Purchase Justification

Purchase Justification Skill

Purchase requests die of vagueness, not price: "we need this tool, it's $8k" gives the approver nothing to weigh, so deferral — the polite no — wins by default. The justification that works is the proposal-skeleton specialized for spend: the problem priced first (what not-buying costs, in the approver's currency — hours, risk, revenue), the ROI math honest and conservative (one inflated payback poisons the requester's every future ask), the alternatives-considered section dispatching the obvious pushbacks ("can't we use what we have?") before they're raised, and the ask sized to its approval tier — because a $4,800 pilot approved this week beats a $50k platform deferred to next fiscal year.

What This Skill Produces

• The justification memo — one page: the priced problem, the proposal, the ROI, alternatives, the ask
• The ROI math — conservative, assumption-labeled, with the payback period the approver can check
• The alternatives table — do-nothing (priced), the cheaper option (honestly assessed), the requested option
• The ask sizing — the request shaped to the approval tier and the pilot-first path where it helps

Required Inputs

Ask for these if not provided:

• The problem's evidence — the hours lost, the incidents, the workaround's cost (meeting-cost-meter-style arithmetic on the status quo); requests without a priced problem are wishes with quotes attached
• The approver and their currency — who signs at this amount, what they weigh (cost-saving? risk? team velocity?), and what's burned them before (exec-vs-working-deck audience-currency logic)
• The real numbers — the quote (vendor-comparison-matrix TCO, not license price), and the honest benefit estimate with its assumptions
• The approval tiers — where the thresholds sit ($5k? $25k?); the sizing strategy needs the map

Framework: The Justification Rules

1. Price the status quo first: "the manual process costs ~9 hours/week across the team (≈$2.3k/month)" — the not-buying number converts the purchase from spending into trading, and makes deferral a priced choice instead of a free one. Conservative estimates only; the approver who checks your math once trusts it forever after.
2. ROI at the approver's altitude, assumptions visible: payback = TCO ÷ monthly benefit, with each assumption labeled ("assumes 6 of the 9 hours recovered — the conservative case") — and the token-cost-style honesty: ranges beat points, and the worst-case-still-works framing ("even at half the benefit, payback is 11 months") is the strongest sentence available.
3. Dispatch the alternatives preemptively: the table — do-nothing (priced, per rule 1), the cheaper/existing-tool option (steelmanned: what it genuinely covers, where it fails the requirement), the requested option — because "can't we just use spreadsheets?" asked in the approval meeting means the memo failed to answer it in advance (spreadsheet-or-database often is the honest answer, and knowing when strengthens every request that isn't).
4. Size the ask to the tier and the trust: below-threshold pilots move in days; cross-threshold platforms move in quarters — the pilot-first path ("3 seats, 3 months, $1.4k, success = [metric], then the real decision") converts the big ask into a small ask plus evidence. Never split purchases to dodge thresholds (that's a career-limiting optimization); do scope genuinely smaller starts.
5. Close with the decision made easy: the specific ask (amount, vendor, start date), the reversibility line ("monthly contract, cancellable"), and the offer of the deeper material (citation-hygiene-clean quotes and comparisons in the appendix). Approvers approve what's easy to approve; the memo's whole job is making yes the low-effort path.