To estimate missed direct orders after an outage, start with the number you would normally expect during the affected period. Subtract direct orders that actually went through, then subtract orders you can reasonably attribute to guests switching to another channel. Use a low and high estimate rather than treating the result as an exact count.
The difficult part is not the subtraction. It is deciding which orders were genuinely interrupted, which guests found another way to buy, and which would not have ordered that day anyway. A short worksheet makes those assumptions visible and gives you a useful basis for comparing future incidents.
1. Record the outage window and affected ordering modes
Write down when guests first could not complete an order and when ordering was working again. If the problem affected only part of the process, record that too. Guests might have been able to browse the menu but not submit an order, or pickup might have worked while delivery did not.
Separate the guest-facing window from the time staff spent investigating or correcting records afterward. For estimating demand, the relevant period is when a guest who wanted to order faced the problem. If the start or end is uncertain, record the earliest and latest plausible times and run the estimate for both windows.
Ask the manager on duty for notes on when staff first noticed the problem and when a successful order appeared again. Compare those notes with order timestamps and any incident information available from your provider. An order timestamp alone may not prove that every ordering mode was working, so note what each record actually shows.
Next, identify the affected paths. These might include ordering through the restaurant's website, pickup, delivery, or a particular device or payment step. Do not automatically include phone, walk-in, or marketplace orders in the affected total. They may have remained available and may help explain where some demand went.
If only delivery failed, use comparable direct delivery orders as the baseline. Including unaffected pickup orders would inflate the estimate. When records cannot separate the modes, say so and make the range wider.
2. Build a baseline from comparable periods
The baseline is your estimate of direct orders that would have been placed during the outage window if ordering had worked normally. Pull several comparable periods where possible: the same day of the week, the same hours, and the same ordering modes.
For example, compare a Friday dinner interruption with other Friday dinner periods rather than an all-day average. Look at the order count as well as the time distribution. A two-hour lunch outage should not inherit a full day's expected orders.
Review unusual conditions before choosing the range. A promotion, local event, holiday, severe weather, staffing limit, or menu change can make one comparison less useful. If a comparable Friday had a special offer that the outage Friday did not, do not simply assume its higher volume would have repeated. Document the adjustment or leave that period out and explain why.
Use actual order records where you have them. Your ordering system may show completed direct orders; the POS may help identify phone and walk-in sales; marketplace reports may show orders placed there. Ask your provider what timestamps and channel details can be exported, and whether failed attempts appear in reports. A failed attempt is not necessarily a lost order: one guest might retry several times or eventually order by phone.
Choose a low and high expected order count that reflects the comparable periods and any known differences. A range is especially useful when you have only a few clean comparisons. If you lack reliable history, ask the manager for a cautious estimate, label it as judgment, and avoid presenting it as a measured result.
3. Subtract completed and shifted orders
First subtract direct orders completed during the affected window. Some orders may have arrived before a partial failure, after a partial recovery, or through an unaffected route. Count each completed order once, and check that it belongs to the same modes included in your baseline.
Then estimate how many affected guests bought through another channel instead. Compare phone, walk-in, and marketplace counts with similar periods. A higher count can be a clue, but it is not proof that every extra order came from an interrupted direct guest. A busy event or promotion could also raise those counts.
For a stronger estimate, use records that indicate a switch, such as a staff note that a caller could not complete an online order. Where there is no direct evidence, use a conservative range and explain the judgment. Do not count the same guest's failed attempts as multiple shifted orders.
Also look for orders placed after service returned. Some may represent delayed demand rather than lost demand, but an unusually busy later period can have other causes. Include a recovery estimate only when there is a reasonable basis, and do not subtract an order already counted as a phone or marketplace switch.
A switched order can preserve a sale while changing the economics. Marketplace fees or additional staff time may make it less valuable than the expected direct order. Keep that channel-cost difference separate from the count of missed orders. The guide to comparing direct and marketplace contribution can help with that second calculation.
4. Use the outage impact worksheet
The following example is hypothetical. A restaurant's direct delivery ordering was impaired for a dinner window. Pickup was unaffected and is excluded. Comparable periods suggest a range of 36 to 44 direct delivery orders. Twelve direct delivery orders were completed during the window, and the owner estimates that 4 to 6 affected orders shifted to other channels. No delayed orders are counted because the owner cannot distinguish them from normal later demand.
For the low missed-order estimate, use the low baseline and high shift estimate. For the high estimate, use the high baseline and low shift estimate. The table shows both calculations.
| Worksheet item | Low missed estimate | High missed estimate |
|---|---|---|
| Expected direct delivery orders | 36 | 44 |
| Less completed direct delivery orders | 12 | 12 |
| Less estimated orders shifted elsewhere | 6 | 4 |
| Estimated missed orders | 18 | 28 |
| Illustrative missed gross sales at $25 per order | $450 | $700 |
| Illustrative missed contribution at $10 per order | $180 | $280 |
The missed-order range is 18 to 28: 36 minus 12 minus 6, and 44 minus 12 minus 4. At an assumed $25 average order value, that represents $450 to $700 in possible gross sales. Gross sales are not profit and do not account for the cost of fulfilling an order.
The separate contribution estimate assumes each missed direct order would have left $10 after variable food, packaging, payment, and delivery costs. That gives $180 to $280. It excludes fixed costs such as rent and subscriptions, as well as exception costs such as overtime, refunds, or incident-related credits. Add those exception costs separately if you want a broader estimate of the incident's impact; do not add gross sales and contribution together.
The $10 assumption needs to fit your own orders. If delivery costs vary sharply or guests pay a delivery charge, calculate an appropriate contribution using your records rather than borrowing the example. If the outage also caused extra marketplace fees on shifted orders, estimate those separately against the direct-order alternative.
Use the result to decide what deserves attention. In this example, even the low estimate suggests meaningful missed demand, so the owner could review whether guests have a clearly communicated alternative when direct delivery is unavailable. If the range were wide mainly because channel records were weak, the first improvement might instead be better staff notes and clearer order-source reporting.
Keep the worksheet with the incident date, outage window, comparison periods, assumptions, and the person who prepared it. Revisit the estimate if better records appear. It is a decision aid, not a count of identifiable guests who definitely would have ordered.
Use this checklist to review your current setup: can you isolate affected modes, export order timestamps, and compare channel totals for the same hours? If you are considering a change, review the online ordering buying guide and your restaurant website ordering path. Then ask Fleksa about restaurant online ordering, reporting options, and verified availability in your US location.
Frequently asked questions
Are all failed online attempts missed orders?
No. A guest may retry, complete an order later, or switch to another channel. Several failed attempts may also come from one person. Use attempts as supporting evidence, not as a direct lost-order count.
How many comparison periods should I use?
Use several genuinely comparable periods when available, matching the day, hours, and affected ordering modes. A few relevant periods are more useful than many that include different promotions or service conditions. Widen the range when the history is thin.
Should shifted marketplace orders be counted as lost sales?
Not if the guest completed a purchase there and you have already counted it as a shift. The restaurant may still face a different fee or contribution on that order, which should be estimated separately from missed sales.
