If half your help desk is home sick next Tuesday, the incidents don’t stop. They get louder. Help desk pandemic planning works the way outage planning does: decide now who covers what, which work waits, and what you’ll tell customers when the answer is “slower than usual.”
That stopped being a hypothetical on April 29, 2009, when the World Health Organization raised its pandemic alert for the new swine flu virus to phase 5. By then the virus had been confirmed in 11 US states, and ABC News reported roughly 160,000 children in 14 states kept home from school as districts closed. Every one of those children has a parent who works somewhere, and a good share of those parents work in a support operation. The question every support manager should be asking is simple: what do we do if our people can’t come in?
I think it’s the right question, and I think most teams are asking it three weeks too late. Here’s how I’d answer it for a team that runs the kind of help desk I described in what a help desk actually is: Tier 1 band-aids, Tier 2 doctors, and a handoff between them that falls apart the moment either side is short-handed.
What actually breaks first when staff are out?
Not the phones. The phones are the easy part, because a Tier 1 seat is the most replaceable seat in the building. What breaks first is the escalation path. Tier 2 is a small group of senior people with narrow skills. When two of them are out, the incidents that would normally reach a specialist sit in a queue with nobody to read them, and the Tier 1 agents who escalated them have no way of knowing.
Staffing a help desk looks like a headcount exercise. It’s a skills exercise. Ten people out of a 100-person floor sounds survivable until you find that all ten were the only ones who could handle one product line. The number of bodies matters less than the number of distinct things each body can do.
So start with a list. Write down every task that has to happen for the help desk to function: answering phones, working the email queue, running the incident tool, escalating, updating customers, handling the on-call rota. Next to each, write the names of the people who can do it. Any line with fewer than three names is a risk, and any line with one name is an emergency you can fix this week.
Cross-training is the cheap answer, and nobody wants to do it when the queue is full. Do it anyway. A Tier 1 agent who can work a single Tier 2 category under supervision is worth more during an absence than a new hire who needs six weeks to learn the product.
Which work should you drop first?
Decide this before you need it. When capacity falls, something has to wait, and the worst way to choose is to let the loudest customer decide. Rank the work in advance. Anything where a customer’s service is down, or a security question is open, stays on. Information requests and account updates can slow down. Reporting, training and process projects can stop entirely for a few weeks, and nobody will miss them.
This is the same logic as the tiered service levels I wrote about in 2007, applied to a bad week instead of a normal one. Your customers already have different response expectations depending on severity. A pandemic plan just says in advance which of those commitments you’ll protect when you can’t keep all of them.
Write the reduced service level down and get your management to sign it. If you wait until the bad week, you’ll be negotiating with a queue behind you and a boss asking why the numbers fell. A signed one-page plan is the difference between “we followed the plan” and “we let you down.”
Also revisit your staffing model. If you schedule using something like the Erlang C approach I covered in the scheduling series, rerun it with a lower headcount and a higher volume. You’ll learn quickly what service level you can actually hold at 80% of your staff, and it’s almost always lower than people assume. Better to know on a spreadsheet than on a Monday morning.
Can your people work from home on day one?
Most help desks can’t, and that’s the finding that surprises managers. Working from home isn’t one switch. A phone agent needs a way to take calls on the company line, access to the incident tool and the knowledge base, and the same security controls that apply on the floor. If any one of those is missing, the person is at home and still unable to work.
Test it. Pick a day, send a handful of people home, and see what fails. You’ll find the VPN licence limit, the softphone that only works on the corporate network, the security question the agent can’t verify without a certain screen. Each of those is cheap to fix in May and expensive to fix in October. A test day also tells you something no memo can: whether your agents can do the job without the person at the next desk to ask.
One more thing about security. When you let agents log into customer accounts from home, you’re extending the trust you’ve built on the floor. Keep the same identity checks. A pandemic is exactly when a social engineer will try a sympathetic story about a sick relative to get an account changed, so the security questions I described for account updates matter more, not less.
What do you tell customers when you’re slower?
The truth, early. Customers forgive a slow help desk that told them it would be slow. They don’t forgive one that kept answering “within the hour” while the queue grew. If your response times are going to stretch, say so on your phone greeting, your email auto-response and your website, with an honest estimate. Put the estimate where people see it before they contact you.
Inside the team, the same rule applies, and it’s the one I’d repeat most. Tier 2 owes Tier 1 an update time even when there’s no answer. In a normal week that’s good practice. In a week with people out, it’s the only thing holding the queue together, because an agent who can say “you’ll hear from us at 3pm” can close a conversation. An agent with nothing to say takes the same call again at 3:05.
Tell your agents what you’re doing about their own health, too. If someone is coughing at the next desk and the policy is to push through, you’ve just published the plan for how the whole team gets sick. Make the policy clear: if you’re ill, stay home, and your work will be covered. It’s an easier sell if the coverage plan already exists.
How do you measure it while it’s happening?
Daily, and with different numbers. Watch the percentage of contacts answered inside your target, the age of the oldest open escalation, and the number of staff available against the number scheduled. The third one is new, and it’s the most useful. It works as one of the leading indicators I wrote about in December. It tells you tomorrow’s problem today.
Keep your reporting simple and keep it honest. Don’t change the definition of a measure mid-crisis to make the week look better. A response-time number that means something different on Wednesday than it did on Monday is noise, and you’ll need those numbers later to show what the event actually cost. Industry groups like HDI publish benchmarking you can compare against once things settle.
After the worst has passed, hold a short review. What did we cover well? What did one person know that nobody else did? What would we cut sooner next time? Put the answers into the plan, and cross-train for whatever was a single point of failure. A review that produces a to-do list nobody owns is theatre.
None of this requires a pandemic. A bad snowstorm, a transit strike or a power outage in one building takes out the same capacity in the same ways. The virus is just the reason it’s on the agenda today, and the plan is worth having whether or not this one turns out to be as bad as the headlines suggest.
You can’t staff for the day everyone’s in. You staff for the day they aren’t.
Hutch Morzaria is a Director-level CX and Support Leadership professional with 19 years of experience building global support organizations across SaaS, Fintech, and enterprise technology. He has hired dozens of support and CX leaders across his career and holds ITIL Expert certification across V3 and V4.
