Case Study: Building a Compensation Structure for a New Function
A founder approves a new AI risk team on Friday; by Monday, HR is staring at five unfamiliar roles, no salary history, and three managers saying, “We need top talent fast.” This is where compensation design becomes visible: if the structure is weak, the company either overpays in panic, loses candidates, or creates internal resentment before the function even launches.
- Start with work, not salary. Define what the new function must deliver, then design roles, levels, pay ranges and governance.
- Compensation structure means the formal architecture of grades, salary ranges, incentives, benefits and rules used to pay people consistently.
- Use a five-part funnel: business need → role architecture → pay philosophy → ranges and mix → governance.
- For a new function, blend external market pricing with internal equity; either one alone creates risk.
- Track compa-ratio, range penetration, offer acceptance rate, critical-role attrition, pay equity gap and variable-pay payout ratio.
- The best answer in an interview is not “match market salary”; it is “design pay to attract, motivate, retain and stay fair.”
Big Picture
Building compensation for a new function is like converting an uncertain business bet into a disciplined pay system. The funnel below is the mental model: wide strategic ambiguity at the top, clear pay rules at the bottom.
Core Explanation
The big idea: do not price people before you have priced the work. A new function is usually messy - titles are fluid, hiring urgency is high, and managers may benchmark against whichever company they admire. Your job is to bring structure.
Think of a new function such as customer success in a SaaS firm, AI risk in a bank, dark-store operations in quick commerce, or ESG reporting in a listed company. Each function needs a different compensation logic because the work differs in scarcity, measurability, risk and time horizon.
The Five-Step Process
Key Design Choices for a New Function
Four choices decide whether the structure will work in the real world.
Choosing the Right Pay Archetype
A new function does not automatically need high fixed pay or high incentives. The answer depends on two questions: is the skill scarce, and can output be measured cleanly?
For example, an AI ethics head or actuarial specialist may require strong fixed pay because skills are scarce and output is judgement-heavy. A sales development team can carry higher variable pay because qualified meetings, pipeline and conversion can be measured. A dark-store team may need stable base pay plus team-level incentives tied to fill rate, picking accuracy and shrinkage.
Metrics to Track Once the Structure Goes Live
Compensation design is not finished when offer letters go out. You track whether the structure attracts, motivates and retains talent without breaking fairness or cost discipline.
Worked Example: Setting a Salary Range
Suppose a company is creating a new Customer Success Manager role. Market data suggests a midpoint of ₹20 lakh annual fixed pay for the relevant level and location. The company wants a salary range spread of 50 percent around the midpoint.
A simple structure can be:
- Midpoint = ₹20 lakh
- Minimum = ₹16 lakh
- Maximum = ₹24 lakh
- Range spread = (₹24 lakh - ₹16 lakh) / ₹16 lakh = 50 percent
If a candidate is offered ₹21 lakh, their compa-ratio is ₹21 lakh / ₹20 lakh = 1.05. That is reasonable for a candidate who is already strong for the level. If a new hire needs ₹26 lakh, that is above range and should trigger an exception review rather than an informal promise.
Definitions
Milkovich, Newman and Gerhart define compensation as “all forms of financial returns and tangible services and benefits employees receive as part of employment.”
Zepto: Compensation Design for a New Operating Function
Zepto's quick-commerce model shows why a new function needs compensation that rewards speed, reliability and control - not just hiring velocity.
Situation. Quick commerce in India expanded rapidly as customers began expecting grocery and daily-need deliveries in very short time windows. For a company like Zepto, this meant the organisation could not rely only on corporate category teams or delivery partners. It needed tightly run operating functions around dark-store management, inventory accuracy, rider coordination, category availability, shift planning and customer issue resolution.
The compensation challenge. This is not a standard retail-store job and not a pure technology job. The work sits at the intersection of operations, analytics, local execution and service reliability. If pay is designed only as low-cost operations pay, the company risks turnover and weak execution. If pay is designed like premium tech pay for every role, unit economics suffer.
The move. A strong compensation design for this kind of new function would separate roles by contribution: store execution, inventory planning, cluster leadership, central category support and operations analytics. Fixed pay would provide stability for frontline and supervisory roles. Variable pay would be tied carefully to team metrics such as order accuracy, stock availability, shrinkage control and attendance reliability. Scarce roles such as planning analytics or cluster leadership would be benchmarked more aggressively against relevant talent markets.
Outcome or lesson. Zepto's operating model is driven primarily by dense dark-store execution and rapid fulfilment discipline, supported by technology, assortment planning, local supply chain and hiring systems. The compensation lesson is clear: when a new function is mission-critical, pay must reinforce the operating model, not merely copy market salaries.
How AI Changes Building a Compensation Structure for a New Function
AI does not remove compensation judgement; it makes the diagnosis sharper and faster.
- Skills-based job architecture: AI can help cluster job descriptions into skill families, identify overlapping roles and flag whether a new function really needs five levels or only three.
- Market intelligence synthesis: Compensation teams can use AI to summarise public salary reports, job postings and competitor role descriptions, then compare how titles differ across industries. Human validation is essential because salary data can be noisy or location-biased.
- Pay equity and anomaly detection: AI models can flag unexplained pay outliers by role, level, tenure, performance and location. The risk is bias amplification, so HR must audit variables and document business reasons.
Use NotebookLM or Claude before an interview: upload a company annual report, recent job postings and your compensation framework, then ask, “What new functions is this company likely building, and how should compensation be structured for one of them?”
Interview Relevance
“Our company is launching a new customer success function for enterprise clients. How would you design the compensation structure?”
Use the phrase “I would first price the role, not the person.” It signals maturity because you are balancing attraction, motivation, affordability and fairness.
Common Mistake
The single biggest mistake is jumping straight to “pay market salary” without defining job architecture. It costs candidates because it ignores internal equity, level clarity, pay compression and governance. One-line fix: define the work and levels first, then benchmark the market and build ranges.