,000 in included sales. The owner records $900 for food and packaging, $300 for order-related labor, $250 for restaurant-paid delivery, $100 for processing, $150 in fixed channel charges, $50 for refunds and other exceptions, and $200 for channel promotion. The costs total $1,950, leaving $1,050 in contribution, or $10.50 per completed order.
This example includes the listed fixed, variable, promotional and exception costs. It excludes other restaurant overhead, such as rent and existing management salaries, as well as any costs not listed. It is not profit for the whole restaurant. Delivery is included as a restaurant expense; no guest delivery charge is included in the $3,000 sales figure. Real costs and order mix will differ.
The owner or bookkeeper can supply fixed charges and processing statements. The kitchen manager can estimate food, packaging and added labor; the person reconciling delivery invoices can supply dispatch expense. Use actual records where possible and mark estimates. If delivery is paid by a guest, show that charge separately rather than treating delivery as free.
For a broader cost comparison, see the guide to comparing direct and marketplace contribution. Compare the same kinds of costs and periods; a direct order is not automatically an additional sale.
Separate finding guests from bringing them back
New customers and repeat customers answer different questions. A launch offer may bring in first-time buyers while producing little contribution. A returning guest may be cheaper to reach, but only if the restaurant can identify that customer consistently across orders.
Choose an identification method before counting. A customer record or contact detail may help match orders, but shared accounts, changed details and guest checkout can cause errors. Count unmatched orders separately instead of assuming every unmatched order belongs to a new customer.
For acquisition, track the number of identifiable first-time direct customers and the promotional spend associated with getting them. Divide that spend by the number of identifiable first-time customers for an approximate acquisition cost. This is a channel measure, not proof that the promotion caused every first order. Some guests may already know the restaurant or have ordered through another route.
For repeat buying, select a group of identifiable first-time direct customers from the same launch period. Then count how many place another completed direct order within a stated window, such as 60 days. Divide by the number in that group who have had the full 60 days to return. A customer who ordered only two weeks ago has not yet had the same opportunity and should wait for the next cohort review.
Consider the trade-off before extending a discount. A promotion that increases first orders but leaves little contribution may still be worth examining if enough guests return at a sustainable price. If repeat buying stays low, try a clearer menu or a different way to reach existing guests before increasing the offer. Do not assume a repeat direct customer is new to the restaurant overall.
Your restaurant website can give guests a clear path to order. Ask your provider what source and customer reports are available, whether guest checkouts can be matched, and how records can be exported for review.
Measure service quality and support work
An order that looks profitable on paper can absorb time if staff must correct addresses, call guests, reenter tickets or handle refunds. Track problems alongside sales, using completed orders as the denominator where appropriate. Keep a separate count of canceled orders so they do not disappear from the picture.
Define an exception as an order requiring a documented correction, refund, remake or manual intervention. Record the reason and whether the order was ultimately completed. One order can have more than one problem, so decide whether your rate counts affected orders or individual incidents. For an affected-order rate, count each completed order once, even if it had two issues.
Track support time as a separate estimate: minutes spent resolving direct-order issues during the period, divided by completed direct orders. This is not a replacement for the labor cost in your contribution calculation. It helps explain that cost and shows whether the channel is creating interruptions. Do not add the same labor expense twice.
During setup, run a simulation with a clearly marked practice ticket. Check that the menu, pickup instructions and staff handoff make sense. Configure the practice so it cannot trigger a real card payment, guest message or courier dispatch. A simulation checks the workflow; it does not show how the channel performs under a dinner rush.
Then review actual orders during a live service pilot using your normal operating controls. Have a manager note delays and corrections as they occur, and reconcile them after service. A small early sample can reveal a serious repeated problem, but it cannot establish a dependable long-term rate. Fix the cause, then watch the next period rather than treating one good shift as proof.
Use one scorecard with clear denominators
The following worksheet puts the measures in one place. Compare each row with the same prior period or a target you set for your restaurant. The denominator matters: a rate based on completed orders cannot be compared directly with one based on all submitted orders.
| Measure | Calculation | What it helps you decide |
|---|---|---|
| Completed orders | Fulfilled direct orders, excluding canceled and fully refunded orders | Whether usable channel volume is growing |
| Contribution per order | Included sales minus defined channel costs, divided by completed orders | Whether volume covers the costs in your model |
| Approximate acquisition cost | Channel promotion spend divided by identifiable first-time direct customers | Whether the spend is reasonable alongside contribution and repeat buying |
| 60-day repeat rate | Eligible first-time customers with another completed direct order within 60 days, divided by eligible first-time customers | Whether a launch cohort returns |
| Completed-order exception rate | Completed orders with at least one recorded exception divided by completed orders | Whether service problems are becoming common |
| Support minutes per order | Recorded issue-resolution minutes divided by completed orders | Whether direct ordering is adding staff work |
Add your own targets beside the worksheet. For example, an owner might decide that contribution must remain positive after the listed costs, while exceptions must be low enough for the shift manager to handle without delaying service. Those are operating choices, not universal benchmarks. A target should reflect your menu, staffing and delivery model.
Also record submitted orders and cancellations as separate counts. If cancellations are a concern, calculate canceled orders divided by submitted orders for the same period. Do not substitute checkout abandonment for the scorecard: it describes one step before an order is placed, not whether completed orders contribute money or guests return.
Schedule reviews and act on the pattern
Review service exceptions and cancellations weekly during launch, while staff can still remember what happened. Review contribution monthly after invoices and processing records are available. Review the first 60-day repeat cohort only after every customer in that cohort has had the full return window.
At each review, write down one decision and an owner. If order volume grows but contribution per order falls, check the mix of delivery orders, discounts and processing or dispatch expense before buying more promotion. If contribution is healthy but exception rates are high, prioritize the menu or handoff issue that generates the most corrections. If repeat data is incomplete, improve customer matching and wait for a mature cohort rather than guessing.
Keep a short note of changes to menu prices, promotions, hours and delivery coverage. These can all affect comparisons between periods. Small order counts, estimated labor and unmatched customer records limit what the numbers can tell you. Use them to ask better questions, not to claim that the channel alone caused a change in restaurant sales.
Before choosing or changing software, work through the online ordering buying guide and ask vendors how they report orders, refunds and customer cohorts. Use this scorecard to assess your current setup, then ask Fleksa about restaurant online ordering and verified US availability for your location.
Frequently asked questions
How many months should I track direct ordering?
Start at launch and review monthly costs as records arrive. Allow at least a full 60-day window before judging a 60-day repeat cohort, and continue tracking as order volume grows.
Is order count enough to judge the channel?
No. Count completed orders, but pair volume with contribution, repeat buying and service exceptions. More orders can also mean more delivery expense or staff work.
Should refunds be counted as a cost?
Yes, account for their effect consistently. Exclude fully refunded orders from completed-order sales and count any remaining refund or remake expense once. Reconcile the treatment with your bookkeeper so it is not deducted twice.
