The Real Seasonal Challenge Isn’t Only Volume. It’s How Fast Your Workforce Can Flex.

So, it turns out I first wrote about this in 2016 here. Things have changed a little since then. With all these ‘changes’, I thought it was about time I revisited the subject, hopefully with some new insights from the last 10 years…

One thing that has not changed in those 10 years? Seasonal workforce management breaks down when the plan can’t move as fast as demand. Holidays, open enrollment, and summer travel arrive on known dates, yet forecasts, schedules, and hiring decisions run on weekly or monthly cycles. Demand moves by the day, or by the hour. CX leaders know about these spikes. They plan for them. But the plan still hits a wall when it meets reality.

One thing has changed, though: AI. It is changing how big the peaks get and how fast we can respond to them. The fix is a shorter loop between spotting a change and acting on it. See change early, decide quickly, and act faster.

Why is seasonal workforce management getting harder?

Because the cost of missing a peak keeps rising. Customers expect more each year, and they leave faster when service falls short.

Verint’s State of Customer Experience 2026 report surveyed 5,000 US consumers in early 2026. Three results stand out for anyone staffing a peak:

• 42% said their service expectations rose over the past year, up from 19% in 2024.
• 79% would switch to a competitor after a single bad experience.
• 78% rank the fastest resolution above using their preferred channel.

The same survey has a twist, though. 61% of consumers currently prefer speaking to a human agent, but 69% said they would switch to automated service if it fully resolved their issue. Among younger customers, that figure rises to 93%.

A peak week is when every gap in the plan turns into a wait. These numbers suggest customers will act on it, and that they care more about resolution than about who or what provides it.

What does history teach us about volume?

Volume doesn’t always return to where it started. After a peak, customers can stay on the channel they moved to, and the new level becomes the baseline.

At Conversocial, which is now part of Verint, we saw it happen in 2020. Across our retail partners, conversations spiked in March 2020, led by delivery questions, and a second wave hit over the holidays when lockdowns pushed shoppers into a “digital Christmas.” We expected the surge to flatten. It didn’t go back.

By September 2021, delivery conversations were still three times the level of the same month in 2019, and volume across the three biggest intents (orders, deliveries, and refunds) was three times higher than two years earlier, well before COVID-19 was a global event. The reason, as we saw it then: customers found private messaging more convenient, and it became their preference.

Obviously, COVID-19 was extreme, and I really hope we never see anything like it again. And I’m not suggesting it’s the model for your next seasonal plan. But it does show what a plan built on last year’s curve can’t absorb: a peak that raises the baseline instead of passing. And that was before AI started changing how customers shop and ask for help.

How will AI shopping change seasonal volume?

Probably upward, and concentrated on the same few peak days. I haven’t found a published number for contact volume yet, so treat this section as direction, not forecast (AI agents buying on customers’ behalf).

Here is what is measurable. Adobe Analytics reports that traffic to US retail sites from generative AI tools rose 693% year over year across the 2025 holiday season, and shoppers who arrived that way converted 31% better than other visitors. For this season, Adobe forecasts another 130% rise, with AI-referred traffic up 159% on Thanksgiving, 95% on Black Friday, and 88% on Cyber Monday.

Two caveats. This is referral traffic, shoppers clicking through from an AI assistant, not AI agents buying on their behalf. And it starts from a small base.

My bet is that service teams will feel it anyway. Every extra order is a possible delivery question, return, or payment issue. If AI helps more shoppers decide in the same few days, those questions arrive together, and a deal an assistant recommends can reach your queue before your forecast has moved.

The traditional playbook answers a bigger peak with a hiring decision. Over-hire and carry the cost after the peak, or under-hire and accept long waits and missed SLAs. Both assume capacity equals headcount, and both get made weeks ahead, on a forecast. If volume lands well above plan, recruiting and training can take longer than the peak lasts.

When the size of the peak is this uncertain, a headcount decision made in advance is the wrong place to put the bet.

What does a hybrid workforce that can flex look like?

It pairs AI agents with human agents and runs both on a real-time view of demand, so capacity moves when volume does.

AI agents can take repeatable conversations end to end, like the delivery status, delay, and return questions behind the 2020 retail spike. Copilots can help human agents with the rest by finding answers and generating wrapups. The bar is resolution, which is what those survey respondents asked for.

AI can also sharpen the forecast, and you should still build one. The difference is that the plan no longer depends on the forecast being right, because it can change mid-peak.

Real-time workforce management aligns the human side with the spike. It shows intraday risk as it builds, recommends the next action, and lets agents adjust their own schedules within business guardrails, so coverage follows volume through the day. Verint Workforce Excellence is designed to do that, and it is built to run on top of your existing contact center as a service (CCaaS) platform.

Together, it works as one loop. Real-time WFM sees the spike. AI agents absorb the repeatable part. Copilots speed up the humans handling the rest. WFM then moves schedules for whatever volume is left. Each step shortens the time between spotting a change and acting on it.

How can you start building a hybrid workforce?

You don’t need a replatforming project to start. You need a clear picture of your last peak.

1. Pull your last two peaks and mark the hours where SLAs slipped. Note what you did about it and how long it took.
2. List the top contact reasons in those hours. The ones with predictable answers are your first candidates for AI agents.
3. Time the loop from spotting a gap to changing a schedule. That delay is what real-time WFM shortens.
4. Agree the guardrails: which schedule changes agents can make without approval, and which still need a supervisor.

Start small. Pick one queue, one peak, and one measure, such as time from signal to schedule change, and run the loop on that before you widen it.

Seasonal peaks aren’t new, and the plan still meets reality on the day. What’s different is that you now have more ways to respond. If you want to see what a real-time approach looks like, start with Workforce Excellence.