One Big Table Consulting ← All Advanced Analytics

Domain 2 — Quota design and setting methodology, dark OBT theme

Domain 2 — Quota design and setting methodology
Analytics core · 6 sections · Regression model included · Quiz at end
Why quota is the most important number in comp
Every single calculation in a comp plan — attainment, payout, accelerator trigger, cost forecast — uses quota as its denominator. A perfect payout curve on a broken quota produces broken incentives. Quota that is too easy overpays the company into missing revenue targets. Quota that is too hard demotivates reps and drives attrition. Getting it right is a statistical and strategic problem simultaneously.
Attainment% = Actual revenue / Quota × 100
If quota is wrong by +20% → attainment is wrong by -20% → payout is wrong → cost model is wrong
Quota error propagates through every downstream calculation. A 10% systematic over-assignment of quota across 500 reps means 500 people operating under false performance signals — and the analytics team's models are all off.
The three properties every quota must have
Achievable — but not easy
Industry benchmark: 60–70% of reps should hit or exceed quota in a well-designed plan. If fewer than 50% hit quota, the plan demotivates. If more than 80% hit quota, the company overpaid relative to the difficulty of the target — and left revenue on the table.
Fair — across territories
Two reps with identical skill should have equal probability of hitting quota given their territory. If territory A has 3× the market opportunity of territory B, equal quotas are inequitable. Fairness is a statistical property — it must be tested, not assumed.
Stable — after assignment
Mid-year quota changes destroy rep trust and break the cost model. A plan that was modeled at $2M quota per rep produces completely different cost outcomes if quotas are raised to $2.4M mid-year. Stability requires a defensible methodology upfront.
Microsoft context
Microsoft's internal benchmark for quota attainment is that the median rep should land between 90–110% of quota in a well-calibrated plan year. When the median falls below 80%, it triggers a plan review. When above 115%, finance asks whether quotas were set too conservatively. The comp analytics team tracks this distribution monthly.
% of reps hitting quota65%
The three quota-setting methods — compared
Top-down allocation
How it works: Finance sets company target → split by segment percentage → by region → by territory → by rep. Each layer applies a factor (growth rate, market share, historical performance).

When it works: Consistent, fast, easy to audit. Good when territories are homogeneous.

When it breaks: Ignores ground-level opportunity signals. A fast-growing territory gets the same growth rate as a saturated one. Reps in hard territories get unfair quotas with no adjustment mechanism.
Bottom-up build
How it works: Reps or managers submit territory opportunity estimates → rolled up through org → reconciled against finance target (usually requires haircut since sum of bottom-up almost always exceeds finance target).

When it works: Captures local market intelligence. Reps feel ownership. Good for new market entries where historical data is thin.

When it breaks: Sandbagging — reps systematically understate opportunity to get lower quotas. Creates political negotiation rather than analytical rigor.
Regression / model-based
How it works: Statistical model trained on historical territory data predicts fair quota for each territory based on observable features. Output is a defensible, data-derived quota with confidence intervals.

When it works: Most fair and most defensible. Appeals process becomes evidence-based. Quotas that feel unfair can be tested against the model.

When it breaks: Requires quality historical data. New territories have no history. Model must be retrained as market conditions shift.
Microsoft's hybrid approach
Microsoft uses top-down as the starting point (Finance sets the total number by segment) then applies territory-level regression adjustments to distribute quota within segments fairly. The comp analytics team owns the regression model. Bottom-up estimates from field managers serve as a sanity check input — not the primary driver. When the model output and the field estimate diverge significantly, it triggers a review.
Say this in the interview
"I'd characterize my approach to quota setting as anchored in regression modeling, used as a correction layer on top of top-down allocation. The top-down gives you the total envelope from finance — which is a constraint, not a methodology. The regression model tells you how to distribute that envelope fairly across territories given observable differences in market opportunity, customer density, and historical penetration. The bottom-up field estimate is your smell test — if the model and the field diverge by more than 20%, something is worth investigating."
The regression-based quota model — built out
The goal is to predict a fair quota for each territory based on observable features. "Fair" means: if two territories have the same features, they get the same quota. If one territory has structurally more opportunity, it gets a proportionally higher quota.
Quota_i = β₀ + β₁(MarketSize_i) + β₂(Penetration_i) + β₃(TenureAdj_i) + β₄(GrowthRate_i) + ε_i
β₀ = baseline quota (intercept) · MarketSize = total addressable revenue in territory · Penetration = current % of market captured · TenureAdj = adjustment for rep ramp-up if <12 months · GrowthRate = market expansion rate · ε = residual (unexplained territory-specific factors)
Live model — adjust territory features, see quota output
Market size ($M addressable)$8M
Current penetration (%)25%
Territory growth rate (%)10%
Rep tenure (months)18 mo
What the model outputs and how you use it
Point estimate
The model's best prediction of fair quota for this territory. Starting point for negotiation.
Confidence interval
The range within which the true fair quota falls with 90% probability. Wide CI = uncertain territory.
Residual
Actual quota minus predicted quota. Large positive residual = over-assigned. Large negative = under-assigned.
Fairness flag
Flag territories where assigned quota is outside the confidence interval. Triggers appeal review.
Your DirecTV story maps here directly
At DirecTV you built DMA-level models quantifying revenue exposure across satellite outages — same regression framework, same territorial decomposition logic. The features change (outage scope → market size, DMA population → penetration rate) but the analytical pattern is identical. That story is your proof point for this section.
Quota fairness testing — the analytical pre-flight
Before quotas go live, the analytics team runs a fairness battery. This is not optional — if quotas are demonstrably unfair across segments, the company faces both financial risk (overpaying easy territories) and legal/HR risk (systematic disadvantage of protected groups if territory assignments correlate with demographics).
The four fairness tests
Test 1 — Attainment distribution test
Run a simulation of expected attainment given the quota assignments. Check that the distribution is roughly normal, centered near 90–100%, with a std dev of 15–25%. A bimodal distribution (reps either crushing it or failing) signals quota is set wrong — some territories are structurally easy, others structurally impossible.
Target: Mean attainment = 90–105%, Std dev = 15–25%
Outside these bounds → quota methodology needs adjustment before launch.
Test 2 — Quota-to-market-size ratio test
Divide assigned quota by total addressable market for each territory. This ratio should be consistent across territories within the same segment. If Territory A has a ratio of 15% and Territory B has 35% for the same role type, Territory B is over-assigned relative to opportunity — a fairness failure.
Penetration target_i = Quota_i / TAM_i
Should be consistent within role/segment. Outliers flag over- or under-assignment.
Test 3 — Historical calibration test
Back-test the new quota methodology against the prior 2–3 years of actual performance data. If the model would have produced quotas that resulted in 85% of reps hitting target historically, it's well-calibrated. If it would have resulted in 40% hitting target, it's too aggressive.
Simulated attainment = Historical revenue / New quota
Back-test across 2–3 years. Target: 60–70% of reps hitting in each year.
Test 4 — Segment and cohort equity test
Check whether quota-to-OTE ratio (quota as a multiple of total target comp) is consistent across demographic cohorts. If female reps systematically receive higher quota-to-OTE ratios than male reps in the same role, that is a pay equity issue embedded in quota design — not just a payout calculation.
Quota-to-OTE multiple = Quota / OTE
Should be consistent within role/level. Disparities by cohort = legal and HR risk.
Live fairness simulator
Mean attainment across reps (%)90%
Std deviation (%)20%
Say this in the interview
"Quota fairness isn't a soft concept — it's a testable statistical property. I run four pre-launch tests: attainment distribution shape, quota-to-TAM ratio consistency, historical back-test calibration, and cohort equity check. If any of those tests fail, the quota methodology needs adjustment before the plan goes live. At DirecTV I ran analogous pre-launch validation on DMA-level targets — same framework, same rigor."
How Microsoft sets quota in practice
The quota-setting calendar
Quota design runs Jan–Apr alongside plan design. Finance publishes segment targets in February. Territory-level model runs March–April. Field manager input collected April. Final quota letters issued to reps in June before July 1 launch. Reps have a formal appeal window in July.
Quota appeal process
A rep who believes their quota is unfairly set files a formal appeal. The comp analytics team pulls the model output for that territory, compares it to the assigned quota, and assesses whether the assignment falls within the confidence interval. If the assigned quota is outside the CI — likely adjustment. If inside — the model defends the quota.
Mid-year quota amendments (MYAs)
If a territory changes materially mid-year (major customer lost, new market entered, rep territory split), a formal MYA process triggers. The analytics team re-runs the quota model on the revised territory, produces a prorated adjustment, and models the cost impact of the change before it is approved.
Quota coverage and finance reconciliation
After all individual quotas are set, the analytics team aggregates to check coverage ratio (total assigned quota / company target). Microsoft targets 110–115% coverage at the segment level to absorb expected attainment shortfall. If coverage comes in below 108%, quotas are reviewed for potential uplift.
Ramp quotas for new reps
New reps get a ramped quota schedule: Month 1–3 at 25–40% of full quota, Month 4–6 at 50–70%, Month 7–9 at 75–90%, Month 10+ at 100%. The comp analytics team models the cost impact of the ramp schedule separately — it affects total comp cost significantly in high-hiring years.
The question this section answers in the interview
"How would you approach quota validation?" — Walk through the four fairness tests in order: distribution shape, TAM ratio, historical back-test, cohort equity. Then add: "I'd also validate that the coverage ratio gives finance sufficient buffer, and that the ramp quota schedule for new hires is modeled separately from the fully-ramped population — because mixing them distorts the attainment distribution analysis."
Say this in the interview
"Quota setting at Microsoft's scale is a modeling problem, not a negotiation problem. The model gives you a defensible answer. The appeal process becomes evidence-based rather than political. And the analytics team's credibility with sales leadership depends on being able to walk into an appeal meeting and show the data behind the quota — not just the number."
Domain 2 quiz — seven questions