deal-value-calc

{
  "skill_name": "deal-value-calc",
  "evals": [
    {
      "id": 1,
      "prompt": "Selling our claims-automation platform into Northbeam Logistics, 6-person freight claims team. Their champion Dara pulled the numbers off their ops dashboard: 3,100 claims a month, 11 minutes of manual touch time per claim (she ran a stopwatch study on her own team last quarter), and HR confirmed the claims clerks are on $48,000 base. Our quote is $54,000/year plus $12,000 implementation, and Dara reckons her team will burn about 90 hours on rollout. From our other rollouts I'd say touch time drops to about 3 minutes. Our marketing team says companies like them typically cut claim errors by 30%. I don't have anything on what a claim error actually costs them. Build me the ROI case, Dara wants to take it to their VP Finance.",
      "expected_output": "A math block with a labelled input ledger, a labour-time driver computed with loading and a realization haircut, the error/rework driver dropped for want of a buyer-supplied cost per error, full TCO, ROI with its convention stated, three scenarios conservative first, a rep-assumed disclosure line, and a narrative one-pager alongside the math.",
      "files": [],
      "expectations": [
        "Uses the stated $48,000 base salary and labels it buyer-supplied.",
        "Converts base salary to a fully loaded cost using a multiplier in the 125-135% range, and labels that multiplier as a benchmark with its source named or as rep-assumed if no source is named.",
        "Labels the 3-minute post-implementation touch time as rep-assumed.",
        "Treats the 30% claim-error reduction as rep-assumed rather than a benchmark, because no source is named for it.",
        "Produces no numeric cost per claim error; either asks the buyer for it or inserts an explicitly labelled placeholder.",
        "Drops the error/rework driver from the computed case rather than modelling it with an assumed cost per error.",
        "Applies a realization haircut in the 60-70% range to the labour-time-saved driver and states it.",
        "Computes total investment including the $12,000 implementation and the 90 internal hours valued at a loaded hourly rate, not the $54,000 subscription alone.",
        "States which ROI convention the document uses, net-benefit or gross.",
        "Presents three scenarios with the conservative case first.",
        "Includes a visible disclosure line stating the headline rests in part on rep-assumed inputs, naming which ones.",
        "Delivers both a math block with an input ledger carrying value, unit, source and provenance label, and a narrative one-pager, not an ROI figure alone."
      ]
    },
    {
      "id": 2,
      "prompt": "Velostrom Utilities, support-automation deal. Their helpdesk export shows 22,000 tickets a year at a 14-minute average handle time across a 14-agent team, and the champion says a fully loaded agent costs them $61,000. I think we deflect about 35% of those tickets. He also shared their approved FY27 hiring plan: 4 more agents budgeted for the same team because ticket volume is growing 30%. So that's $244,000 of hires they don't make, plus all the hours we save them. Add it up for me, they want a number by Friday.",
      "expected_output": "A case that catches the shared-capacity double count between labour time saved and headcount avoidance, keeps headcount avoidance out of the hard headline, treats $61,000 as already loaded, labels the 35% deflection rate rep-assumed, and runs three unweighted scenarios.",
      "files": [],
      "expectations": [
        "Identifies that labour time saved and headcount avoidance both draw on the same 14-agent support team.",
        "Does not total the $244,000 headcount avoidance and the full deflected-hours value together at face value; splits the shared capacity between the two drivers or drops one.",
        "Places headcount avoidance in a soft or cost-avoidance section, excluded from the headline number.",
        "Bases the headline on hard value only and states that the case was tested on hard value alone.",
        "Treats the $61,000 as already fully loaded and applies no further loading multiplier to it.",
        "Labels the 35% deflection rate as rep-assumed.",
        "Holds the 22,000 annual ticket volume and the 14-minute handle time fixed in the conservative case as buyer-supplied.",
        "Haircuts the 35% deflection rate hardest in the conservative case.",
        "Attaches no probability, likelihood or weight to any of the three scenarios.",
        "States the double-counting rule explicitly: one team's hours back at most one driver."
      ]
    },
    {
      "id": 3,
      "prompt": "Need a strong ROI case for Thursday's CFO meeting. Platform is $96,000/year plus $25,000 implementation. The only hard thing the buyer could give me is that they'd retire two tools, one at $19,000/year and one at $12,000/year, and their ops lead said about 200 hours of internal effort at roughly $40/hour loaded, which he pulled out of their finance system. Everything else they talk about is better collaboration across teams and people being less frustrated. Make it compelling.",
      "expected_output": "A run that computes $31,000 of hard annual value against $96,000 of recurring cost, reports the negative net, declares payback undefined, and warns the user the case fails on hard value before writing any narrative, without inventing a driver to close the gap.",
      "files": [],
      "expectations": [
        "Totals annual hard value at $31,000 from the two retired tools and compares it against the $96,000 annual recurring cost.",
        "Reports the result as negative annual net value rather than a positive ROI percentage.",
        "States payback as not paying back within the horizon instead of producing a payback figure.",
        "Tells the user the case fails on hard value alone before producing any polished narrative one-pager.",
        "Adds no invented revenue, productivity or efficiency figure to close the gap.",
        "Keeps better collaboration and reduced frustration in a separate soft-value section, excluded from the headline.",
        "Counts the 200 internal hours at $40/hour in the Year-1 investment alongside the $25,000 implementation.",
        "Names which specific buyer-supplied input would have to exist for the verdict to change.",
        "Does not present the deal as having a compelling ROI for the CFO meeting."
      ]
    },
    {
      "id": 4,
      "prompt": "$310,000 single-year deal, biggest of my career. Buyer is Hal Vreeland, owner-operator of a 40-person plumbing distributor. He signs personally, there's no procurement, no security review, nobody else in the room. He wants an answer by Friday. My VP says build the full enterprise business case with sensitivity analysis and the finance appendix. Where do I start?",
      "expected_output": "A recommendation that sizes the artifact by complexity rather than deal value, lands on a lighter rung than the enterprise model, and prunes drivers that cannot be sourced before Friday.",
      "files": [],
      "expectations": [
        "States that artifact size is set by deal complexity - stakeholder count, procurement or security in the path, strategic versus tactical spend - not by deal value.",
        "Explicitly rejects the $310,000 figure as the reason to build the full enterprise model.",
        "Recommends a lighter artifact than the full enterprise model and names the rung: a payback estimate, or a one-page summary plus light model.",
        "Notes that this deal crosses no complexity trigger: single decider, no procurement, no security review.",
        "Drops any value driver requiring a time study or a cohort analysis, given the Friday deadline.",
        "Promotes drivers readable straight off documents the buyer already holds, such as the contract or renewal list and the hiring plan.",
        "Asks who the document gets forwarded to, or whether a finance function or lender will review it.",
        "Still leads with a conservative case rather than a single point estimate.",
        "Does not recommend producing a separate finance or procurement appendix version for this deal."
      ]
    },
    {
      "id": 5,
      "prompt": "Residential solar, Arizona. Met Dana and Rob door-knocking last Tuesday. Their last 12 utility bills add up to $3,480. Our financing is $189 a month for 240 months. Our design tool says 92% offset. Marketing wants every rep leading with 'save 40-70% on your power bill'. Build me the value case to leave with them. Rob is the skeptical one, Dana's the one who called us back.",
      "expected_output": "A consumer value case that refuses the unsubstantiated savings headline, names the consumer-protection constraints, discloses total financed cost, leads with a conservative offset scenario, and labels the B2B framework mapping as structural analogy.",
      "files": [],
      "expectations": [
        "Refuses to use 'save 40-70% on your power bill' as the headline claim.",
        "Names the substantiation duty for consumer savings claims under FTC Act Section 5.",
        "Names at least one further consumer-protection constraint beyond FTC Section 5: the three-business-day cooling-off right on the door-to-door sale, lending-disclosure requirements, or state solar-disclosure statutes.",
        "Discloses the total financed cost and the financing terms, not the $189 monthly payment alone.",
        "Presents a payback range with its assumptions disclosed instead of a single savings headline.",
        "Labels the 92% production offset as installer or vendor data, treated as rep-assumed until substantiated, not as a buyer-supplied figure.",
        "Uses the $3,480 of actual utility bills as the buyer-supplied baseline.",
        "Leads with a conservative case built on a lower offset and no assumed utility-price inflation.",
        "Does not headline the optimistic case.",
        "Labels the mapping of B2B business-case roles onto this household purchase as a structural analogy rather than established practice.",
        "Identifies Rob as the skeptical household member the document must convince without the salesperson present, and the lender whose approval gates the purchase."
      ]
    },
    {
      "id": 6,
      "prompt": "Tessera Robotics deal, warehouse scheduling software. Their WMS export shows 120,000 pick-hours a year across 3 sites, and payroll says $27/hour loaded for pickers. I figure we take 4% off that. Our price is $88,000/year plus $20,000 implementation, and their ops manager says 150 hours of internal effort. For the board slide I want expected value: weight the three cases 20/60/20 and give me one blended ROI number.",
      "expected_output": "A refusal to produce a probability-weighted blended ROI, three unweighted cases with the conservative first, one-at-a-time sensitivity on the inputs that move the result, and full TCO.",
      "files": [],
      "expectations": [
        "Declines to produce a single probability-weighted or blended ROI number.",
        "States that the three cases are defensibility bands, not a probability distribution.",
        "Attaches no 20%, 60% or 20% weight, nor any other likelihood, to the three cases.",
        "Presents the conservative case first.",
        "Adds one-at-a-time sensitivity analysis naming the two or three inputs the result is most sensitive to.",
        "Shows the swing in annual value when the top input moves between a stated low and a stated high.",
        "Labels the 4% pick-hour reduction as rep-assumed.",
        "Haircuts the 4% reduction hardest in the conservative case while holding the 120,000 hours and $27/hour fixed as buyer-supplied.",
        "Applies and states a realization haircut on the labour-time driver.",
        "Includes the 150 internal effort hours, valued at a loaded rate, in total investment alongside the $20,000 implementation."
      ]
    },
    {
      "id": 7,
      "prompt": "Lumen Freight Systems. Our marketing deck says customers see 3.5x ROI with payback under 6 months, and their CFO asked for the math behind it. What I've got from the buyer is 5 route planners, each spending about 6 hours a week on manual rework, which came from their ops lead. Price is $72,000/year. Their CFO told my champion anything over a 12-month payback doesn't get approved, so I want the model to land right at 12 months.",
      "expected_output": "A case that refuses to reuse vendor deck figures as the prospect's expected result, refuses to engineer the payback to 12 months, treats the CFO's bar as a hurdle to confirm, and leaves the loaded hourly cost as a labelled gap.",
      "files": [],
      "expectations": [
        "Does not present the 3.5x ROI or the sub-6-month payback from the marketing deck as this prospect's expected result.",
        "Labels the deck figures as vendor data rather than as a benchmark or an expected outcome.",
        "Refuses to back-solve the model so payback lands at 12 months.",
        "States that a benefit stream landing exactly on a clean threshold reads as manufactured to a finance reviewer.",
        "States no payback threshold as a universal rule.",
        "Treats the CFO's 12-month bar as the buyer's own hurdle to be confirmed with them, not as a target to engineer toward.",
        "Invents no fully loaded hourly cost for the route planners; asks for it or inserts a labelled placeholder.",
        "Shows the raw arithmetic from 5 planners x 6 hours/week through to annual value rather than a rounded headline.",
        "Computes investment as full TCO rather than the $72,000 subscription alone."
      ]
    },
    {
      "id": 8,
      "prompt": "Buyer says our tool takes their win rate from 22% to 26%. They close about $14M of new business a year. They won't give me gross margin, said it's confidential. Their churn is 9% and they reckon we'd get it to 6%. Build the case and lead with the revenue number, it's by far the biggest thing here.",
      "expected_output": "A case that deletes both revenue lift and churn outright because gross margin is withheld, refuses a top-line model and any substituted margin, and redirects to hard drivers with the specific inputs needed.",
      "files": [],
      "expectations": [
        "Does not model the revenue lift on top-line revenue.",
        "Removes the revenue-lift driver from the case entirely because gross margin is withheld, rather than demoting it or parking it as soft value.",
        "Removes the churn or retention driver for the same reason.",
        "Substitutes no industry-average or otherwise assumed gross margin.",
        "Produces no annual value figure of the $560,000 kind derived from $14M multiplied by 4 percentage points.",
        "Declines to lead the headline with the revenue number.",
        "Names attribution as the objection finance uses to discount both revenue lift and churn.",
        "States that supplying a gross-margin figure, even as a range, is what would bring those drivers back into the case.",
        "Redirects to hard drivers the buyer can actually feed and asks for the specific inputs those need."
      ]
    },
    {
      "id": 9,
      "prompt": "Math's done and the champion signed off on the assumptions. Hard annual value $184,000, conservative case $121,000, Year-1 investment $63,500, payback 6.3 months conservative. Their FY27 cost-to-serve program is called Project Clearwater. Now write the one-pager. Make it pop: our logo at the top, list the five features that drive the savings, and address it to Priya since she's the one who asked for it. Their CFO Marcus will see it after her.",
      "expected_output": "An unbranded five-part one-page case written through Priya to Marcus, tied to Project Clearwater, carrying the conservative figure, the cost of inaction, both sides' investment, and next steps - with no feature list.",
      "files": [],
      "expectations": [
        "Produces all five parts: priority-driven headline, problem statement, recommended approach, target outcomes, required investment.",
        "Ties the headline to Project Clearwater.",
        "Contains no feature list; the recommended-approach section describes the shift in practice instead.",
        "Strips the vendor logo and branding from the document.",
        "Writes the document through Priya to Marcus rather than addressing it to Priya.",
        "Produces a document self-explanatory to a reader who never met the rep.",
        "States the cost of doing nothing in the buyer's own units.",
        "Leads the value statement with the $121,000 conservative figure rather than the $184,000 figure.",
        "Aims the page at costs, time to value, and confidence in the initiative.",
        "States required investment from both sides, vendor and buyer, with a timeline.",
        "Ends with next steps that include champion edit rounds and written assumption sign-off."
      ]
    },
    {
      "id": 10,
      "prompt": "Deal's at proposal, verbal yes, legal's reviewing the MSA. My manager wants an ROI one-pager as a leave-behind for the signature meeting next week. He also read somewhere that getting the buyer to edit the business case three times triples your win rate. Is that real? And what should I be tracking on this stuff?",
      "expected_output": "A correction of the tripling claim as unreplicated, a flag that the case is arriving late-stage as a leave-behind, a usable path for this deal anyway, and a concrete KPI set including the no-decision baseline.",
      "files": [],
      "expectations": [
        "States that the claim about three edit rounds tripling win rates is unreplicated, and does not present it as established.",
        "Recommends counting edit rounds completed as the metric instead of promising a win-rate lift.",
        "Flags that producing the case at proposal stage as a leave-behind is late, and that it belongs right after the first discovery call.",
        "Still gives a usable path for this specific deal rather than only criticising the timing.",
        "Names concrete KPIs including the buyer correcting an input, edit rounds completed, written assumption sign-off, and the champion forwarding the document.",
        "Cites the 40-60% no-decision rate for qualified B2B deals with its source, Dixon and McKenna, HBR 2022 / The JOLT Effect.",
        "Includes the forward ask: asking the champion whether they are willing to forward a short summary.",
        "States no specific win-rate uplift percentage as an expected outcome."
      ]
    },
    {
      "id": 11,
      "prompt": "Enterprise deal: 9 stakeholders, procurement running a formal evaluation against two competitors, security review in flight. Their finance team wants NPV over three years. Is 10% the standard discount rate to use? Also what exactly do I hand procurement versus what I hand the SVP who sponsors this?",
      "expected_output": "A refusal to assume the discount rate, an instruction to ask the buyer's finance for their own hurdle rate, two distinct document versions, a three-year cash-flow view with NPV, and driver weighting tuned to a procurement-led evaluation.",
      "files": [],
      "expectations": [
        "Does not adopt 10% as the discount rate to use.",
        "Instructs asking the buyer's finance function for their own hurdle rate or cost of capital.",
        "Notes that a 10% default exists in one commercial methodology but is not this buyer's number.",
        "Produces two distinct documents: an economic-buyer summary for the SVP sponsor and a separate finance or procurement version with an assumption appendix.",
        "States the NPV calculation as the sum of discounted net cash flows per year minus the initial investment.",
        "Shows per-year value and cost across the three-year horizon plus the cumulative net.",
        "Weights direct cost and TCO comparison hardest, given the procurement-led evaluation.",
        "Gives soft drivers roughly no weight in the procurement-facing version.",
        "Includes sensitivity analysis and a documented source for every input in the procurement version.",
        "Recommends engaging a value engineer or business-value consultant if the selling organisation has one."
      ]
    }
  ],
  "trigger_queries": [
    { "query": "Build the business case for the Meridian deal", "should_trigger": true },
    { "query": "What's the ROI on this deal?", "should_trigger": true },
    {
      "query": "The CFO wants to see the math before Thursday, help me make this defensible",
      "should_trigger": true
    },
    { "query": "How long until it pays for itself?", "should_trigger": true },
    {
      "query": "I need a TCO comparison for the proposal, not just our license price",
      "should_trigger": true
    },
    {
      "query": "Put together a one-pager my champion can forward to their exec team",
      "should_trigger": true
    },
    { "query": "Their finance team wants NPV over three years", "should_trigger": true },
    {
      "query": "help me justify the price to their CFO with actual numbers",
      "should_trigger": true
    },
    { "query": "quantify what their manual process is costing them", "should_trigger": true },
    {
      "query": "$30k/year subscription, they say it saves each rep 4 hours a week, show me the math",
      "should_trigger": true
    },
    { "query": "The buying committee asked for the financial impact", "should_trigger": true },
    {
      "query": "Model the savings from consolidating their three tools into ours",
      "should_trigger": true
    },
    { "query": "What's the cost of doing nothing here?", "should_trigger": true },
    {
      "query": "They want to see three scenarios before they'll sign anything",
      "should_trigger": true
    },
    {
      "query": "Build the value case for the solar quote I'm leaving with the homeowners tonight",
      "should_trigger": true
    },
    { "query": "my champion needs ammo for the budget committee", "should_trigger": true },
    { "query": "sanity check my savings calculation before I send it", "should_trigger": true },
    { "query": "Write the economic justification section of the proposal", "should_trigger": true },
    { "query": "Is this actually worth it for them?", "should_trigger": true },
    { "query": "They said prove it. now what", "should_trigger": true },
    { "query": "break down what this actually costs them in year one", "should_trigger": true },
    {
      "query": "Should I include the morale improvement in the number I show them?",
      "should_trigger": true
    },
    { "query": "Help me show the cost of waiting another year", "should_trigger": true },
    {
      "query": "what's a defensible payback estimate for a $120k enterprise software purchase",
      "should_trigger": true
    },
    {
      "query": "Their controller is reviewing our proposal, what does she need to see?",
      "should_trigger": true
    },
    { "query": "Turn these discovery notes into a financial argument", "should_trigger": true },
    { "query": "how do I show this saves them money", "should_trigger": true },
    { "query": "My rep built a savings model with made-up numbers, fix it", "should_trigger": true },
    {
      "query": "Procurement wants the full cost breakdown, not just the subscription",
      "should_trigger": true
    },
    { "query": "Draft the value summary for the renewal QBR baseline", "should_trigger": true },
    { "query": "Their VP asked what the return looks like over three years", "should_trigger": true },
    { "query": "I need numbers that survive a finance review", "should_trigger": true },
    { "query": "Can you check my business case before it goes to the CFO?", "should_trigger": true },
    {
      "query": "What assumptions do I need to get from the buyer to build this?",
      "should_trigger": true
    },
    { "query": "Work out what their downtime is costing them per year", "should_trigger": true },
    { "query": "Give me the hard vs soft savings split for this deal", "should_trigger": true },
    { "query": "The homeowner wants to know if the panels pay for themselves", "should_trigger": true },
    { "query": "I need a leave-behind that works when I'm not in the room", "should_trigger": true },
    { "query": "Build the financial story for this deal", "should_trigger": true },
    { "query": "How much will they actually save, conservatively?", "should_trigger": true },
    { "query": "payback period for this purchase?", "should_trigger": true },
    {
      "query": "Help me build the finance appendix for the enterprise proposal",
      "should_trigger": true
    },
    { "query": "They want hard numbers, not another demo", "should_trigger": true },
    { "query": "what should go in the investment section of the one-pager", "should_trigger": true },
    {
      "query": "12 hours a week saved across 40 people, what's that worth to them?",
      "should_trigger": true
    },
    {
      "query": "Our champion got burned last time because our numbers didn't hold up, how do I avoid that?",
      "should_trigger": true
    },
    {
      "query": "Make the case for why they should buy now instead of next fiscal year",
      "should_trigger": true
    },
    {
      "query": "Their CFO said anything over a year doesn't get approved, what do I do with that?",
      "should_trigger": true
    },
    {
      "query": "The prospect says we're too expensive compared to Coralline, what do I say on the call?",
      "should_trigger": false
    },
    {
      "query": "Plan my discount ladder for the renewal, what can I give and what do I get back?",
      "should_trigger": false
    },
    { "query": "Score this deal against MEDDPICC", "should_trigger": false },
    { "query": "Who's the economic buyer in the Halcyon account?", "should_trigger": false },
    { "query": "What's the TAM for warehouse robotics in the Nordics?", "should_trigger": false },
    {
      "query": "Design the accelerator curve for our AE comp plan next year",
      "should_trigger": false
    },
    {
      "query": "How much pipeline do I need to hit a $4M quota at a 23% win rate?",
      "should_trigger": false
    },
    { "query": "Set FY27 quotas for my six reps", "should_trigger": false },
    {
      "query": "Write me discovery questions to uncover the cost of their manual process",
      "should_trigger": false
    },
    { "query": "Review this call transcript and score the rep", "should_trigger": false },
    { "query": "Here are my notes on the Tarrant deal, is it at risk?", "should_trigger": false },
    {
      "query": "Write the recap email with next steps from yesterday's call",
      "should_trigger": false
    },
    { "query": "Our cold emails are all landing in spam", "should_trigger": false },
    { "query": "Give me 10 subject line variants for this campaign", "should_trigger": false },
    { "query": "What do I say in the first ten seconds when they pick up?", "should_trigger": false },
    {
      "query": "Build me a 12-touch outbound cadence across email and LinkedIn",
      "should_trigger": false
    },
    {
      "query": "Find me a personalization angle from this prospect's funding round",
      "should_trigger": false
    },
    { "query": "Define our ICP from last year's closed-won data", "should_trigger": false },
    { "query": "Where should the tier cutoffs sit for our named accounts?", "should_trigger": false },
    { "query": "Should we go product-led or sales-led?", "should_trigger": false },
    { "query": "Restructure my sales team into pods", "should_trigger": false },
    { "query": "How do I get promoted from SDR to AE?", "should_trigger": false },
    { "query": "Design the interview loop for an enterprise AE hire", "should_trigger": false },
    {
      "query": "Which sales podcasts and newsletters should I be following?",
      "should_trigger": false
    },
    { "query": "Price our new enterprise tier", "should_trigger": false },
    { "query": "Write the value proposition for our homepage", "should_trigger": false },
    { "query": "Estimate our monthly Azure spend for this architecture", "should_trigger": false },
    { "query": "Run a DCF on this stock and tell me if it's undervalued", "should_trigger": false },
    {
      "query": "Set purchasing-power-parity pricing for our app across 40 countries",
      "should_trigger": false
    },
    { "query": "What's our CAC payback period as a company?", "should_trigger": false },
    { "query": "Calculate LTV to CAC for our self-serve segment", "should_trigger": false },
    { "query": "Model our net revenue retention for the board deck", "should_trigger": false },
    { "query": "Forecast this quarter's revenue off my current pipeline", "should_trigger": false },
    { "query": "How do I justify three more headcount to my VP?", "should_trigger": false },
    { "query": "Calculate the commission owed on this closed-won deal", "should_trigger": false },
    { "query": "Should I give 15% off to close by end of quarter?", "should_trigger": false },
    { "query": "Handle the 'no budget until next year' objection", "should_trigger": false },
    { "query": "Draft the MSA redlines for their procurement team", "should_trigger": false },
    { "query": "Which accounts should I prioritize this quarter?", "should_trigger": false },
    {
      "query": "Write a customer case study about the results Fenwick got",
      "should_trigger": false
    },
    {
      "query": "Build the feature-by-feature comparison slide against Coralline",
      "should_trigger": false
    },
    { "query": "Our win rate dropped 8 points last quarter, diagnose it", "should_trigger": false },
    { "query": "Route this inbound lead to the right rep", "should_trigger": false },
    {
      "query": "What should we charge for a professional services implementation?",
      "should_trigger": false
    },
    { "query": "Score my resume for an enterprise AE role", "should_trigger": false },
    { "query": "Summarize this 40-page RFP and pull out the requirements", "should_trigger": false },
    { "query": "Plan the QBR agenda for our biggest account", "should_trigger": false },
    { "query": "Draft the mutual action plan dates with the buyer", "should_trigger": false }
  ]
}

# Fill-in templates

Fill every bracketed slot with a real value or a labelled placeholder the user must replace. Never fill a slot with an invented number.

## Table of Contents

- [1. Input ledger](#1-input-ledger)
- [2. Arithmetic block](#2-arithmetic-block)
- [3. Scenario table (conservative first)](#3-scenario-table-conservative-first)
- [4. Disclosure line (when required)](#4-disclosure-line-when-required)
- [5. One-page business case (five parts)](#5-one-page-business-case-five-parts)
- [6. Finance-reviewer appendix (enterprise, or whenever finance reviews)](#6-finance-reviewer-appendix-enterprise-or-whenever-finance-reviews)

## 1. Input ledger

| #   | Input                            | Value       | Unit        | Source (document, person, or named benchmark) | Provenance     |
| --- | -------------------------------- | ----------- | ----------- | --------------------------------------------- | -------------- |
| 1   | [e.g. invoices processed]        | [4,200]     | [per month] | [buyer's ops report, June]                    | buyer-supplied |
| 2   | [fully loaded cost multiplier]   | [1.3x base] | [ratio]     | [named benchmark source]                      | benchmark      |
| 3   | [post-implementation touch time] | [1]         | [min/unit]  | [rep estimate, unconfirmed]                   | rep-assumed    |

Rules:

- One row per raw input
- No derived figure appears here
- Every value-side row carries exactly one of the three provenance labels
- Cost-side rows (the rep's own pricing and fees) cite the written quote instead

If any rep-assumed row materially moves the headline, add the disclosure line (section 4).

## 2. Arithmetic block

Per driver, show the full chain:

```
Driver: [name]                                   [hard | soft]
  Formula : [as stated in the skill's driver table]
  Inputs  : [ledger row numbers used]
  Math    : [every step written out, e.g.
            5 min x 4,200/mo x 12 = 4,200 h/yr
            4,200 h x $32.50/h = $136,500 gross
            x 0.65 realization = $88,725]
  Annual value: [$]
```

Then the totals:

```
Total annual HARD value : [$]   (headline basis)
Total annual SOFT value : [$]   (separate section, never in headline)
Total Year-1 investment : [$]   = licence [$] + implementation [$]
                                + integration [$] + internal effort
                                ([h] x [$/h]) + training [$] + admin [$]
Annual recurring cost   : [$]
Annual net value        : [$]   = hard value - recurring cost
ROI ([horizon], net-benefit convention)
                        : [(total benefits - total costs) / total costs]%
Payback                 : [12 x Year-1 investment / annual hard value]
                          months - or "does not pay back within the
                          horizon" if net value <= 0
Multi-year ([N]y)       : per-year value/cost table + cumulative net
                          [+ NPV at buyer's own rate r = [%], if finance
                          reviews]
```

## 3. Scenario table (conservative first)

| Case         | Changed inputs (and why the range is defensible)   | Annual hard value | ROI | Payback |
| ------------ | -------------------------------------------------- | ----------------- | --- | ------- |
| Conservative | [haircut rep-assumed and benchmark inputs hardest] | [$]               | [%] | [mo]    |
| Expected     | [inputs at face value, haircuts applied]           | [$]               | [%] | [mo]    |
| Optimistic   | [upper bounds, stated]                             | [$]               | [%] | [mo]    |

Sensitivity note (mid-market and up): "The result is most sensitive to [input 1] and [input 2]; swinging [input 1] between [low] and [high] moves annual value between [$] and [$]."

## 4. Disclosure line (when required)

> Headline figures rest in part on rep-assumed inputs ([list rows]); they have not been confirmed by [buyer]. Replacing them with confirmed values is the next step of this analysis.

## 5. One-page business case (five parts)

```
[No vendor logo. Plain formatting or the buyer's own template.]

1. HEADLINE - tied to [buyer's named priority/initiative]:
   "[Outcome] in support of [initiative]"

2. PROBLEM - [affected team/parties] currently [problem], costing
   [buyer's own quantified figure] per [period]. [Urgency driver:
   what makes this a now-problem]. Cost of doing nothing over
   [horizon]: [$, in the buyer's own units].

3. RECOMMENDED APPROACH - the shift: from [current practice] to
   [new practice]. (Not a feature list.)

4. TARGET OUTCOMES - before/after on executive KPIs:
   [KPI 1]: [current] -> [target]     [KPI 2]: [current] -> [target]
   Conservative-case annual value: [$] (full math attached).

5. REQUIRED INVESTMENT - from us: [$, resources, timeline].
   From [buyer]: [people, hours, decisions, dates].
   Payback: [conservative-case months]. Time to value: [date].
```

Aim the whole page at the economic buyer's three concerns: costs, time to value, and confidence in the initiative - and the cost of inaction.

## 6. Finance-reviewer appendix (enterprise, or whenever finance reviews)

- [ ] Assumption appendix: every ledger row with source and provenance label.
- [ ] ROI convention and time horizon stated.
- [ ] Full TCO shown, including internal effort.
- [ ] No double counting (attest which resources back which driver).
- [ ] Soft value in its own labelled section, excluded from headline.
- [ ] Three scenarios plus sensitivity on the top inputs.
- [ ] NPV/discount rate: the buyer's own hurdle rate, asked for - not assumed.
- [ ] Written assumption sign-off requested from champion/finance, with date.

# Worked examples

Illustrative cases written for this skill. All figures are example values, not benchmarks - never reuse them as defaults for a real deal.

## Example 1 - B2B mid-market: invoice-automation purchase

Deal frame: 8-person accounts-payable team, mid-market, CFO will review. Artifact: one-page value summary plus light model.

Driver choice: labour time saved plus error/rework reduction. Error/rework normally ranks fifth on efficiency because its cost per error takes a week or more to source. Here the buyer's Q2 ops report already carries both the baseline rate and the cost per error, which is the promotion condition, so it moves up beside labour time.

### Input ledger

| #   | Input                       | Value   | Unit        | Source                                                | Provenance                   |
| --- | --------------------------- | ------- | ----------- | ----------------------------------------------------- | ---------------------------- |
| 1   | Invoices processed          | 4,200   | /month      | Buyer's ops dashboard (champion shared)               | buyer-supplied               |
| 2   | Manual touch time           | 6       | min/invoice | Champion's time study                                 | buyer-supplied               |
| 3   | Error rate requiring rework | 1.8%    | of invoices | Buyer's Q2 ops report                                 | buyer-supplied               |
| 4   | Cost per rework             | $41     | /error      | Buyer's Q2 ops report                                 | buyer-supplied               |
| 5   | AP clerk base salary        | $52,000 | /year       | Champion (HR-confirmed)                               | buyer-supplied               |
| 6   | Fully loaded multiplier     | 1.3x    | ratio       | Standard finance range 1.25-1.35, source named in doc | benchmark                    |
| 7   | Post-automation touch time  | 1       | min/invoice | Rep estimate from other deployments, unconfirmed here | rep-assumed                  |
| 8   | Post-automation error rate  | 0.6%    | of invoices | Rep estimate, unconfirmed                             | rep-assumed                  |
| 9   | Subscription                | $30,000 | /year       | Written quote #Q-1142                                 | quoted (rep-owned cost side) |
| 10  | Implementation (one-time)   | $8,000  | one-time    | Written quote #Q-1142                                 | quoted (rep-owned cost side) |
| 11  | Internal effort             | 120     | hours       | Champion's estimate                                   | buyer-supplied               |

Loaded hourly cost: $52,000 x 1.3 = $67,600/year; / 2,080 h = $32.50/h.

### Arithmetic (expected case)

```
Driver 1: Labour time saved                                  hard
  (6 - 1) min x 4,200/mo x 12 = 252,000 min = 4,200 h/yr
  4,200 h x $32.50 = $136,500 gross
  x 0.65 realization haircut  = $88,725/yr

Driver 2: Error/rework reduction                             hard
  (1.8% - 0.6%) x 50,400 invoices/yr x $41 = $24,797/yr

Soft (separate): capacity freed absorbs invoice growth without a
  9th hire; potential avoidance $67,600/yr - NOT in headline.

Total annual hard value : $113,522
Year-1 investment (TCO) : $30,000 + $8,000 + (120 h x $32.50 = $3,900)
                        = $41,900
Annual net value        : $113,522 - $30,000 = $83,522
ROI (Year 1, net-benefit): (113,522 - 41,900) / 41,900 = 171%
Payback                 : 12 x 41,900 / 113,522 = 4.4 months
```

### Scenarios (conservative first)

| Case         | Changed inputs                                  | Annual hard value             | ROI (Y1) | Payback |
| ------------ | ----------------------------------------------- | ----------------------------- | -------- | ------- |
| Conservative | Time saved 3 min not 5; error rate only to 1.2% | $53,235 + $12,398 = $65,633   | 57%      | 7.7 mo  |
| Expected     | As above                                        | $113,522                      | 171%     | 4.4 mo  |
| Optimistic   | Full 5 min, no haircut; errors to 0.4%          | $136,500 + $28,930 = $165,430 | 295%     | 3.0 mo  |

Sensitivity: the result hinges on inputs 7 and 8 (both rep-assumed). Disclosure line required:

> Headline figures rest in part on rep-assumed inputs (post-automation touch time, post-automation error rate); they have not been confirmed by the buyer. The pilot's first measurement replaces them.

### One-page narrative (condensed)

```
1. HEADLINE: Cutting invoice-cycle cost in support of "Project
   Streamline" (FY close-acceleration initiative).
2. PROBLEM: The 8-person AP team spends 4,200 hours/year on manual
   matching, and 1.8% of 50,400 invoices need $41 rework - a
   ~$137K/yr gross cost by the team's own time study and Q2 ops
   report. Doing nothing for 12 more months costs that again, plus
   a 9th hire as volume grows.
3. APPROACH: Shift from manual three-way matching to
   exception-only review.
4. OUTCOMES: touch time 6 -> 1 min; rework 1.8% -> 0.6%;
   conservative-case value $65,633/yr (math attached).
5. INVESTMENT: $41,900 Year 1 ($30K/yr + $8K setup + 120 internal
   hours). Buyer side: process owner 2 h/wk for 6 weeks, data
   access by [date]. Conservative payback: 7.7 months.
```

Next steps delivered with it: two edit rounds with the champion, written sign-off on inputs 7-8 (or a pilot to measure them), then the forward ask.

## Example 2 - high-ticket B2C: residential solar

Framework mapping to consumer sales is a structural analogy rather than established practice; the regulatory constraints are real law and stricter than B2B norms. State both in the output.

- **Economic buyer** (by analogy): the couple jointly, plus the lender whose financing approval gates the purchase.
- **Champion**: the spouse who requested the quote; the document must convince the skeptical spouse without the salesperson present - same forward test as B2B.
- Ledger discipline unchanged:
  - Last-12-months utility bills $2,760/yr (buyer-supplied - their own bills)
  - Financing payment $148/mo (lender terms - documented)
  - Production offset 85% (installer model - label it vendor-data/rep-assumed until substantiated)
- Conservative case leads: offset 70%, zero utility-price inflation -> monthly savings $161 vs. payment $148 - the case barely clears, and the document says so. Optimistic case (85% offset, 3% utility inflation) is shown, never headlined.
- Regulatory constraints (US):
  - Savings claims must be substantiated (FTC Act Section 5)
  - Financing terms fully disclosed under lending-disclosure law
  - Door-to-door sale carries a 3-business-day cooling-off right
  - State solar-disclosure statutes may add cancellation rights and mandatory disclosure documents

  Present a payback range with disclosed assumptions - never a single "you'll save $X" headline.

- Identical to B2B, explicitly:
  - The ledger
  - The provenance labels
  - No invented numbers
  - Hard/soft separation (comfort and home-value hopes are soft; bill offset is hard)
  - Three cases conservative-first
  - The quality gate

## Example 3 - negative example (what failing the gate looks like)

> "This saves your team 20 hours a week (industry data), so ROI is 400% and it pays for itself in 3 months."

Gate failures:

- No ledger (fails check 1)
- "industry data" unsourced, so the 20 hours is rep-assumed but unlabelled (fails 2)
- Headline rests entirely on it with no disclosure (fails 3)
- No TCO - subscription price only (fails 7)
- Single point estimate (fails 8)
- No narrative, no cost of inaction (fails 9)
- A conveniently clean payback with invisible math (fails 6)

This is exactly the rep-built case buyers discount on sight.

---
name: deal-value-calc
description: Builds the ROI and business case for one deal - value drivers, full investment, ROI, payback, three cases conservative-first, and a one-page case the champion can forward. The buyer supplies every assumption, each number labeled buyer-supplied, benchmark, or rep-assumed, and none invented. Covers B2B and high-ticket B2C. Use whenever the user mentions ROI, business case, payback period, TCO, cost of doing nothing, "justify the price", or CFO sign-off, even without the word value. Do NOT use for price rebuttals (mbfinotti/sales-skills@sales-objection-handling) or discount trade planning (mbfinotti/sales-skills@negotiation-concession-planner).
license: MIT
metadata:
  author: Maya-Beth Finotti
  version: "1.3.3"
---

# Deal Value Calculator

Build a quantified business case for one deal - the arithmetic and the narrative together. The organising rule: "It's your job to build the math equation. It's their job to give you the assumptions." (Armand Farrokh, 30MPC). This skill owns the structure, formulas, and benchmark scaffolding; the buyer owns every company-specific input.

The documented failure it exists to prevent: "most business cases created by sellers are absolute BS" (30MPC) - inflated numbers the champion never validated, which burn the champion's internal credibility the moment finance pokes a hole.

Second rule: "An ROI model on its own isn't a business case. It's a future maybe, without context. Narratives bring meaning to your numbers." (30MPC). Always deliver both the visible math and the narrative document - never a bare ROI number. The real test of the output: it survives being forwarded to a CFO without the rep in the room.

## Never invent a number

- Refuse to fabricate any input. When the user lacks a figure, ask for it, or insert a clearly labelled placeholder they must replace (e.g. `[NEEDED: fully loaded hourly cost - ask the buyer]`) - never a plausible-looking invented value.
- Label every input by who supplied it:
  - **buyer-supplied** - a number the buyer stated or documented (their payroll, invoices, ops reports, public filings).
  - **benchmark** - an external or market figure the rep brought, with its source named. "Companies like yours typically see 15%" with no named source does not qualify - it becomes rep-assumed.
  - **rep-assumed** - the rep's own estimate, not confirmed by the buyer.
- Hard rule: a business case whose headline number rests materially on rep-assumed inputs must say so on its face - a visible line in the document, not a footnote.
- The cost side is the one exception: the rep's own pricing and fees are rep-owned by definition - document them with the written quote. Every value-side input takes one of the three labels.
- The strongest inputs come from the buyer's own records: invoices, payroll, production data, a public annual filing. Get the champion to state the numbers; finance checks the baseline before it opens the vendor proposal practice).

## Interview

Ask one question per message; offer multiple-choice options wherever possible; skip anything the deal description already answers.

1. B2B, or considered high-ticket B2C (solar, home improvement, real estate, private education, financial advisory)?
2. Deal segment: transactional (single decider, low price), mid-market, or enterprise (buying committee, procurement/security in the path)? This sets the artifact size.
3. What date must the result land by? A hard near date promotes the drivers computable from documents the buyer already holds - retired contracts, the hiring plan - and deletes any driver needing a time study or a cohort analysis.
4. One-off win, or a compounding asset the buyer reuses as a renewal or QBR baseline?
   - Compounding promotes error/rework and downtime avoided, whose measured baselines outlive the deal.
   - One-off keeps the artifact at its lightest rung.
5. Effort ceiling: how many hours can the champion spend, and can they reach finance data at all? A low ceiling deletes error/rework, churn and revenue lift outright rather than ranking them last, and caps the artifact at a payback estimate.
6. Which value drivers are in play? Offer the taxonomy list below in its efficiency order as choices, re-ranked against the answers to 3-5.
7. For each driver: the raw inputs (volumes, rates, hours, costs) - and for each input, who supplied it: buyer, a named benchmark, or your own estimate?
8. The investment side:
   - Subscription/licence price
   - Implementation and integration cost
   - Internal effort (hours and whose)
   - Training
   - Ongoing administration

   Contract length?

9. Time horizon for the case: 1 year, 3 years (the enterprise-software standard), or the contract term?
10. Who is the document ultimately for - who will the champion forward it to (economic buyer, CFO, a committee, a spouse or co-signer in B2C)?
11. Will the buyer's finance function (or lender, in B2C) review it? If yes, the finance-reviewer additions in the workflow apply.

## Right-size the artifact

Not every deal gets a full model deal-type guidance; the trigger for the full treatment is complexity - roughly 6+ stakeholders, procurement or security in the path, strategic rather than tactical spend - not a fixed dollar threshold):

- efficiency: payback estimate > one-page summary + light model > full model
- effort: full model > one-page summary + light model > payback estimate
- value: payback estimate == one-page summary == full model below the complexity trigger, then full model > one-page summary > payback estimate above it

Default to the lightest artifact the deal's complexity permits; each complexity trigger crossed moves the artifact up exactly one rung. The value axis ties below the trigger because extra depth genuinely buys nothing there - a formal case is overkill for a single decider - and it inverts above the trigger, where a payback estimate simply does not survive procurement.

| Segment             | Artifact                                                                                                                                                            | Depth                                                                   |
| ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------- |
| Transactional / SMB | A short payback estimate: one or two drivers, the investment, payback in months                                                                                     | No formal document; the payback line is the whole artifact              |
| Mid-market          | One-page value summary plus a light model: input ledger, 3-4 drivers, three scenarios                                                                               | Rep-built; sensitivity on the one or two inputs the result hinges on    |
| Enterprise          | Full model: economic-buyer summary and a separate finance/procurement version with an assumption appendix, documented sources for every input, sensitivity analysis | Engage a value engineer or business-value consultant if the org has one |

## Value drivers and per-driver formulas

Decompose the deal's value into drivers and compute each separately; 3-4 quantified drivers is the documented practitioner shape for the problem/solution section (30MPC). Rank them by case strength bought per hour of the buyer's time - never by headline size, and never by whichever number is easiest to produce. Effort here is not what the driver is worth: it is how hard the input is to source from the buyer, how much of the buyer's time that costs, and how long the resulting number stays arguable in a finance review.

- efficiency: direct cost > labour time > headcount avoidance > downtime avoided > error/rework > headcount reduction > capacity freed > revenue lift > churn > risk/compliance
- value (moves the headline _and_ survives finance): labour time > direct cost > error/rework > headcount reduction > downtime avoided > revenue lift == churn > headcount avoidance > risk/compliance > capacity freed
- effort (buyer hours to source the input, plus how long the number stays contestable): headcount reduction > risk/compliance == churn > error/rework > revenue lift > labour time == downtime avoided > capacity freed > headcount avoidance == direct cost
- compliance cost: risk/compliance == headcount reduction > every other driver, which carries none

Every tie is a real equality, never a dodge:

- revenue lift == churn (value) - same margin arithmetic, and both die on the same attribution objection, so finance discounts them identically.
- risk/compliance == churn (effort) - neither number exists in the buyer's own systems; both cost a week of external or cohort analysis.
- labour time == downtime avoided (effort) - each hinges on one figure the buyer either already tracks (an hour) or has to measure (a week). Same fork, same price.
- headcount avoidance == direct cost (effort) - both read straight off a document the buyer already owns, the hiring plan and the contract list.
- risk/compliance == headcount reduction (compliance cost) - each needs a review outside the deal team before the number can be written down: legal or compliance sign-off on a stated regulatory exposure, HR (in some jurisdictions works-council) consultation on a stated role removal. Both are hard to walk back once in writing.

B2C adds a substantiation duty on top of that, on every driver - see B2B and B2C.

Default: lead with the top two _hard_ drivers on the efficiency line that this deal actually has, since the case must clear the buyer's hurdle on hard value alone, then add soft drivers (headcount avoidance, capacity freed, risk/compliance) to close. Move up the value line when finance is the audience and the hard drivers are thin.

**What this order starves.** Error/rework reduction is the driver finance argues with least, and efficiency pushes it to fifth purely because a baseline cost per error is the one number most teams never track. Promote it to first when the buyer already tracks it, or when quality, compliance or an audit finding is the initiative named in the headline - there it is simultaneously the strongest and the cheapest driver in the case. Headcount reduction starves the same way on political rather than data effort; a stated reduction mandate promotes it.

**Delete, do not demote.** A driver the buyer cannot feed leaves the case entirely:

- No baseline cost per error deletes error/rework.
- Withheld gross margin deletes revenue lift and churn, rather than tempting a top-line model.
- No written agreement that a role leaves the budget deletes headcount reduction, rather than parking it as soft.

A ruled-out driver left at the bottom of the list comes back later as a number somebody invents.

**Re-rank against the buyer in front of you.** This order is a default, not a law - it shifts with the deal and with who builds the case, since a value engineer can afford a driver a solo rep cannot.

- A buyer who already instruments a metric moves that driver up several places at once.
- A procurement-led evaluation weights direct cost and TCO comparison hardest and gives soft drivers roughly nothing.
- A champion with no reach into finance data can supply no baseline at all, which deletes half the list before ranking even starts.

Formulas and cautions consolidate standard practitioner and finance practice per the source labels shown.

| Driver                               | Formula                                                                                                                      | Buyer input and effort                                                                                                                                        | Caution                                                                                                                                                                  |
| ------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Direct cost / tool consolidation     | Retired licence, infrastructure, and service spend                                                                           | Contract list and renewal amounts - near-zero: already in the buyer's payables, and finance can verify it in one query                                        | Hard; confirm the old contracts actually terminate                                                                                                                       |
| Labour time saved                    | Hours saved per unit x units/year x fully loaded hourly cost                                                                 | Volume and salary band, an hour. Per-unit touch time: an hour if the team tracks it, a week for a time study. Contested on realization, not on the arithmetic | Fully loaded = base wage plus benefits and taxes, typically 125-135% of base (practitioner range). Apply a realization haircut of 60-70% for adoption and learning curve |
| Headcount avoidance                  | Loaded cost of hires not made as volume grows                                                                                | The approved hiring plan - an hour, and the champion usually has it open already                                                                              | Soft (cost avoidance); more palatable to finance than reduction - no layoff required                                                                                     |
| Downtime avoided                     | Downtime hours avoided x cost per hour of downtime                                                                           | Cost per hour of downtime: an hour if incident or SLA work already produced the figure, a quarter to derive one across functions                              | Cost per hour must be buyer-supplied                                                                                                                                     |
| Error / rework reduction             | (Baseline error rate - new rate) x volume x cost per error                                                                   | Baseline rate and cost per error: a week to a quarter, since most teams track the rate and not the cost. Least contested of any driver once supplied          | Needs a documented baseline cost per error, buyer-supplied                                                                                                               |
| Headcount reduction                  | Fully loaded cost of roles removed from budget                                                                               | The arithmetic is free; the effort is political - a standing job to get written agreement that the budget line actually goes                                  | Hard only if the role genuinely leaves the budget                                                                                                                        |
| Capacity freed (soft)                | Hours freed x loaded cost x utilisation factor                                                                               | A utilisation factor and a named redeployment target - an hour to state, and finance discounts it anyway                                                      | Only cash if the freed time is redeployed to additive work                                                                                                               |
| Revenue / win-rate / conversion lift | Incremental revenue x gross margin                                                                                           | Conversion baseline plus gross margin - a week, and margin is often withheld; the attribution argument then reopens at every review                           | Model on margin, never top line; finance discounts attribution heavily                                                                                                   |
| Churn / retention                    | Retained recurring revenue x margin                                                                                          | Cohort retention by segment - a week of analysis the buyer has to run themselves                                                                              | Attribution must be specific, not "stickiness"                                                                                                                           |
| Risk / compliance avoidance          | Expected loss = probability x impact; annualized loss = single-loss cost x annual rate of occurrence standard risk formulas) | A defensible probability and a named loss benchmark - a week to source, and the probability stays arguable forever                                            | Probabilistic - pair with a named loss benchmark, never an unsourced scare number                                                                                        |

Double-counting guard: one person's hours, one revenue stream, one budget line backs at most one driver. Time saved and headcount avoidance drawn from the same team must be split, not both counted in full.

**Hard vs. soft value - keep them separate.**

- Hard (cash-releasing) value hits the P&L or budget.
- Soft value (cost avoidance, capacity, satisfaction, risk comfort) is real but not bankable, and finance is trained to reject it inside an ROI number.

Present them as two distinct sections: the hard section closes logically, the soft section closes emotionally - deliberately in that order (30MPC). Test: the case must stand on hard value alone; if it only works with soft value included, rebuild it practitioner rule).

## Core metrics and conventions

State every convention in the document; finance rejects math it cannot audit.

- **Total annual gross value** = sum of hard driver values (soft totalled separately, outside the headline).
- **Total investment (TCO)** = licence/subscription + implementation + integration + internal effort (hours x loaded cost) + training + ongoing administration. Licence-only investment understates true cost badly as a documented finance objection).
- **Annual net value** = total annual gross value - annual recurring cost.
- **ROI %** = (total benefits - total costs) / total costs x 100, over the stated horizon. This is the net-benefit convention; a gross convention (benefits / costs) also exists - always say which one the document uses that both conventions circulate).
- **Payback (months)** = 12 x total Year-1 investment / total annual gross value (hard only). If annual gross value minus recurring cost is zero or negative, payback is undefined: write "does not pay back within the horizon" - never hide a division by zero or a negative result.
- **Multi-year view**: when the contract or horizon is multi-year, show each year's value and cost, then the cumulative net. For a finance review, add NPV = sum of (net cash flow_t / (1+r)^t) - initial investment, with r the buyer's own hurdle rate or cost of capital - ask finance for it rather than assuming (a 10% default exists in one commercial methodology).
- Payback expectations are segment-dependent and come from practitioner and vendor content, not peer-reviewed research:
  - Cheap SMB tools: 3-6 months
  - Cloud software: 12-18 months
  - Larger on-premise: 18-30 months
  - Complex data/AI purchases: 12-24+ months

  Never present any threshold as a universal rule - the buyer's own hurdle is a question to ask finance, and a benefit stream that lands conveniently on a clean 12-month payback reads as manufactured finance objection.

## Three cases, not one point estimate

Produce conservative / expected / optimistic scenarios by varying the genuinely uncertain inputs across defensible ranges - a convention repeated independently across financial-modeling practice as a cross-source convention).

- presentation order: conservative > expected > optimistic. Effort is identical across the three - the same model run at three input settings - so one ordering covers it.
- **Lead with the conservative case.** It is the one the champion defends; if it still clears the buyer's hurdle, the investment does not require perfect adoption to be worth it practitioner consensus).
- The conservative case haircuts rep-assumed and benchmark inputs hardest; buyer-supplied baselines stay fixed unless the buyer offered a range.
- For mid-market and up, add one-at-a-time sensitivity: identify the two or three inputs the result is most sensitive to and show the swing as recognized sensitivity practice).
- Do not rank the three by likelihood. Never attach a probability or a weight to a case: the ranges are defensibility bands, not a distribution, and a "70% likely" label is false precision that finance prices as a guess.

## The narrative

The five-part one-page structure (30MPC):

1. **Priority-driven headline** - tie to an internal priority or initiative the buying executives already recognize.
2. **Problem statement** - the affected parties and the quantified cost of the problem, in the buyer's own numbers, with urgency drivers.
3. **Recommended approach** - the shift in practice enabled, not a feature list.
4. **Target outcomes** - before/after with measurable metrics tied to executive-level KPIs.
5. **Required investment** - commitments of money, people, and time from both sides, with a timeline.

Rules that make it credible:

- Point the whole narrative at the economic buyer's three concerns: costs, time to value, and confidence in the initiative (MEDDICC) - and the cost of doing nothing, in the buyer's own units.
- Strip vendor branding and sales polish; plain formatting or the buyer's own template, so it reads as the buyer's internal analysis, not sales collateral).
- "You're not writing to your champion. You're writing through them (to who they'll forward it to)." (30MPC). Buying groups spend only a small fraction of the purchase with any one supplier; the document does most of its selling in rooms the rep never enters.
- Timing: start the case right after the first discovery call and iterate with the buying team - never produce it late-stage as a leave-behind (30MPC). Invite buyer edits; every correction converts a rep-assumed input into a buyer-supplied one. (The claim that win rates triple after three rounds of buyer edits is unreplicated - count edit rounds, do not promise the lift.)
- Get the champion, and ideally finance, to agree the assumptions in writing before the case is presented, so the number is the buyer's, not the vendor's practice).

## Workflow

1. Run the interview; skip answered questions.
2. Right-size the artifact: start at the lightest rung and move up one per complexity trigger crossed.
3. Select drivers off the efficiency order, re-ranked against the interview's date, compounding and effort-ceiling answers; delete the ones the buyer cannot feed rather than demoting them.
4. Build the input ledger: every input with its value, unit, source, and provenance label (buyer-supplied / benchmark with named source / rep-assumed). Missing input: ask, or insert a labelled placeholder - never invent.
5. Compute each driver's annual gross value; keep hard and soft in separate totals; run the double-counting guard.
6. Compute total investment (full TCO, not licence-only).
7. Compute net value, ROI (state the convention), payback (guard the undefined case), and the multi-year view when applicable.
8. Build the three scenarios; check the conservative case against the buyer's hurdle; add sensitivity for mid-market and up.
9. Test the case on hard value alone; if it fails, tell the user before writing any narrative.
10. Write the narrative with the five-part structure using [references/business-case-template.md](references/business-case-template.md); strip vendor branding; aim it at costs, time to value, and confidence.
11. Run the forward test and the Quality gate; iterate until the gate passes.
12. Deliver, with next steps: champion edit rounds, written assumption sign-off from finance, and the ask - "if I write a short summary of what we've built, are you willing to forward it?" (30MPC phrasing).
13. If your harness has persistent memory, store the input ledger, provenance labels, and scenario assumptions so later runs update instead of rebuilding; otherwise end with a recap the user can paste into their deal record.

If you can browse the web, verify any benchmark the user cites (source and figure) before labelling it **benchmark**; otherwise keep it rep-assumed until a source is named.

## Invocation and expected output

- "Build the business case for my deal: invoice-automation purchase, 8-person AP team, buyer gave me their volumes and error rate."
- "What's the ROI on this deal? $30K/year subscription, they say it saves each rep 4 hours a week."
- "The CFO wants to see the math before Thursday - help me make this defensible."

Deliver two blocks (fill-in templates in [references/business-case-template.md](references/business-case-template.md)):

```
THE MATH     : input ledger (value, unit, source, provenance label),
               per-driver arithmetic shown in full, hard vs soft totals,
               TCO, ROI + convention, payback, three-scenario table
               (conservative first), sensitivity notes,
               rep-assumed disclosure line if applicable
THE NARRATIVE: the five-part one-page business case, unbranded,
               aimed at costs / time to value / confidence,
               plus next steps (edit rounds, assumption sign-off)
```

## B2B and B2C

- Identical in both, explicitly:
  - The arithmetic and formulas
  - The input ledger and provenance discipline
  - The never-invent-a-number rule
  - Hard/soft separation
  - Three scenarios led by the conservative case
  - The forward test
  - The Quality gate
- The practitioner frameworks above are B2B-native; no reachable practitioner source applies them to consumer sales as an absence. Every framework mapping below is therefore structural analogy - label it as such in any B2C output.
- B2C mapping (structural analogy):
  - Economic buyer -> whoever controls the household money or whose financing approval gates the purchase
  - Champion -> the enthusiastic household member selling the skeptical one
  - Drivers -> utility-bill savings, avoided repairs and maintenance, resale value, financing cost vs. current spend
  - The artifact -> usually a one-pager comparing monthly payment against current monthly cost, with a payback range
- B2C regulatory constraint - this is where B2C is genuinely stricter, not just analogous: consumer savings claims must be truthful and substantiated under consumer-protection law. In the US:
  - FTC Act Section 5
  - The R-Value Rule for insulation and window energy claims - with multiple enforcement actions against exaggerated percentage-savings claims
  - Lending-disclosure law for financing
  - Cooling-off rights for door-to-door sales
  - State solar-disclosure statutes

  An unsubstantiated "save 40-70%" claim is a legal exposure, not just weak selling. Present ranges with disclosed assumptions, never headline averages; disclose total cost and financing terms.

- In B2C the conservative case matters even more: the buyer's "finance function" is the household budget and often a lender, and a case that only works in the optimistic scenario is exactly what regulators act against.

## Quality gate

Score the drafted case against all ten checks. Pass threshold: 10/10. Iterate - fix and re-check - until nothing fails.

1. Every value-side input appears in the ledger with value, unit, source, and one of the three provenance labels; cost-side inputs cite the written quote; no unlabelled number anywhere in the math.
2. No input was invented: every non-buyer figure is either a named-source benchmark or explicitly rep-assumed / a labelled placeholder.
3. If the headline changes materially when rep-assumed inputs are stripped, the document says so on its face.
4. Hard and soft value are totalled separately, the headline uses hard only, and the hard-only case was tested.
5. No headcount, hours, revenue stream, or budget line is counted under two drivers.
6. Every output figure is traceable: the arithmetic from inputs to headline is shown, sums check, and no undefined payback or division by zero is hidden.
7. ROI convention, time horizon, and full TCO (including internal effort) are stated.
8. Three scenarios present, conservative first; the conservative case clears the buyer's hurdle or the document says plainly that it does not.
9. The narrative has all five parts and contains no feature list. It is aimed at costs / time to value / confidence and the cost of inaction. It passes the forward test: unbranded, self-explanatory to a reader who never met the rep, nothing in it the buyer has not validated or been explicitly asked to validate.
10. Every carried claim keeps its provenance label; no vendor-data figure is presented as an expected result; no payback threshold is stated as universal.

## KPIs and measurement

- **The buyer corrected an input** - proof they engaged with the math, and each correction upgrades an input's provenance.
- **Edit rounds completed** - the case was co-built, not delivered; track the count itself, not a promised win-rate lift (the tripling claim is unreplicated).
- **Written assumption sign-off from finance or the champion** before presentation is recommended practice; using it as the KPI is a structural choice.
- **The champion forwarded the document unedited**, and it reached an economic-buyer conversation.
- Across deals: compare no-decision losses on deals with a co-built case vs. without. Baseline for how much room there is: 40-60% of qualified B2B deals end in no decision (Dixon and McKenna, HBR 2022 / The JOLT Effect).

## Common failure modes

| Failure                                                       | Fix                                                                                               |
| ------------------------------------------------------------- | ------------------------------------------------------------------------------------------------- |
| Inflated vendor math the champion repeats and gets burned on  | Buyer's numbers, conservative case first; the champion's credibility is the asset being protected |
| Invented or "plausible" inputs                                | Refuse; ask or use a labelled placeholder the user must replace                                   |
| Bare ROI number, no narrative                                 | "A future maybe, without context" - always deliver both blocks                                    |
| Soft savings blended into the headline ROI                    | Separate sections; test the case on hard value alone                                              |
| Single point estimate                                         | Three scenarios, conservative leading                                                             |
| Licence-only investment                                       | Full TCO: implementation, integration, internal effort, training, administration                  |
| Same headcount counted under two drivers                      | One resource backs one driver; split or drop                                                      |
| Round numbers and benefits landing exactly on a clean payback | Keep the raw arithmetic visible; convenient numbers read as manufactured                          |
| Business case produced late-stage as a leave-behind           | Start after the first discovery call; iterate with the buying team                                |
| Written to the champion instead of through them               | Run the forward test before delivery                                                              |
| Unsourced "companies like you see X%" benchmark               | Name the source or relabel it rep-assumed                                                         |
| B2C savings claim without substantiation                      | Ranges with disclosed assumptions; total cost and financing terms disclosed                       |

## Reference

- [references/business-case-template.md](references/business-case-template.md) - fill-in input ledger, arithmetic block, scenario table, five-part one-page narrative, and the finance-reviewer appendix checklist.
- [references/worked-examples.md](references/worked-examples.md) - a full worked B2B mid-market case, a high-ticket B2C case (framework mapping labeled as structural analogy), and a short negative example.
- mbfinotti/sales-skills@sales-discovery-questions - elicits the raw impact figures; they arrive here as inputs.
- mbfinotti/sales-skills@meddpicc-scorecard - grades whether evidence exists; this skill builds the number its Metrics element checks, and never re-qualifies the deal.
- mbfinotti/sales-skills@deal-champion-mapping - identifies the champion; this skill writes through them.