Year one Proposed $196,800 vs status quo $142,500 = $54,300 more cash out. The insurance loading is already inside the $142,500, so its removal is captured in that difference — it is not an additional offset. The one genuine offset in year one is capital: the $46,000 server replacement that no longer happens. Net before counting any outage reduction: $8,300 behind. Year one, conservative case Assume outages reduced by 50% only: +$38,000 → net benefit ≈ $29,700. Payback inside year one. Year one, aggressive case Assume outages reduced by 85%: +$64,600 → net benefit ≈ $56,300. Year two onward $162,800 vs $142,500 = $20,300 more cash out, with the insurance removal already inside that figure. Offset by the conservative outage reduction of $38,000 → net ≈ $17,700/yr conservative, $44,300 aggressive. Not counted in any of the above Dave's redeployment to the stalled document management project. We deliberately left it out of the numbers and mentioned it in one sentence — an unquantified benefit stated modestly is credible; the same benefit with a number attached invites an argument you can't win. Assumptions, stated on the page Charge‑out rate $140/hr (theirs). Outage average from their own incident recollection, not our modelling. Seat count 92 with ±5 tolerance before repricing. Insurance loading removal confirmed by their broker in writing, not assumed.