At 11:45 a.m. Eastern, Riley is about to confirm a noon gift invitation for employees in three U.S. offices. Four support agents are on the schedule. A case left over from the morning test still needs an answer, and the only person who can verify a shipment is handing over the queue later that day. Riley has fifteen minutes to decide whether to send the whole wave.
Riley and the company are fictional. In the staffing workbook, Riley will enter the work already waiting, the new cases expected after the send, and the minutes the four agents can actually spend. The result gives the sponsor a choice to make while there is still time to change the schedule.
Start with Riley's noon queue
Riley opens the workbook at the desk rather than counting the names on the schedule. The question is how much work those four people can finish in the hour after the send. The answer also depends on which cases they are allowed to handle.
The workbook estimates work in minutes. It starts with the unresolved work carried into a window and adds the expected handling time of new cases. It then divides that workload by the number of minutes one assigned person can realistically devote to cases during the same window. The result rounds up to a whole number of people. It does not forecast the minute an individual will receive a reply. A window with enough work capacity can still have a long queue if contacts arrive within its first five minutes, if a particular skill is scarce, or if a carrier answer is required before a case can move.
This is useful before sending invitations, opening a seasonal choice program, or publishing a sponsor update likely to draw questions. The program owner already has a send plan. The support owner brings recent case volumes and handling times, and someone with scheduling authority confirms the people who will actually be available. If the model identifies a gap, the sponsor can reserve named people, lower the size of the wave, or adjust when it is released. The model is most useful when the team has time to make one of those decisions.
Workforce forecasting commonly uses incoming volume and average handling time. Microsoft's workforce-management overview describes those measures in the context of contact center forecasting. The workbook here uses its own workload arithmetic and synthetic examples, with assumptions intended for the program owner to replace. Microsoft workforce-management forecast overview
Version 1, dated November 7, 2026, provides editable examples and blank rows in the US English edition. Use the sample values to check how the workbook behaves, then replace them with the cases, queue owners, and time windows of your own program. Leave the sample off any operational dashboard that could be mistaken for live planning evidence.
Download the gift campaign support staffing workbook
Which calls would count as new work after noon?
The morning test left one partially researched case. Riley asks the agent how many minutes are left, then checks what the same employee sent through chat. That second message may add work even if it does not open a second case.
Pick one queue and one reporting window per row. A 60-minute planning window can be useful for a steady async inbox. A 15- or 30-minute window may better expose a burst after a bulk invitation. Record the time zone along with the queue label, especially when an Eastern team releases a campaign that a Mountain or Pacific team will support. Compare the scheduled start and end in each person's working time. A plan with a date but no time zone can shift the only experienced staff away from the actual launch.
Define a case before counting it. If the recipient's email and web chat share one case history, the underlying issue counts once while the time spent reading both messages belongs in the handling estimate. If the channels create two separate work items and the team must manually reconcile them, measure the additional workload as it actually occurs. At the next data review, determine whether the process can merge those items earlier. A raw total of messages can overstate new cases in one campaign and understate work in another because it misses internal review.
Estimate handling time from a comparable slice. Include the first read, research, replies, ordinary wrap-up, and record updates when the assigned person performs them. A three-minute first reply may be followed by ten minutes checking an invitation, confirming the correct destination, or escalating a mismatch. Use observed handling time for that case type if you can. Time waiting for a carrier or approver does not necessarily occupy an agent's hands, even when the case remains open. Model the follow-up work in the interval where someone is likely to do it.
Carryover means remaining minutes of work when the window opens. Suppose an agent already performed most of the work on a case and needs two minutes to send the next approved message. The case contributes two minutes to carryover. Another case blocked on an outside response may require no immediate work until a due check. Keep the open case visible and give it an owner even if no handling minutes fall in the next interval. Aging, escalation, and workload answer different scheduling questions.
For a new program without historical data, use a bounded scenario and mark the contact-rate assumptions as estimates. Start from eligible invitations, identify which are likely to be acted on in the window, and use a plausible contact rate based on a pilot or a comparable release. Check whether the release will contain a new kind of choice, delivery path, or recipient group. Treat a small sample from one site as a starting point, not a national benchmark. Keep a low, working, and stressed view with named assumptions. Change the volume or handling mix explicitly between them.
Turn the four names on the roster into work minutes
| Input or result | Calculation or definition | What to check |
|---|---|---|
| Interval length | Clock minutes in the window | Time zone and overlap with staff hours |
| Carryover | Work minutes remaining from earlier cases | Cases requiring action now |
| New cases and handling time | Expected count and average minutes per new case | Count rule, sample and case mix |
| Demand | Carryover + new cases × handling time | Time units match |
| Shrinkage | Share of paid window unavailable for casework | Breaks, training, meetings and coordination |
| Occupancy limit | Share of remaining time planned for work | Buffer within the interval |
| Effective minutes per person | Window × (1 − shrinkage) × occupancy | Nonzero and feasible |
| Staff needed | Demand ÷ effective minutes per person, rounded up | Same queue and eligible skill group |
| Gap minutes | Demand − scheduled staff × effective minutes per person, floored at zero | Residual work under current staffing |
The workbook uses workload minutes for comparisons. The highest headcount and largest work-minute gap can occur in different rows.
Shrinkage and occupancy represent different deductions. A paid hour may include coaching, breaks, and scheduled meetings. Those activities reduce the time available for the queue. An occupancy limit then reserves some of the remaining time for variation and interruptions. If a break is deducted from the rostered window before entering it, do not also put the same break in shrinkage. Likewise, if all occupied minutes already include the usual wrap-up, avoid subtracting wrap-up as a non-case activity. State what the two rates cover in the planning note.
Only count staff who can perform the work in that window. An agent scheduled for general support may lack the permission to resolve a high-value order hold. A warehouse contact who will be loading trucks is not equivalent to a dedicated shipping specialist. Calculate distinct queues when work cannot be freely shared, then inspect overlaps. Never add two assignments for the same person during the same period. For a small part-time block, reduce the counted capacity by using a separate row for the interval the person can cover or split the interval at the schedule change.
The row has an input check. Enter a queue or window label and all seven numerical inputs. Zero new cases or scheduled staff are legitimate when known. An empty value calls for investigation and leaves that row without outputs. A zero occupancy limit also leaves the row uncalculated because the division would have no meaning. When you copy numbers into another planning system, keep that distinction. Showing zero as the demand of an incomplete queue can turn a missing estimate into an apparently healthy launch.
The first row leaves 39 minutes uncovered
Riley's first row covers noon to 1 p.m. Eastern. The team expects 22 new cases at eight minutes each, and the morning pilot left 16 minutes of work. Demand is 16 + 22 × 8 = 192 minutes. The schedule has four people. With 25% of the window set aside for breaks or other commitments and an 85% occupancy limit on the remainder, each supplies 60 × 0.75 × 0.85 = 38.25 effective minutes. Together they provide 153 minutes. That leaves 39 minutes of work. The modeled requirement rounds up to six because five people would provide 191.25 minutes.
This narrow gap between five people and the rounded answer makes the decision sensitive to the assumptions. A second review can check whether the carryover estimate includes tasks waiting for a decision that will happen later. Confirm whether all four scheduled people can investigate invitations and authorization. If a fifth agent joins only for the last ten minutes, they will add much less than a full person under the hour-long calculation. You can split the forecast into shorter windows and see whether the burst can be moved to when the extra agent is present.
The sponsor has several defensible options. It can reserve two people with the right permissions for the full hour. It can hold back part of the invitation wave, assuming the volume estimate actually scales for that group. It can release a clearly scoped wave to one office and watch the first 15 minutes before a second one. In each case, record the planned send time, the specific trained people or reduced cohort size, and the observation that will trigger the next decision. An extra name on a staffing sheet will not help if the agent is simultaneously listed for another queue.
Suppose two unexpected exception types arrive right after sending. The support lead can separate the issues, assign the fraud or policy-sensitive ones to the authorized queue, and keep the regular inquiries with the planned team. Check whether the change affects the time to respond to ordinary cases. The sponsor can delay the next wave before it goes out. A pause decision made against observed demand is easier to explain when the original plan already identifies the owner and next check.
Later in the day, a regional handoff has a different gap
Riley adds a second row for the regional handoff. This fictional hour starts with 35 minutes of carryover and adds 26 cases at nine minutes each. With the same 25% shrinkage and 85% occupancy assumptions, the six scheduled people supply 229.5 minutes against 269 minutes of work. The gap is 39.5 minutes, and the modeled requirement rounds up to eight people. This time the team must also check how many of the scheduled people can handle shipment changes.
Now suppose a subset of those cases concerns address changes after dispatch. Only a small group has access to the verified fulfillment record. The support manager should not assume the general pool can absorb those minutes. Split the workload by skill and track how much work comes from immediate triage and how much requires the specialist. If the specialist group has a different occupancy limit or can only work the first half of the period, model its window separately. Two people cannot be counted as separate headcount for both pools at the same time.
The operations lead could arrange staggered coverage so the agent who can verify orders remains available during the handoff. Another option is to send approved status updates for known delivery delays before recipients open new cases, then measure the actual change. Reserve a clear next step for cases that require carrier confirmation. Keep the cases visible even when waiting does not consume work minutes in the current window.
At the handoff, share the latest verified shipment or invitation state, the next action, and its due time. If a scheduled outbound notification slipped, update the current forecast using the observed send time. Do not leave the forecast peak under the previous hour simply because the schedule had been approved there. In this example, the sponsor could delay the final invitation wave until the owner sees whether the handoff holds. After the wave, compare predicted demand with received cases and actual remaining work, then adjust the next row with evidence.
Separate a busy inbox from a problem that staffing cannot fix
In Riley's launch chat, an agent asks if all the new cases require another pair of hands. If they are mostly reporting one broken link, the next person needed may be the owner who can repair the invitation. The timing of the cases and the kind of work each takes help answer that question.
An hour's total demand may look manageable while the first quarter-hour is not. Plot cases by arrival time if the data is available and check the earliest work an employee can do without a separate approval. A customer-facing deadline may require a more detailed queue model, response monitoring, or real staffing software. The simple workload estimate provides a first decision about whether the scheduled capacity is plausible. A team should separately verify its service target with observed waits, abandonment where relevant, and the portion of cases that genuinely met the response promise.
Forecasts are especially sensitive when past campaigns used different channels, countries, or products. A choice program launched through email may have a different distribution of questions than an on-site program with a QR sign. Initial signup questions and post-dispatch claims can have very different handling times. If case types are mixed, calculate a weighted mean using the expected number of cases in each type, or make separate rows. Averaging the means of three categories with very different case counts distorts workload. Keep the breakdown behind the forecast so that the sponsor can check why one wave is assigned more time than another.
A high contact rate may reveal an unclear invitation or a delivery configuration problem. Before adding shifts across the board, inspect the most common reason. If the first 30 contacts are asking where to find the choice deadline, updating an approved help notice may address the cause. If the contacts report an outage or inconsistent order records, route the issue to its responsible owner and hold the affected send. A lower anticipated volume after a change is a planning assumption until measured. Keep a baseline and a revised row and note which release decision used them.
The workbook also has a recovery use. If an agent is unexpectedly unavailable, update the assigned people and confirm whether the remaining staff can actually handle all case types. If a market opens late, move the relevant future windows rather than editing the past staffing record. Measure what remains at the end of the affected window. Give that work a next owner. The headcount requirement in the next window then reflects what is still to be done, rather than every case that ever entered the queue.
Give the sponsor a decision before the second wave
The sponsor needs more than the number six. Riley can provide the noon model, the actual staffing roster and permissions, and the exact point at which someone must delay the next send.
Review the file with the sponsor and the queue owner. They should be able to point to the release schedule, the source of the expected case estimate, the carryover check time, and each person with appropriate access. Ask what action follows if the first 15 minutes run above the stressed forecast. Ask who can authorize a delayed wave. Check whether any case type requires someone who will be in a different time zone, is on leave, or can see only a subset of orders. A revised number is useful only when it reaches the person who can act on it.
Try three scenarios against the workbook. Enter a quiet window with zero new cases and an existing carryover, and verify that it still shows the residual work. Enter zero staffed people with known work, and verify that all demand appears as the gap. Remove a required input, and check that the row stops reporting a staffing result. Confirm that the busiest row in the top summary matches the detailed calculation after changing one assumption. An estimate with clear failures is easier to use under pressure than a table where every row shows a number.
At launch, record actual counts at the chosen interval. The support lead can update the next row with the amount of work genuinely remaining. Separate late cases due to an outage from cases waiting for an authorized decision. Route sensitive inquiries through their own approved support channel and retain the amount of work without copying private case notes into the planning file. When the campaign settles, compare predicted and actual volume, handling time, and available staff. Keep the next forecast's assumptions grounded in those observations.
Before archiving the schedule, note who approved the wave size and which forecast version was used. Save the original inputs alongside any revision. If actual demand exceeded the estimate, the next owner can compare the released population, observed interval, and contact mix without asking the last shift to reconstruct it.
Set response ownership with the recipient support and escalation guide. Estimate warehouse capacity separately with the gift fulfillment capacity planner so packing labor stays out of the support calculation.
What Riley releases at noon
Riley cannot get another trained agent for the first hour. The sponsor holds a separable group expected to produce five of the 22 cases, keeping the other assumptions the same. The revised row has 17 new cases at eight minutes plus 16 minutes of carryover. Demand is 152 minutes against the four agents' 153 effective minutes. That is enough on the planning arithmetic with only a minute to spare. They check observed arrivals after fifteen minutes before committing the held group, and leave the published help route open for the people already invited.
The carrier inquiry remains with the original order at the regional handoff. Riley delays the held group until the specialist's replacement has confirmed access to shipment records and the support lead has checked the remaining workload. The sponsor can now show which invitations went out, the workbook version used, and who authorized the next send. Nobody needs to count the morning case twice to explain the schedule.
Using Giftpack within an approved gifting program
Once the sponsor has approved the gift choices, recipient scope, release calendar, and support ownership, it can consider Giftpack for the gifting work itself. Giftpack's help materials describe configurable choice sets, spending limits, country rules, and fallback options. Confirm which support duties the company, provider, and any fulfillment partner will each handle. The worksheet can model each confirmed queue and handoff from the organization's own staffing and case data. Giftpack corporate gifting and personalization