One Big Table Consulting ← All Advanced Analytics

Domain 5 — Cost impact analysis and ROI modeling for comp plans, dark OBT theme

Domain 5 — Cost-impact analysis and ROI modeling
CFO-facing output · 6 sections · Live decomposition engine · Elasticity models · Quiz at end
What cost-impact analysis actually answers
Scenario modeling and Monte Carlo quantify what a plan costs. Cost-impact analysis answers the harder question: "Is the cost justified by what we get back?" It's the bridge between the analytics team and the CFO conversation.

Every comp plan change is an investment. The company pays more in incentive cost and expects to receive more in revenue. Cost-impact analysis quantifies both sides — ΔCost and ΔRevenue — and produces a net impact and ROI. The behavioral response assumption is the contested number in the middle that requires the most analytical rigor and intellectual honesty.
ΔCost = Σ [Payout_new(rep_i) − Payout_old(rep_i)]
ΔRevenue = Σ [ΔAttainment_i × Quota_i]
Net impact = ΔRevenue − ΔCost
ROI = ΔRevenue / ΔCost
ΔAttainment is the behavioral response — how much more a rep sells because of the plan change. This is the hardest number to estimate and the one finance will interrogate most. It must be explicitly flagged as an assumption, not a fact.
The three components you always decompose
Component 1 — Fixed cost delta (ΔBase)
If the plan change includes a base salary adjustment, this is the simplest component — headcount × base change. Fully deterministic. No distribution, no uncertainty. But it's often zero in a plan redesign where only the incentive structure changes.
Component 2 — Variable cost delta (ΔTI exposure)
The change in incentive payout driven by the new curve, threshold, or accelerator parameters — holding attainment constant. This isolates the "plan mechanics" effect from the "rep behavior" effect. Calculated by running both old and new formulas on the same historical attainment distribution.
Component 3 — Behavioral response delta (ΔAttainment × Quota)
The incremental revenue from reps performing better because of the new plan. This is a model-derived estimate — not a fact. Must be sourced from historical behavioral response data or academic elasticity research. Always presented with confidence bounds. This is the number that either makes the ROI positive or negative.
Your DirecTV story maps here directly
The DirecTV satellite outage model was a cost-impact analysis: ΔRevenue (revenue at risk from lost viewership) versus ΔCost (restoration and retention spend). The decomposition was DMA-level, the behavioral response was subscriber churn elasticity, and the output was a net impact range for leadership. Same three-component structure. Same analytical discipline. Different domain.
Cost decomposition — waterfall from old plan to new plan
The waterfall chart is the clearest way to communicate cost impact to leadership. It starts from the current plan cost, adds each component of change, and lands on the new plan cost. Every step is labeled and quantified. No number appears without a source.
Live cost decomposition — adjust plan change parameters
Rep count150
Old TI per rep ($K)$50K
New TI per rep ($K)$65K
Old accelerator rate (%)150%
New accelerator rate (%)175%
Average attainment (%)88%
Say this in the interview
"I always present cost impact as a waterfall — starting from current plan cost, adding each component of change, landing on new plan cost. Every bar in the waterfall has a labeled source: TI change, accelerator change, threshold change. That transparency is what gives finance confidence in the number. A single lump-sum delta with no decomposition is not a model — it's a guess."
The ROI model — is the investment worth it
The ROI model answers the CFO's real question: "For every additional dollar we spend on incentive comp, how many dollars of incremental revenue do we get back?" An ROI above 1.0× means the plan change pays for itself. An ROI of 3.0× means every $1 of additional comp cost generates $3 of revenue. Below 1.0× means you're paying more than you're getting back — plan change is financially destructive.
ROI = ΔRevenue / ΔCost
Break-even attainment lift = ΔCost / (Reps × Avg_Quota)
Cost of sales % = Total comp cost / Total revenue generated
Break-even attainment lift = the minimum behavioral response needed for the plan change to pay for itself. If break-even requires a 15% lift in attainment but historical behavioral response data shows 5–8% is realistic, the ROI case is weak. Flag this explicitly.
Live ROI calculator
ΔCost — additional comp expense ($K)$1,500K
Expected attainment lift (%)8%
Rep count150
Avg quota per rep ($M)$2.0M
Avg attainment before change (%)88%
Cost of sales benchmarks — where Microsoft sits
Enterprise SaaS
8–14% cost of sales. Microsoft cloud products typically target 10–12%.
Cloud infrastructure
6–10%. Lower margin business — tighter comp cost envelope.
SMB / transactional
5–8%. High volume, low touch — comp cost must stay lean.
Warning threshold
>16% triggers finance review — plan may be structurally over-paying.
Say this in the interview
"The ROI model is how I translate the analytics into a CFO-ready recommendation. I present three numbers: ΔCost (what we're paying more in comp), ΔRevenue (what we expect back based on behavioral response), and the ROI ratio. But I always also present the break-even attainment lift — the minimum behavioral response needed for the plan change to pay for itself. If that break-even is realistic given historical data, the case is strong. If it requires a response the data doesn't support, I say so explicitly."
Behavioral elasticity — the contested number in every ROI model
Elasticity is the ratio of behavioral change to incentive change. In comp analytics: how much does attainment increase when you increase the incentive opportunity by 10%? An elasticity of 0.5 means a 10% increase in incentive opportunity produces a 5% increase in attainment. This single assumption can flip an ROI model from positive to negative — which is why it requires explicit sourcing and transparency.
Elasticity (ε) = % change in attainment / % change in incentive opportunity
ΔAttainment = ε × (ΔIncentive_opportunity / Old_incentive_opportunity)
ΔRevenue = ΔAttainment × Quota × Reps
Incentive opportunity = TI × curve(attain%). If you raise the accelerator, you've increased the incentive opportunity for reps above quota. The elasticity tells you how much more they sell as a result. Academic research (Lazear, Bandiera) suggests elasticities of 0.2–0.8 for performance pay, higher for simpler tasks.
Elasticity by rep segment — live model
Incentive opportunity increase (%)20%
Avg quota per rep ($M)$2.0M
Rep population100
How to source and defend your elasticity assumption
Source 1 — Natural experiments from prior plan changes
When Microsoft changed an accelerator rate for one segment but not another in a prior year, you have a quasi-experiment. The segment that got the change is your treatment group. Measure the attainment delta vs. the control segment. That measured delta is your elasticity estimate — the most credible source because it's from your actual rep population.
Source 2 — Academic benchmarks
Lazear (2000) showed productivity increased 44% when Safelite Glass shifted from hourly pay to piece rates. Bandiera et al. (2007) showed elasticity of ~0.5 for performance pay in similar sales contexts. Use these as sanity checks on your own estimates — if your model assumes ε=1.5, something is wrong.
Source 3 — Sensitivity framing (when you don't have data)
If you have no historical basis, present the ROI model at three elasticity levels: conservative (ε=0.2), base case (ε=0.5), optimistic (ε=0.8). Show finance what ROI looks like under each assumption. This is more honest than presenting a single elasticity as fact — and it reveals how dependent the ROI case is on behavioral response.
Say this in the interview
"The elasticity assumption is where I'm most explicit about uncertainty. I never present it as a fact — I present it as an assumption with a source. If we have natural experiment data from a prior plan change, I use that and state it. If we don't, I present the ROI at three elasticity levels and let finance decide what confidence level they need to approve the change. Intellectual honesty about the contested assumption is what builds long-term credibility with finance."
How Microsoft runs cost-impact analysis in practice
The plan change package — what goes to the governance committee
Every proposed plan change above a materiality threshold requires a formal plan change package submitted to the SIC governance committee. The analytics team owns the financial modeling section, which includes: current plan cost baseline (P50 from Monte Carlo), proposed plan cost (P50 and P90), ΔCost decomposition waterfall, behavioral response assumption with source, ΔRevenue estimate with confidence bounds, ROI ratio, break-even attainment lift, and cost of sales % under old vs. new plan.
Materiality thresholds
Not every plan tweak requires a full package. Microsoft uses materiality thresholds: changes with expected ΔCost below $500K at P50 can be approved by VP Finance. $500K–$5M requires SVP Finance + HR. Above $5M requires CFO-level sign-off. The analytics team calculates the expected ΔCost as part of initial triage — determining which approval level applies is the first deliverable.
Cost of sales monitoring — monthly cadence
The analytics team tracks cost of sales % monthly against the plan target. If actual cost of sales tracks above the P50 forecast by more than 1.5 percentage points for two consecutive months, it triggers an early warning report to finance. This report includes: current trajectory, projected full-year cost at current run rate, variance decomposition (is the overrun from higher attainment, plan mechanics, or headcount growth), and recommended actions.
Windfall and sandbag analysis
Windfall: identify reps whose payout is driven by a single large deal rather than sustained performance. Flag for windfall review — the deal may have landed regardless of rep effort. Analytics deliverable: regression of payout on deal count and deal size to separate "one big deal" reps from "sustained high performers."

Sandbagging: identify reps who consistently hit quota in Q4 but underperform in Q1–Q3. This is a plan timing signal — reps are holding pipeline to manage their attainment trajectory. Analytics deliverable: quarterly attainment pattern analysis by rep cohort.
The ROI conversation with sales leadership
Sales leaders want higher TI and steeper accelerators. Finance wants lower cost. The analytics team is the neutral party that quantifies both sides. The ROI model is the common language. When sales says "raising the accelerator will drive 15% more revenue," the analytics team's job is to test that assumption against data and either validate, modify, or challenge it — with evidence, not opinion.
The question that wins the interview
"How would you build the ROI case for raising the accelerator rate?" Walk through the five steps: (1) calculate ΔCost by running old and new curves on historical attainment distribution, (2) source the behavioral elasticity from prior plan change natural experiments, (3) calculate ΔRevenue as elasticity × incentive opportunity change × quota × reps, (4) compute ROI and break-even lift, (5) present with explicit confidence bounds on the elasticity assumption. Then add: "I'd also check whether the cost of sales % under the new plan stays within benchmark range — because even a positive ROI can be a bad plan if it pushes cost of sales above 15%."
The end-to-end answer — Domains 1–5 in one paragraph
"A comp plan change is a financial investment decision. To model it, I start with the plan architecture — fixing base, TI, threshold, and accelerator parameters. I set the quota correctly so the denominator is defensible. I model the attainment curve to understand the behavioral levers. I run scenario analysis for leadership alignment and Monte Carlo for finance budget setting. Then I build the cost-impact model: decompose ΔCost into its components, estimate ΔRevenue using a sourced elasticity assumption, and produce an ROI ratio with break-even attainment lift and cost of sales %. That full package — scenario, Monte Carlo, sensitivity, cost-impact, ROI — is what goes to the governance committee. Each model serves a different stakeholder. The analytics team's job is to run all of them and synthesize them into a recommendation the business can act on."
Domain 5 quiz — eight questions