Peak season planning

Forecasting · Scheduling · ~7 minute read

The known unknown of the planning year

Almost every contact centre has a peak season. Retail and delivery operations peak around Black Friday and Christmas. Tax-related operations peak around self-assessment deadlines. Insurance peaks around renewal cycles and weather events. Energy peaks around price-cap and tariff changes. Travel peaks around booking windows and disruption events. The peak is the most predictable hard moment of the year, and yet it is the moment most planning functions handle badly — not because the peak itself is unknown, but because the lead times required to staff it are longer than the operational rhythm of most planning teams. This article walks through how to plan peak well, working backwards from the peak itself through the recruitment, training, and operational decisions that need to be in place months in advance.

Forecasting the peak

The first job is the volume forecast. Peak forecasting differs from normal forecasting because the comparison points are scarce — one peak per year, with a few years of history at best. The risk of over-fitting last year is high, and the risk of missing a structural change in customer behaviour is real. Three habits help. First, decompose the peak into its components — a Christmas peak is not a single event but a cascade of pre-Christmas enquiries, in-week service spikes, and post-Christmas returns and complaints. Forecast each separately. Second, hold the peak against multi-year trend rather than year-on-year — the question is whether next year’s peak will be 8% higher than last year or 15% higher, not whether it will look like last year exactly. Third, scenario the upside and downside — produce a base case, a high case, and a low case, and design the operational plan around the base case with explicit triggers for moving toward the higher case.

Back-solving from the peak

Peak headcount on the floor on the busiest day is the output. Working backwards from that date, several deadlines fall into place. New-hire training typically requires six to eight weeks from start date to full productivity, depending on the role’s complexity. Recruitment from advert to start date typically requires four to eight weeks, longer if security clearance or specialist skills are needed. Schedule design and shift-pattern changes need to be agreed and communicated to existing staff at least four weeks before the new patterns take effect. Working back from a December peak through training, recruitment, and design, the planning function must have agreed the peak headcount requirement by late August or early September at the latest. Operations that try to plan peak in October consistently arrive at peak under-staffed because the lead times do not bend.

Temp vs. permanent

Most peaks justify some additional headcount. The decision is what type. Permanent hires deliver better quality, better retention, and more flexibility across the year, but cost more per FTE and carry the ongoing payroll commitment into the trough that follows peak. Temporary hires (agency staff, fixed-term contracts, seasonal workers) are cheaper, easier to flex, and easier to release post-peak, but typically deliver lower productivity through the peak itself and contribute more to quality and CSAT variability. The right mix depends on the shape of the peak: a short, sharp peak (a one-week post-Christmas spike) leans toward temporary; a sustained peak (a six-week pre-Christmas push) often leans toward permanent because the temporary worker’s ramp-up curve consumes too much of the peak itself.

A pattern that works for many operations is a layered approach: bring permanent hires in early enough that they are fully ramped by the start of peak, supplement with temporary staff for the in-peak spike, and use overtime among the existing workforce as the flex layer that absorbs day-to-day variation. The three layers stack rather than compete, and each one has different commercial terms.

Training schedule design

Peak training is more constrained than normal-period training because the existing workforce cannot afford to give up training time at the moment volume is highest. Most operations push training schedules earlier in the year — pre-peak rather than during peak — and accept that this requires a longer wait before benefits are seen on the floor. Refresher training for existing agents on peak-specific topics (returns processing, complaints handling, gift orders, holiday closures) is usually best delivered in the four to six weeks before peak starts, when volume is still manageable but the content is fresh in the agent’s mind. Training during peak is reserved for new hires who must be brought up regardless.

Real-time during peak

The real-time function carries more weight during peak than during normal operation. Volumes are higher, agent absence is more impactful, and the cost of a poor real-time decision compounds across many more agents. Three habits help. First, extend the real-time review cadence — intraday reviews every hour rather than every two hours during the heaviest weeks. Second, pre-agree the lever menu and the authorisation levels — the question of who can authorise emergency overtime should not be debated for the first time at 10am on Black Friday. Third, brief team leaders explicitly on the day-by-day expected pattern — many peak operations follow a known shape (Monday is the worst day, Tuesday recovers, Wednesday is moderate), and getting the team leaders aligned on the pattern before peak starts reduces the volume of reactive decisions.

Post-peak review

The most under-rated part of peak planning is the post-peak review, conducted in January or whenever the peak winds down. Most operations are exhausted by the end of peak and let the lessons slip away. The operations that build a planning function over time hold a structured review within two weeks of peak ending: what was forecast vs. actual, where did volume surprise, how did training quality play out, what did the temporary workforce deliver, what did real-time get right and wrong, what would we change next year. The output is a one- or two-page document filed somewhere the planner will find again in August next year. Over three or four cycles, the document becomes the single most valuable asset the planning function owns for peak preparation.

Common mistakes

Five patterns recur. Planning peak too late, missing the recruitment and training lead times. Forecasting last year’s peak rather than next year’s. Over-relying on temporary staff who cannot fully ramp in the peak window. Under-investing in pre-peak refresher training and paying the cost in AHT and quality during peak. And failing to capture the post-peak learning, so the same surprises happen again next year.

Conclusion

Peak season is the moment of the year when planning quality is most visible to the rest of the business. A well-planned peak makes the operation look composed at exactly the moment everyone outside expects it to be in crisis. A badly planned peak burns through agent goodwill, breaks SLAs, generates customer complaints, and reinforces every stereotype senior management already holds about the difficulty of the operation. The difference is mostly about lead times: starting early, working backwards, agreeing the temp-vs-permanent mix in good time, briefing the real-time function before the heaviest days, and taking the post-peak learning seriously. None of it is easy, but all of it is achievable, and the planners who do it consistently become the planners senior management trusts with everything else.

Pair this with the contact centre planning cycle for the layered rhythm peak planning sits inside, and the true cost of attrition for the workforce economics of seasonal hiring.