← Back to the guideThe full worked example: Check a new commission plan before it's signed
campo.
Field guide / Sales / UC-001 · SC-29
The result

"Top performers earn 22% more"? Top by bookings. The plan cuts pay for 29 of 36 service reps, who sell the multi-year agreements we said we want.

A VP Sales at a 15,000-person electrical equipment maker checks next year's commission plan before the CRO and CHRO sign it, with any AI chat her company allows.

The resultElena's page for the CRO and CHRO. Fictional company.
OSTRANDER ELECTRIC · SALES · NORTH AMERICA
FY27 commission plan: check before Friday's sign-off
For: Dev Malhotra, CRO, and Ruth Calloway, CHRO
From: Elena Vargas · Monday, December 14

THE ANSWER

Don't sign this version on Friday; sign a revised one. Re-run on FY26 deals, the plan costs $15.8M against the $15.2M US budget, pays less for the multi-year service agreements we named as a priority, and covers 214 of our 245 reps: Mexico isn't modelled yet. · Inputs 1, 3, 4, 5

WHO GAINS, WHO LOSES (US REPS, FY26 DEALS)

Top 43 by bookings+22%
None earn less
Other 135 equipment+2%
3 earn less
36 service reps-14%
29 earn less, incl. 8 of the top 20 by margin
31 Mexico repsNot modelled
Extract due January 8

BEFORE WE SIGN, AND BEFORE JANUARY 11

FIX FIRST
Service crediting: pay for the multi-year agreements we want
First-year-only crediting cuts service reps' credited bookings by about a third; their quotas fall 25%. Ask Ben for an option that rewards contract term by Thursday, December 17. · Inputs 1, 2, 4
CORRECT
The summary: cost, accelerators and the 85%
On FY26 deals, $15.8M is over the $15.2M budget, before Mexico. The 2x accelerator starts at 110%, and 41 reps finished FY26 between 100% and 110%. The 85% counts US reps only. · Inputs 2, 3, 4, 5
HR AND COUNSEL CONFIRM
Notice period and Mexico
Plan terms give 30 days' written notice; January 11 to February 1 is 21 days. Marta confirms how the 30 days count. Mexico's counsel reviews before anything is communicated there. · Input 5
Caught on review
1
"Top performers" means top by bookings; 8 of the 20 highest-margin reps earn less
2
"Cost-neutral"? $15.8M modelled against a $15.2M budget, before Mexico
3
"85% earn the same or more" counts 214 US reps; 29 of 36 service reps earn less
UC-001 · SC-29 · Rev A
FOR
VP Sales, North America
SECTOR
Electrical equipment manufacturing
COMPANY
15,000 people · 14 sites · US, Mexico
TASK
Check next year's commission plan before the CRO and CHRO sign it
WRITTEN FOR
ChatGPT · Claude · Gemini · Copilot
AI TEST
Claude, Nov 14, 2026 · 8 of 8 found
Step 0 · The mess

What landed on Elena's desk

A short email from the CRO, a confident plan summary, the modelling by group, and two notes from sales operations and HR. Every average in the table is right. The claims built on them are another matter.

EMAIL · CRO[1]
Check it before we sign
“Our FY27 priorities are new customers and growing multi-year service agreements.”
1
The yardstick.
PLAN SUMMARY[2]
Sales Compensation
“Top performers earn 22% more. 85% of reps earn the same or more. Cost-neutral.”
2
Three claims.
MODELLING[3]
FY26 deals re-run
Aftermarket & service reps: 36 … -14% … 29 earn less.
3
Who's in the 85%?
SALES OPS[4]
Mechanics
1x applies from 100% to 110%, and 2x above 110% … Mexico is not modelled yet.
4
The small print.
HR NOTE[5]
Budget and terms
Budget $15.2M … 30 days' written notice … counsel has not reviewed.
5
Dates and sign-offs.
You're not alone
97%
of companies in Alexander Group's sales compensation survey changed their plans for 2026. Only about one in five rated the previous year's plans very effective.
Alexander Group, 2026 Sales Compensation Trends Survey, companies in 11 industries (sample size not published). Source
Steps 1–3 · How it works

Three steps, one chat

Here the AI checks a pay plan's claims against its own modelling. The prompt makes it say who every percentage counts.

1
Open a new chat

Use the AI tool your company approves: ChatGPT, Claude, Gemini or Copilot.

2
Paste the prompt and the inputs

Copy the prompt and the five fictional inputs from the appendix into one message.

3
Check one date yourself

Count the days from the announcement to the start date, and look up the weekday of any deadline the AI gives you. Then decide what to lead with.

The prompt, part by part
Copy-ready version in the appendix

I'm VP Sales for North America at an electrical equipment manufacturer, with 245 quota-carrying reps in the US and Mexico. Our CRO and CHRO sign off next year's commission plan on Friday, December 18, and we announce it at sales kickoff on Monday, January 11. Today is Monday, December 14. The CRO has asked me to check the plan holds up before they sign. Below are his email, the plan summary, the modelling, a sales operations note and an HR note.

1
Who you are, who signs, the sign-off and announcement dates, and today's date.

Please give me:

  1. Each claim in the summary: holds, doesn't hold, or unverified, with the input that shows it.
  2. Who gains and who loses, by group, and what the averages hide.
  3. Does the plan pay for what we said we want next year?
  4. What must happen before sign-off and before the announcement, and who confirms each item.
  5. My recommendation to the CRO and CHRO, in five lines or fewer.
2
A verdict on each claim with its source, who gains and who loses, whether the plan pays for the priorities, what must happen first and who confirms, and a five-line recommendation.

For every percentage, say who it counts and who it leaves out. Check the arithmetic. Don't add facts that aren't in the inputs; if you assume something, say so. Don't give legal conclusions: say what needs legal or HR confirmation.

3
The heavy lifting: for every percentage, say who it counts and who it leaves out. It found all eight planted problems.
Where to paste
Every tool has one message box. Paste the prompt, add a blank line, paste the inputs, send. Modelling spreadsheets can be attached as files if your tool allows it.
Safe to try
Practise with the fictional inputs first. Real pay data is confidential and names people: use only your company's approved tool, and use group figures, not individuals.
The AI's first answer

What Claude actually wrote

Exactly as it came back, not edited. Read it the way Elena did: check each point against the source it cites.

UNEDITED
Claude · fresh session, no other context · Nov 14, 2026 · one run

FY27 commission plan check (for Elena, VP Sales North America)

Arithmetic check: the group averages and totals in the modelling [3] add up. 43 + 135 + 36 = 214 US reps, and with Mexico's 31 that makes 245. FY26 plan total: 43×$112,000 + 135×$56,600 + 36×$68,000 = $14,905,000. FY27 plan total: 43×$136,600 + 135×$57,700 + 36×$58,500 = $15,769,300. The problems are in what the summary [2] claims on top of these figures.

1. Each claim in the summary

Claim [2] Verdict Evidence
Accelerators raised from 1.5x to 2x Doesn't hold as written [4]: in FY26, 1.5x paid above 100% of quota. In FY27, 1x pays from 100% to 110%, and 2x only above 110%. So between 100% and 110% the rate drops from 1.5x to 1x. 41 US reps finished FY26 in that band. The summary leaves out the threshold change. found
New-customer bookings credited at 1.25x Unverified It is a plan design statement. Nothing in [3] or [4] shows whether the model applied it, or how many FY26 deals would have qualified.
Service: first-year value only, "quotas reduced to match" Doesn't hold [4]: about half of service reps' credited bookings were multi-year, average 3 years. Crediting year 1 only gives 0.5 + 0.5×(1/3) = 0.67, so credit falls about a third. Quotas fall only 25%. Attainment becomes 0.67 / 0.75 = 89% of before, so a rep at 100% of quota in FY26 lands at about 89%. Matching the credit cut would need a quota cut of about 33%. found
Top performers earn 22% more Holds only narrowly $112,000 to $136,600 = +$24,600, +21.96%. Who it counts: the 43 US equipment reps in the top 20% by FY26 bookings, and it is their average. Who it leaves out: service reps and Mexico. "Top 20%" is also unclear: 43 is 20% of all 214 modelled reps, but 24% of the 178 equipment reps the row label names. "Top" means bookings, not margin. Of the 20 US reps with the highest FY26 gross margin, 8 are service reps and all 8 earn less [4]. found
85% of reps earn the same or more Holds for US only 182 of 214 = 85.0%. Who it counts: the 214 modelled US reps. Who it leaves out: the 31 Mexico reps (13% of 245), who are not modelled. Across all 245 reps, the share known to earn the same or more is 182 / 245 = 74%. "The same or more" also hides that 135 reps gain only about $1,100 on average. found
Cost-neutral, within budget Doesn't hold FY27 plan $15.77M against a $15.2M US budget [5]: about $0.57M (3.7%) over. That is US only, and Mexico would be extra. It is +5.8% over the FY26 plan re-run ($14.9M), against a budget set at +2%. It also rests on FY26 deals; FY27 cost moves with FY27 results. found
Takes effect Monday, February 1 Conflicts with notice terms [5]: reps get 30 days' written notice before a change takes effect. An announcement on January 11 is 21 days before February 1. HR needs to confirm how the notice applies (see 4). found

There is also an internal inconsistency to reconcile, not a conclusion. 41 reps lose the 1.5x rate on their 100% to 110% band, yet only 3 equipment reps earn less in the model. Possible explanations are the new-customer bonus offsetting the loss, or most of the 41 being top-20% reps, but the inputs don't say. Priya should confirm.

2. Who gains and who loses (US modelled reps only; Mexico unknown)

  • Top 43 equipment reps: +$24,600 each on average, none lose. As a group they get +$1,057,800, which is more than the whole net increase of $864,300 (122% of it).
  • Other 135 equipment reps: +$1,100 each on average (+2%), 3 lose. Group total +$148,500.
  • 36 aftermarket and service reps: -$9,500 each on average (-14%), and 29 of 36 (81%) lose. Group total -$342,000. found

What the averages hide:

  • The service group's -14% is an average that includes the 7 reps who don't lose. If those 7 are flat or up, the 29 who lose must average more than -14%. The individual figures aren't given (unverified).
  • 8 of the 20 highest-margin reps are among the losers.
  • The "all modelled" row leaves the averages blank. They work out at about $69,650 under FY26 and $73,690 under FY27 (+5.8%), and that figure blends a large gain at the top with a cut for service.
  • No individual distribution, minimums or maximums are given.

3. Does the plan pay for what we said we want?

The two FY27 priorities are new customers and growing multi-year service agreements [1].

  • New customers: partly. The 1.25x credit points the right way, but its effect isn't shown in the modelling (unverified).
  • Multi-year service agreements: no. First-year-only crediting means a 3-year agreement earns the same credit as a 1-year one, so the plan removes the reward for the term we say we want to grow. The people who sell these agreements take the cut, with 29 of 36 earning less. found
  • Where the money goes: the 2x accelerator above 110% sends most of the extra pay to the top bookings reps. Neither priority is about that.

4. What must happen, and who confirms

Before sign-off on Friday, December 18:

  1. Reconcile the cost against the budget ($15.77M vs $15.2M, US only), or change the design. Ben Harlow (Comp) and Marta Ochoa (HR) confirm; Finance should confirm the budget figure. I am assuming Finance owns the budget; the inputs don't say.
  2. Fix or explain the service quota cut: 25% against a credit cut of about 33%. Also decide how multi-year agreements are rewarded, given the stated priority. Ben and Priya Nair (Sales Ops) confirm; Dev decides.
  3. Reconcile the 41 reps in the 100% to 110% band with the 3 equipment losers, and confirm whether the 1.25x bonus is in the model. Priya confirms.
  4. Correct the summary's wording on accelerators, "top performers" (bookings, US, equipment only), the 85% (US only) and cost-neutral. Ben owns this.
  5. Decide whether to sign for 245 reps while 31 are unmodelled, or sign for the US only. Dev and Ruth decide.

Before the announcement on Monday, January 11:

  1. Notice: 30 days' written notice before February 1 means notice by about January 2 if counted in calendar days. Alternatively, keep January 11 and move the effective date to February 10 or later. HR (Marta) confirms how the notice term applies; legal should confirm if HR advises it. A: check the weekday
  2. Mexico: the extract arrives January 8, leaving one working day before kickoff. Counsel's standing advice is that any change that can lower commission is reviewed before it is communicated, and this plan lowers pay for some reps. Mexico can't be announced on January 11 until counsel has reviewed the plan. Marta and Mexico employment counsel confirm. Mexico's cost also has to be added to the budget check. found

5. Recommendation to Dev and Ruth

  1. Don't sign Friday as drafted: it runs about $0.57M over the $15.2M US budget, and Mexico isn't costed. B: lead with the priority
  2. The service change works against our own priority: it removes credit for multi-year terms, and quotas fall 25% while credit falls about 33%, so 29 of 36 service reps, including 8 top-margin reps, earn less.
  3. The summary overstates "2x accelerators" (no accelerator at all from 100% to 110%), and "85%" and "22%" cover US reps only.
  4. A January 11 announcement doesn't give 30 days' notice before February 1, and Mexico needs counsel review before anything is communicated. HR and counsel to confirm.
  5. Ask Ben for a revised version this week covering the service crediting/quota fix, the cost against budget and the notice timing. Sign the US plan once those are confirmed, and Mexico after counsel review. C: a date, and the right fix

Everything above rests only on inputs [1] to [5]. Assumptions are labelled, and legal points are flagged for HR or counsel rather than concluded.

foundcaught a planted problemA–CElena changes it; see the next section
The human part

What Elena changed, and why

Claude found every planted problem, checked the arithmetic and kept legal points with HR and counsel. Elena's edits check one date and decide what the CRO hears first.

What Claude found: 8 of 8 planted problems
One run
✓“Top” means top by bookings
✓Accelerator threshold moved
✓Average hides the service cut
✓Plan works against a priority
✓Over budget
✓85% is US reps only
✓Mexico counsel review
✓Notice period
Three edits, one check and two judgment calls
A
CLAUDE WROTE
“30 days' written notice before February 1 means notice by about January 2 if counted in calendar days.”
ELENA'S VERSION
January 2 is a Saturday, and the plan isn't final yet. Elena asks Marta to confirm how the 30 days count and gives the CRO two options, if the 30 days are calendar days: announce on January 11 and start the new plan no earlier than Wednesday, February 10, or keep February 1 and give written notice by Thursday, December 31, the last working day before January 2.
WHY · ACCURACY
A deadline that falls on a weekend isn't a plan. Checking the calendar turns "about January 2" into a choice the CRO and CHRO can make.
B
CLAUDE WROTE
“Don't sign Friday as drafted: it runs about $0.57M over the $15.2M US budget, and Mexico isn't costed.”
ELENA'S VERSION
Elena leads with the CRO's own priority: the plan pays less for multi-year service agreements, and the people who sell them, including 8 of the top 20 reps by margin, take the cut. The budget comes second.
WHY · PRIORITY
A budget gap can be closed by trimming. A plan that works against the strategy it's meant to drive is the point Dev and Ruth need to hear first.
C
CLAUDE WROTE
“Ask Ben for a revised version this week covering the service crediting/quota fix, the cost against budget and the notice timing.”
ELENA'S VERSION
Elena asks Ben for the revised version by Thursday, December 17, the day before sign-off, with an option that rewards contract term. A deeper quota cut would restore service reps' pay, but a 3-year agreement would still earn no more than a 1-year one.
WHY · DECISION
The quick fix restores the numbers, not the behaviour the CRO asked for. And an ask without a date leaves Friday's version as the default.
What only you can do
Check the weekday, lead with what the plan is for, and put a date on the fix. The AI found who each number counts. You decide what the people who sign it hear first.
Your turn

Make it yours

Use it whenever a pay or incentive change is summarised for you in averages: a commission plan, a bonus scheme, a quota change, a territory redesign.

Your prompt: fill in the brackets

I'm [your role] at a [industry] company, with [how many people] affected by [the plan]. [Who] signs it off on [date], and we announce it on [date]. Today is [date]. Below are [the summary, the modelling, the mechanics, the budget and any notice or legal terms].

Please give me: 1. Each claim in the summary: holds, doesn't hold, or unverified, with the input that shows it. 2. Who gains and who loses, by group, and what the averages hide. 3. Does the plan pay for what we said we want? 4. What must happen before sign-off and before the announcement, and who confirms each item. 5. My recommendation, in five lines or fewer.

For every percentage, say who it counts and who it leaves out. Check the arithmetic. Don't add facts that aren't in the inputs; if you assume something, say so. Don't give legal conclusions: say what needs legal or HR confirmation.

Five checks before you sign
1
Every percentage says who it counts and who it leaves out.
2
Each group that earns less is named, with how many people.
3
The plan pays more for what the strategy asks for, not less.
4
The modelled cost is set against the budget, for every country.
5
Notice and legal reviews fit the dates, on working days.
Make it repeatable (optional)
Save the prompt once and reuse it every plan cycle: a custom GPT in ChatGPT, a project in Claude, a Gem in Gemini, or an agent in Copilot. Some companies switch these off; copy and paste always works.
With real data
Pay plans and earnings are confidential. Use only the AI tool your company approves, follow its rules, and work with group figures rather than named people.
What this example shows
One run of this prompt in Claude found all eight problems planted in fictional inputs and kept legal points for HR and counsel.
What it doesn't show
That every tool, setting or run will do the same, how much time it saves you, or what a good commission plan looks like. Ostrander Electric and its people are fictional.
Appendix

Copy-ready: the prompt and the five inputs

Copy the prompt, then all five inputs, into one message in a new chat. Everything here is fictional and safe to paste.

The prompt

I'm VP Sales for North America at an electrical equipment manufacturer, with 245 quota-carrying reps in the US and Mexico. Our CRO and CHRO sign off next year's commission plan on Friday, December 18, and we announce it at sales kickoff on Monday, January 11. Today is Monday, December 14. The CRO has asked me to check the plan holds up before they sign. Below are his email, the plan summary, the modelling, a sales operations note and an HR note.

Please give me:

  1. Each claim in the summary: holds, doesn't hold, or unverified, with the input that shows it.
  2. Who gains and who loses, by group, and what the averages hide.
  3. Does the plan pay for what we said we want next year?
  4. What must happen before sign-off and before the announcement, and who confirms each item.
  5. My recommendation to the CRO and CHRO, in five lines or fewer.

For every percentage, say who it counts and who it leaves out. Check the arithmetic. Don't add facts that aren't in the inputs; if you assume something, say so. Don't give legal conclusions: say what needs legal or HR confirmation.

[1] Email from Dev Malhotra, CRO

Elena, Ben's FY27 commission plan summary is below. Our FY27 priorities are new customers and growing multi-year service agreements. Ruth and I sign it off Friday, and we announce at sales kickoff on Monday, January 11. Can you check it holds up before we sign?

[2] FY27 commission plan: summary (Ben Harlow, Sales Compensation)

What changes from FY26: - Accelerators raised from 1.5x to 2x. - New-customer bonus: bookings with new customers credited at 1.25x. - Simpler crediting for service agreements: first-year value only (FY26: full contract value, up to 3 years). Service quotas reduced to match.

What it does: - Top performers earn 22% more. - 85% of reps earn the same or more. - Cost-neutral: within the FY27 variable pay budget.

The plan takes effect at the start of FY27, Monday, February 1.

[3] Modelling extract: FY26 deals and quotas, re-run under the FY27 plan (US reps)
Group Reps Avg variable pay, FY26 plan Avg variable pay, FY27 plan Change Earn less under FY27
Top 20% by FY26 bookings (all equipment reps) 43 $112,000 $136,600 +22% 0
Other equipment reps 135 $56,600 $57,700 +2% 3
Aftermarket & service reps 36 $68,000 $58,500 -14% 29
All modelled 214 32

Total variable pay: FY26 plan $14.9M; FY27 plan $15.8M.

[4] Sales operations note (Priya Nair)
  • Accelerators: in FY26 the 1.5x rate applied above 100% of quota. In FY27, 1x applies from 100% to 110%, and 2x above 110%. 41 US reps finished FY26 between 100% and 110%.
  • About half of service reps' FY26 credited bookings were multi-year agreements (average term 3 years). First-year-only crediting cuts their credited bookings by about a third; their quotas are cut 25%.
  • Of the 20 US reps with the highest FY26 gross margin, 8 are aftermarket & service reps. All 8 earn less under FY27.
  • Mexico (31 reps) is not modelled yet. Its FY26 data is in a separate CRM; extract due January 8.
[5] HR note (Marta Ochoa, HR Business Partner, Sales)
  • FY27 variable pay budget for US sales: $15.2M (FY26 actual plus 2%).
  • Our plan terms give reps 30 days' written notice of a plan change before it takes effect.
  • Mexico: our employment counsel there has not reviewed the plan. Their standing advice is that any change that can lower a rep's commission must be reviewed before it is communicated.