A professional services firm replaced its annual headcount plan with a 24-month workforce shape. Instead of asking how many people to add next year, it asked which role categories would compound and which would taper as delivery automation matured. The team modeled five role families against a demand forecast, protected two that were undersupplied, and let attrition thin two others rather than backfilling. Twelve months in, revenue per delivery head rose 14 percent and the firm stopped hiring into roles it was about to automate. The shape, not the number, became the planning object leadership actually governed.
The plan that kept being wrong
The firm was a 1,900-person professional services business with roughly 1,400 people in billable delivery. Every October it ran the same exercise. Each practice leader submitted a headcount request for the coming fiscal year, finance summed the requests, trimmed them by a blanket 8 percent to hit the margin target, and the number became the plan. By the following June the plan was already wrong. Two practices had overhired into work that softened; one had underhired into demand it could not staff and had turned away 2.3 million dollars of qualified pipeline. The annual number was precise and useless.
The deeper problem was that the plan counted people without describing the shape of the workforce. It could not answer whether the firm was building toward the capability mix its market was moving to, because it never named the role categories or forecast which ones would compound. Delivery automation was quietly changing the economics underneath the plan. Junior analysts spent 40 percent of their hours on document assembly and data cleanup that tooling was starting to absorb, while the firm was chronically short of the senior engineers who designed the automated workflows. The annual plan treated a junior analyst and a delivery engineer as interchangeable units of headcount, when in reality one was becoming cheaper to automate and the other was the scarce resource the whole model depended on. Stratenity was engaged to replace the annual headcount plan with a 24-month workforce shape that leadership could actually govern, and to instrument it so the shape could be steered quarter by quarter rather than relitigated once a year.
Five role families, forecast against demand
The engagement began by collapsing 61 job titles into five role families defined by what the work would become, not what it was called today. Each family was forecast against a 24-month demand model built from the firm's own pipeline, win rates, and the automation roadmap. The output was not a single number but a trajectory for each family: compound, hold, or taper. Roles that compounded were protected and funded ahead of demand; roles that tapered were thinned through attrition rather than layoffs, with no backfill authorized.
| Role family | Headcount at start | 24-month trajectory | Planned action |
|---|---|---|---|
| Delivery engineers (workflow and automation design) | 90 | Compound to 165 | Fund ahead of demand, external hire plus reskill |
| Quality stewards (review, assurance, client sign-off) | 70 | Compound to 120 | Protect and expand, promote from analyst pool |
| Client leads (relationship and scope ownership) | 240 | Hold near 250 | Steady hiring to replace attrition only |
| Junior analysts (assembly, data prep, first-draft work) | 520 | Taper to 380 | Freeze external hiring, redirect via reskill track |
| Coordination and admin support | 200 | Taper to 150 | No backfill, absorb through natural attrition |
The most contested cell was the analyst taper. Freezing junior hiring felt like starving the pipeline that produced future seniors, and two practice leaders argued hard against it. The resolution was a reskill track: 85 analysts were routed into a six-month path toward delivery engineer and quality steward roles, funded from the salary line that external analyst hiring would have consumed. Because the funding was a reallocation rather than a new ask, finance signed off without a fresh budget fight. The shape held the analyst number down while lifting the two families that were undersupplied, and it did so without a single involuntary exit in the first year. Each family also carried a leading indicator, so leadership could see a trajectory drifting off target within a quarter rather than discovering the miss at the next annual plan.
The demand model itself was deliberately kept simple enough that a practice leader could argue with it. It took three inputs per family: the pipeline the firm expected to win, the automation coverage expected to land in each period, and the average productive capacity of a person in that family. Multiplying those together produced a required headcount trajectory the leaders could challenge input by input, rather than a black-box number handed down by finance. When a practice leader disagreed, the argument was about an assumption they could name, not about who had lobbied hardest. That transparency is what converted the shape from a planning artifact into something the executive team was willing to govern quarter after quarter.
What moved in twelve months
- Revenue per delivery head rose 14 percent, from 218,000 to 249,000 dollars, as work shifted from assembly hours to engineered and reviewed output.
- The delivery engineer family grew from 90 to 131 on the way to its 165 target, with 38 of the additions coming from the internal reskill track rather than external hiring.
- The firm stopped hiring into the analyst family entirely and absorbed a 27-person reduction through attrition, avoiding an estimated 2.9 million dollars in salary it would have spent backfilling roles it was automating.
- Turned-away qualified pipeline fell from 2.3 million dollars to under 400,000, because the undersupplied quality steward and engineer families were funded ahead of the demand rather than behind it.
- The October headcount ritual was retired. Leadership now reviews the five-family shape quarterly against actual demand and adjusts trajectories, rather than negotiating a single annual number.
What the engagement taught
- A headcount number answers how many. A workforce shape answers which capabilities compound. The second question is the one that determines margin, and the annual plan never asked it.
- Role families beat job titles. Collapsing 61 titles into 5 families made the trajectory legible to executives who could not have reasoned about 61 separate lines.
- Tapering through attrition and reskilling is slower than layoffs but far more durable. It preserved institutional knowledge and avoided the morale collapse that would have undermined the compound families.
- Funding compound roles ahead of demand is uncomfortable and correct. The firm had always hired behind demand, which is exactly why it kept turning away pipeline.
- The reskill track only worked because it was funded from a real line, the frozen analyst hiring budget. Reskilling framed as goodwill fails; reskilling framed as a reallocation succeeds.
How to run this in your firm
- Collapse your job titles into five to seven role families defined by what the work is becoming, not what it is called today.
- Build a 24-month demand model from your own pipeline, win rates, and automation roadmap, and forecast each family as compound, hold, or taper.
- Protect and fund the compound families ahead of demand, and authorize no external backfill for the taper families.
- Fund a reskill track explicitly from the hiring budget you are freezing, and set enrollment targets tied to the compound-family gaps.
- Replace the annual headcount negotiation with a quarterly shape review, and instrument revenue per delivery head as the metric the shape is accountable for.