Workforce Planning That Holds Up Under Uncertainty

The organizations that held capacity through volatile demand were the ones that stopped planning technology and workforce separately. When you plan them together you can see where capacity actually lives and what is consuming it. When you plan them apart you are always optimizing half the picture and wondering why headcount decisions don't hold.

We had a brutal stretch in 2019 where two major clients churned in the same month and demand dropped 40% overnight. I watched other 3PL owners panic and lay off their best warehouse workers, only to scramble when volume returned six weeks later and pay 30% more for temp labor that couldn't hit their quality standards.

My decision rule became stupidly simple: protect roles that touch product or customers, pause everything else. When volume drops, the instinct is to cut warehouse staff because labor is your biggest variable cost. Wrong move. Those people know your systems, your clients' SKUs, their packing quirks. Replacing them costs way more than carrying them through a slow period.

Instead, we paused all hiring for supervisory and admin roles. We stopped the search for a new operations manager. Froze our plans to add another customer service rep. I even took over some procurement tasks myself for three months. The warehouse team stayed intact. When demand rebounded, we were the only 3PL in our market that could scale immediately without quality drops.

The scenario planning practice that saved us was tracking what I called the "ramp time metric" for every role. How long does it take someone new to reach 80% productivity? For warehouse associates picking and packing, it was 6-8 weeks in our operation. For a customer service rep who knew our clients, it was 12 weeks. But for a finance person or a recruiter? You could replace them in two weeks if needed.

So the rule became: if ramp time exceeds your cash runway for that role by more than 2x, you keep them. If a warehouse associate takes 8 weeks to train and you've got 16+ weeks of cash, you hold. If demand craters and you're at 6 weeks of runway, different conversation.

The brands winning right now aren't the ones who cut fastest. They're the ones who know exactly which people are impossible to replace.

A practice that worked well for us was planning around workload changes instead of relying on top line growth. We built staffing plans based on the type of work coming in such as steady work spike work and strategic work. Each type needed a different response from our team and resources. Steady work supported stable hiring while spike work needed short term support or better task sharing.

This approach gave us a clear way to move people without hurting future progress. We focused on employees who could adapt across roles and placed them into strategic work first. This helped us stay focused on long term goals while reducing waste. We reviewed plans every month with one rule that we do not cut skills that take longer to rebuild than the market takes to change.

When demand shifts quickly, the most reliable approach has been to anchor workforce decisions to a capability-based planning model rather than static job roles. At Invensis Learning, demand signals are mapped against clusters of critical skills—such as agile delivery, cybersecurity, and cloud operations—rather than individual titles. A consistent decision rule has been to protect and continue hiring for roles aligned with high-growth, future-critical capabilities, while pausing or redeploying talent tied to declining or cyclical demand areas. This reduces the risk of eroding long-term capacity while still enabling short-term flexibility.

One scenario planning practice that has proven particularly reliable is maintaining a "70-20-10 workforce allocation model"—where 70% of capacity supports current demand, 20% is flexibly redeployable, and 10% is continuously upskilled for emerging priorities. This structure creates built-in agility without reactive overcorrections. According to the World Economic Forum, nearly 44% of core workforce skills are expected to change by 2027, reinforcing the importance of dynamic skill-based planning over rigid headcount strategies. Organizations that embed continuous reskilling into workforce planning are better positioned to absorb volatility while sustaining long-term growth.

I faced this when demand jumped as we moved into wholesale and pharmacy—orders increased, but the real bottleneck was education and support, not packing boxes. Early on I nearly overhired for operations, but what actually drove growth was people who could train staff and answer questions properly. Now my rule is simple: hire or redeploy into roles that remove the current constraint to product use or reorder, not just the visible workload. I map where things are slowing—stockists unsure, customers confused, or fulfilment delays—and adjust the team around that. When demand shifts, don't default to more hands; fix the point where progress stalls. It protects long-term capacity because you're building capability where it matters, not just reacting to noise.

What we always try to avoid is making quick decisions that end up not actually helping in the way that we need them to. I've found that one of the best ways to avoid making a decision that's wrong in these kinds of scenarios is being in communication with leaders from all of the departments impacted. The more perspectives, and the more ideas about what could be done to help, the better the decision and outcome.

We focus on what's most needed. When there is a shift in demand, that means that there are going to be certain departments or teams that see more of an uptick in work than others. You don't necessarily need to add one new team member for every single department, because that can end up wasting money and time while simultaneously not giving enough support to the specific teams that need it most.

The first thing to do when there is a sudden change in demand is to separate short-term demand from long-term capacity. As a CEO, I look for positions that protect delivery, customer experience, and future growth. These are the areas that we will try not to cut too much, no matter how the conditions change.

One of the thumb rules that we have used for many years is to plan for capability, not just on head count. If a position supports a capability that will be required in six to twelve months, we will re-deploy, train someone else to fill that role or adjust the priority of that position before terminating it altogether. Recruiting can slow down, but capability doesn't go away. This allows companies to react quickly without making rash decisions, which may open larger gaps down the road. Workforce planning should be aimed at protecting the company's future, not just addressing the immediate quarter.

By cutting horizontally, firms generally mistake when demand shifts and their long-term capacity is eliminated. The solution is to use the Core Architecture guideline by determining which of your engineering roles are responsible for 20%, or less of your organisation's institutional knowledge, which should not be subject to any type of reduction. As an example, if you are using a lay-off to reduce payroll costs, you will not be able to replace that expertise when your market regains momentum because the cost of replacing that institutional knowledge will far exceed any savings you realised from layoffs.

To retain the remaining 80% of your engineering resources, we employ what I refer to as a flow-state utility; where we gauge the capabilities of developers on the current project phase as opposed to job descriptions that are static during typical business cycles. As an example, when an individual developer is working on a feature set and the project stalls due to market changes, rather than cut the resource we reassign him to a project that has been pushed back because of high growth cycles; specifically, technical debt reduction or documentation.

Using flow-state utility allows engineering teams to remain intact and productive until such time as the market returns, thus allowing the engineering team to scale as the market returns.

Navigating the shifts is more about trust than about math; thus when you retain your core team during periods of volatility, they will recall the stability during that period, creating a significant competitive advantage in the normalisation of the marketplace.

I'm Runbo Li, Co-founder & CEO at Magic Hour.

The honest answer is that we've built Magic Hour to millions of users with a two-person team, so our "workforce planning" looks nothing like what you'd read in an HR textbook. And that's exactly the point. The decision rule that's proven most reliable for us is simple: never hire for a function until AI can't do it, and even then, wait one more month.

Here's why that works. When David and I launched Magic Hour, the conventional wisdom said we needed a customer support team, a marketing team, a DevOps person, maybe a designer. Instead, we built AI-powered systems to handle support tickets, automate deployment pipelines, generate marketing content, and run analytics. Every time demand shifted, like when an NBA edit went viral and brought hundreds of thousands of new users overnight, we didn't scramble to hire. We tuned our systems. We redeployed our own time from one priority to another. That's the real redeployment question in 2024 and beyond. It's not about moving people between seats. It's about moving your attention between AI-powered workflows.

The one practice I'd recommend to any founder or operator is what I call the "last human standing" test. Before you open a role, ask: if I gave a sharp generalist access to the best AI tools available today, could they cover this function at 80% quality? If yes, don't hire a specialist. Give that generalist the tools and the mandate. You preserve long-term capacity because generalists with AI fluency can flex across functions when demand shifts. Specialists without AI fluency become bottlenecks the moment priorities change.

When Mark Cuban became a paying customer and the Mavericks reached out, we didn't spin up a sales team. I handled it personally, used AI to prep materials, and closed the relationship. That's a pattern we've repeated dozens of times. The constraint was never headcount. It was creativity in how we applied the tools.

The companies that will struggle most in the next five years aren't the ones that hired too few people. They're the ones that hired too many before understanding what AI could replace. Hiring is a one-way door with high switching costs. AI workflows are infinitely adjustable. Build on the adjustable layer first, and you'll never panic when demand moves.

The decision rule that proved reliable: distinguish between demand shifts that affect revenue-generating functions versus those that affect support functions, and protect the former unconditionally.

At Dynaris, we've navigated demand volatility as an early-stage AI startup — periods of rapid customer acquisition followed by stretches requiring infrastructure consolidation. The mistake many founders and CHROs make is treating all headcount decisions symmetrically during a slowdown. They pause across the board or redeploy based on who's available rather than where value is created.

Our framework: we map every role to one of three categories — (1) revenue-critical: directly tied to customer acquisition or retention, (2) capacity-critical: required for product delivery at current scale, (3) growth-optional: roles that accelerate but don't enable. When demand shifts, we protect categories 1 and 2 unconditionally and redeploy or pause category 3 first.

The scenario planning practice that helped: we run a quarterly "70% revenue scenario" — what does the org look like if revenue drops 30%? Which roles become redundant? Which gaps open? This exercise forces explicit mapping between roles and value generation before a crisis happens, so decisions under pressure are pre-reasoned rather than reactive.

The key insight from running this: most redeployment decisions that feel difficult in the moment are actually easy if you've done the scenario mapping in advance. The emotional difficulty of in-the-moment workforce decisions comes from lack of preparation, not from the decisions themselves being genuinely hard.