Black Friday customer support: a staffing and returns plan

Your Black Friday plan has enough agent hours on paper. By mid-afternoon, chat is backed up, nobody can approve an address change before dispatch, and the refund owner is off shift. The total headcount was only part of the plan.

Black Friday–Cyber Monday customer support planning should connect the sales forecast to human work, schedule people for when that work arrives and make sure decisions can be made across support, fulfillment and finance. It also needs to cover the delivery and return waves after the promotion ends.

A recent Shopify Community preparation discussion raises two issues: testing promotions, and handling address changes before parcels leave. The thread includes service-provider perspectives, not verified industry statistics. Use its questions to check your own operation rather than copying its assumptions.

What data should drive the staffing forecast?

Use your own order volume, human case volume, arrival times and handling effort. Distinguish a case from a message: one customer issue can produce several replies. Decide whether follow-up work is included in the case’s handling time so it does not disappear or get counted twice.

Break out work that does not track orders closely, such as presale compatibility questions or an existing technical backlog. Segment by channel, language and complexity where those differences change staffing.

Gorgias describes a forecasting approach based on orders, contact volume and actual team capacity. The useful principle is linking demand to capacity; its example benchmarks are no substitute for measuring your own products and team.

Base the forecast on both a recent ordinary period and a relevant peak. A new product, shipping region or promotion can change the case mix, so show a range of assumptions instead of treating last year’s ratio as guaranteed.

A worked example: orders to handling hours

For order-related work, start with forecast orders × human cases per order + existing cases to clear. Multiply by active handling minutes per case and divide by 60 to get handling hours. Add any separately forecast work, such as presale questions, if it falls outside that case ratio.

All numbers below are hypothetical planning inputs, not industry benchmarks, CXharbor results or staffing recommendations. The example estimates human handling work across a four-day promotion.

Input or calculation	Illustrative value
Forecast orders	1,400
Human cases per order	0.30
New human cases	1,400 × 0.30 = 420
Existing cases to clear in this period	60
Total cases to handle	480
Total active handling time per case	12 minutes
Required active handling hours	480 × 12 ÷ 60 = 96

The 0.30 assumption counts work that still needs a human after self-service has genuinely been completed. The 12 minutes covers reading, replies, notes and related follow-up, but not passive time spent waiting for a customer. If your reporting counts these differently, adjust the inputs before using the calculation.

Measure presale demand separately if it is not represented in the ratio. Likewise, identify return work likely to arrive after these four days; do not assume it has been staffed just because the original order is in this forecast.

How do handling hours become shifts?

An eight-hour shift does not provide eight uninterrupted hours of case handling. Allow for breaks, training, meetings and necessary availability between contacts.

For illustration, assume 25% of each shift is unavailable for queue work and plan active handling for 80% of the remaining time. These are chosen assumptions, not recommended industry targets:

8 hours × 0.75 × 0.80 = 4.8 planned handling hours per shift.

The example needs 96 ÷ 4.8 = 20 shifts, averaging five shifts per day across four days. That is a workload estimate. It does not show that five people can cover every hour, language or live-channel service target.

Check sensitivity. If the human-case ratio becomes 0.40 with other assumptions unchanged, the plan contains 620 cases, requires 124 handling hours and rounds up to 26 shifts. A fairly small change in one assumption can change the roster significantly.

Map the shifts onto actual arrival times. Review language coverage, product skills, breaks and supervisor availability. Phone and live chat need extra attention to simultaneous contacts and wait times; a daily average cannot tell you what service level you will reach.

Which decisions need coverage as well as agents?

A queue can stall even when frontline capacity is available. Name the decision owner, executor and backup for the events most likely to cross team boundaries.

Event	Decision or handoff to prepare
Promotion behaves differently from the published offer	Who confirms the intended terms and authorizes correction
Address-change request before dispatch	Who checks whether fulfillment can still stop or edit the shipment
Promised shipment cannot be made	Who approves the applicable customer options and communication
Returned hardware needs inspection	Who receives it and supplies findings to the case owner
Refund exception or dispute needs action	Who decides and who executes within the relevant deadline

Have each owner state when they are available and how to reach a replacement. Give the customer an update time the team can actually control, not an invented delivery or approval promise.

For covered US orders, the FTC’s Mail, Internet, or Telephone Order Merchandise Rule sets the shipping obligations. You need a reasonable basis for any shipping time you advertise; if you state none, you must ship within 30 days (50 days if the buyer applies for credit to pay for the order). Once you learn you cannot ship on time, and before that deadline passes, you must offer the buyer a choice: consent to the delay, or cancel and receive a prompt refund.

What happens if the buyer does not reply depends on the new date. According to the FTC’s business guide to the rule, silence counts as consent only to a definite delay of 30 days or less. For a longer or indefinite delay, you need the buyer’s express consent; without it, you must cancel and refund any order you have not shipped within 30 days of the original deadline. Refunds are due within seven working days, or within one billing cycle if you extended the credit yourself. “Shipped” means handed to the carrier, so a carrier delay after dispatch does not trigger the same notice.

When should Black Friday support planning start?

Start from the first date customers can see the offer, which may be an early-access email rather than Black Friday itself, and count backward. List the jobs that must be finished by then:

1. Pull last year’s peak data and build this year’s forecast range.
2. Agree the roster, languages and decision owners with each team lead.
3. Select and contract a provider or temporary agents, if you need them.
4. Update the knowledge pack: offer terms, shipping cut-offs and the return route.
5. Train the team, then rehearse complete cases with the people who will work the queue.
6. Fix what the rehearsal exposes and run those cases again.

Put a duration beside each job from your own records, not from a template: how long last year’s hiring took, the onboarding lead time a provider quotes in writing, how long finance needs to approve refund wording. Add up the jobs that must happen in sequence, run the rest alongside them, and the result is your latest safe start date. Leave room for one round of rework in case the rehearsal exposes a gap.

If that date has already passed, reduce scope rather than skipping rehearsal. Covering fewer channels or languages properly is safer than launching all of them untested; a universal number of weeks cannot account for an untrained team or an unresolved return route.

What should you test before the promotion?

Walk through the actual offer, its product restrictions and customer messages. Have support explain a failed discount, an out-of-stock choice, a cancellation request and a delivery question using approved information.

Check whether the person changing fulfillment can act before the parcel leaves. Check whether the return instructions send an item to a location that can actually process it. These are useful rehearsals even if you have no new software to launch.

Use the outsourcing onboarding checklist for temporary or external agents. Confirm what they have been trained to handle before assigning them to a peak queue. Put the AI handoff conditions through the same rehearsal if automation is involved.

Plan the return wave before reducing coverage

Forecast post-promotion delivery inquiries, setup questions and returns as well. Many of those delivery inquiries can be prevented with proactive order-status updates. Separate each platform’s deadlines from your internal targets. Our guides to TikTok Shop return clocks and Shopify return settings explain why a single generic refund timer is not enough.

During the event, compare arrivals with the forecast, and track unresolved waits, missed commitments and cases blocked on another team. Adjust the roster and decision coverage together. Faster acknowledgment alone will not clear a stalled inspection queue.

Afterward, compare the original assumptions with actual case mix, handling time and repeat work. Keep the version of the forecast you used so the next campaign starts from what happened, rather than another guess at how many agents feel sufficient.

If you need external peak coverage, discuss the workload with CXharbor using your promotion dates, forecast range, channels and coverage gaps. Include the return period in the same brief.