The cheapest way to handle a predictable peak is to not hire for it.
Renewal season, a storm, a book roll, a carrier change. The volume arrives on a schedule you can mostly see coming, and then it leaves again. Staffing to cover the peak means carrying that cost through every quiet month. Staffing to the average means the peak lands on whoever is most senior and least able to refuse it.
This piece covers why both of the usual options are bad, how to tell a genuine peak from a permanent step up in volume, what has to be true before an operation can flex at all, and how to plan for the next one.
Both of the usual options are bad, for different reasons
Hiring for the peak fails on arithmetic and on timing. The cost is permanent while the volume is not, so the new seat is underused for most of the year. And hiring takes longer than the peak lasts. A CSR posted in September, interviewed in October, and productive in January has missed the thing they were hired for.
Absorbing the peak fails more quietly. The work does not disappear, it redistributes to the people with the least room to say no, which is usually your two most experienced people. What you actually spend is not overtime. It is their attention on the work only they can do, plus a quiet increase in the chance one of them starts taking recruiter calls in February.
Most agencies alternate between these two, hiring after a peak that hurt and then carrying the cost until the next quiet stretch makes it look like a mistake.
Tell a peak from a step change before you decide anything
The two need opposite responses, and they are easy to confuse in the middle of one. A peak returns to baseline. A step change does not, and a step change genuinely does justify a hire.
The way to tell them apart is to look at request volume rather than at how busy everybody feels. Busy is not a measurement, and by the time it registers, both situations feel identical. Count requests by type, week over week, across the last two cycles. If last year returned to the previous baseline within six to eight weeks of the peak, you have seasonality. If it settled at a new, higher floor, you have growth, and you should staff for it.
Agencies that cannot answer this question usually cannot answer it because service work was never counted as discrete requests in the first place, which is the same reason the peak is hard to absorb.
You cannot flex work that is not defined
This is the part that gets skipped, and it is the reason “we will get temporary help” rarely works.
Capacity can only be added to work that can be described, handed over, and checked. If a renewal in your agency is a thing a specific person knows how to do rather than a defined sequence of steps, then nobody can help with it at short notice, and any extra hands need weeks of supervision from the people already underwater. That is why bringing in help during a peak often makes the peak worse.
Defining the work is the unglamorous prerequisite. Once a request type is a set of steps with owners and a definition of done, it becomes something you can add capacity to on a Tuesday. Until then, every option is a hire.
Adding capacity without adding headcount
The version that works is mechanical. You send the requests in and they come back complete, routed by license and skill so the steps that need a licensed person get one and the rest do not wait behind them. Cost and speed are visible per task, so the peak has numbers attached to it rather than a feeling.
Nothing is carved out. The book stays yours, the client relationships stay yours, and you decide request by request. Your own staff stays the default, and this is extra capacity when you need it. When the peak passes, you send fewer requests, which is the part a permanent hire cannot do.
COVU runs on top of your AMS, not instead of it, with 8 AMS integrations live, so none of this asks you to change your system of record in the eight weeks you can least afford it.
What the production data shows
Across the agencies we operate:
- About 40,000 tasks a week are classified and routed with no manual intervention, which is what makes volume something you can absorb rather than something you have to feel.
- AI triage filters about 68% of inbound as noise, meaning carrier notifications, duplicates, and automated confirmations that never needed a person. Classification runs at a 98% triage success rate.
- Roughly 44% of completed service tasks are executed by AI.
- Escalations, meaning a task where someone is stuck and has to hand it on, are down 20%, from 435 a day to 348, and still declining.
- Median handle time on a task is 3.8 minutes, down 16% from 4.5.
Licensed work is handled by licensed agents, commercial and personal lines, U.S.-based with 8 or more years of experience, licensed in all 50 states.
S&G Mitchell went from 17.9% to 60%+ EBITDA in 12 months. Same agency, same customers, same book of business. What changed was the operating model underneath.
A plan for the next peak
Three steps, and the first one is the only one that has to happen before the volume arrives.
Count what happened last cycle, by request type, so you know the shape of the thing you are planning for rather than its reputation. Pick the one or two request types that made up most of the increase, which for most agencies is certificates and renewal administration. Define those as steps with an owner and a definition of done, so they can be handed to anyone, including someone outside the building.
Then, when the peak arrives, you have a choice that is not between a permanent hire and a hard eight weeks. Staff for your average and route the difference.
Frequently Asked Questions
How do I know whether a busy period is seasonal or permanent growth?
Count requests by type, week over week, across your last two cycles. Seasonality returns to the previous baseline within roughly six to eight weeks of the peak. Growth settles at a new, higher floor and does justify a hire. Feeling busy cannot distinguish the two, which is why the count matters more than the impression.
Why does hiring for renewal season usually not work?
Timing and arithmetic. The hire is permanent while the volume is not, and recruiting, onboarding, and ramp usually take longer than the peak itself lasts, so the new person becomes productive after the reason for hiring them has passed.
Can we just bring in temporary help during the peak?
Only for work that is defined well enough to hand over. If a request type lives in one person’s head rather than as a documented sequence, extra hands need supervision from the people who are already the bottleneck, which can make the peak worse. Defining the work first is what makes flexing possible, and it is the same groundwork that makes AI tools worth anything.
Which work should move first during a peak?
The high-volume, low-judgment request types, usually certificates and renewal administration. They are frequent enough to relieve real pressure and well-defined enough to hand over safely. There is a fuller argument for starting with certificates.
Does flexing capacity mean replacing our service team?
No. Your own staff stays the default and this is extra capacity when you need it. Nothing is carved out, the book stays yours, and you decide request by request. When volume drops you send less work, which is the flexibility a permanent seat cannot give you.
What does a peak actually cost if we absorb it internally?
Mostly not overtime. It is the senior time diverted from work only senior people can do, the small things that slip while attention is elsewhere, and the retention risk that follows a hard quarter. Some of that shows up later as churn, which is the mechanism behind clients leaving at renewal.