Instrumentation ROI model — where automation earns its keep first¶

Purpose¶

Translate ranking (Instrumentation priority matrix) into decision-grade framing: what must be measured before scaling, how cost and benefit sit in the books, and how instrumentation interacts with labor (triage, false positives, upkeep) and security. Instrumentation does not automatically reduce labor—it may reduce trips or document risk; net hours can be negative if triage dominates.

Non-goals: Vendor ROI calculators; claiming payback without filled inputs.

Controls: Smart technology and telemetry strategy, Automation stop rules.

Relationship to the priority matrix¶

Tool	Role
Priority matrix	Qualitative score + phase fit + manual-first gates
This ROI model	Quantitative shell: trips, hours, $/yr placeholders—pilot vs fleet

Use both: a row can score high on priority but fail ROI until evidence is logged.

Required inputs¶

Input	Description
Instrument ID	Matches matrix row (`I1`, `I2`, …)
Baseline	Manual operating procedure and measured trips/hours (pre-pilot)
Pilot scope	Count of assets covered; duration (weeks)
Capex	Hardware + install + your labor hours valued at shadow rate
Annual OPEX	Subscriptions, SIMs, spares, power increment
Triage / false-positive / upkeep	Structured fields below—not optional footnotes
Risk value (optional)	Avoided loss events—range, not single hero number
Split-site context	Extra value if instrument reduces `COMMUTE_ONE_WAY` trips (`trip policy`)

Triage time fields (labor — real)¶

Symbol	Meaning	How to measure
`H_TRIAGE_WK`	Hours/week on this instrument cluster: acknowledging alerts, dashboard checks caused by this cluster, correlation with other signals	Time log 4+ weeks
`H_TRIAGE_MO`	`H_TRIAGE_WK × 52 / 12` (or direct monthly sum)	Worksheet
`N_ALERT_WK`	Raw alerts/week from this cluster (before suppression)	Broker / tooling stats
`N_TRUE_WK`	Alerts week that required action or were true positives	Post-hoc label during pilot
`FP_RATE`	False-positive rate placeholder = `(N_ALERT_WK − N_TRUE_WK) / max(N_ALERT_WK, ε)`	Dimensionless; trend matters

Ceiling: Family H_TRIAGE_MAX total across all instruments per Enterprise unit economics—overflow → de-scope before fleet.

False-positive cost fields¶

Symbol	Meaning
`C_FP`	Cost per unnecessary triage event (shadow $/hr × minutes + morale tax optional scalar) — placeholder
`N_FP_MO`	Count/month of triage events classified as false positive after review
`FP_COST_YR`	`N_FP_MO × 12 × C_FP` or `H_FP_MO × 12 × R_SHADOW` where `H_FP` = hours on false alarms

Rule: FP_COST_YR enters net benefit as subtract—same status as OPEX.

Upkeep burden fields (non-triage)¶

Symbol	Meaning
`H_PATCH_MO`	Patch / upgrade / container pulls / dependency fires attributable to this cluster
`H_FIELD_MO`	Physical site visits for sensor maintenance only (not normal farm work)
`H_ADMIN_MO`	Vendor portals, SIM, billing, certificate renewal
`SPARES_YR`	Expected replacement parts cost ($) /yr (antennas, radios, sensors)
`UPKEEP_YR`	`(H_PATCH_MO + H_FIELD_MO + H_ADMIN_MO) × 12 × R_SHADOW + SPARES_YR`

Decommission criteria (kill / freeze)¶

An instrument cluster must de-scope, freeze fleet expansion, or remove when any:

ID	Criterion (pass/fail)
DC-1	`FP_RATE` > `FP_RATE_MAX` for `T_FP_EVAL` consecutive weeks after tuning attempt
DC-2	`H_TRIAGE_WK` > H_TRIAGE_BUDGET_WK` for this cluster where `H_TRIAGE_BUDGET_WK` is allocated share of `H_TRIAGE_MAX`
DC-3	Net annual benefit ≤ 0 for two review periods after pilot — see worksheet
DC-4	Security incident or unpatchable CVE on cluster with no owner within `T_PATCH_MAX`
DC-5	Enterprise path changes such that instrument no longer applies (e.g. no grazing — stock water ROI N/A)

Action: Document kill in wiki/log.md refactor or policy entry; update priority matrix row status deprecated.

Outputs supported¶

Output	Decision
$/yr avoided trip cost	`trips_saved × (mile + time cost)` — compare to OPEX + FP_COST_YR + UPKEEP_YR
Net hours to farm	`(field_hours_saved + travel_saved) − H_TRIAGE − H_FP − H_PATCH − H_FIELD − H_ADMIN` — can be negative
Payback range	`capex / (annual net benefit)` — only when inputs filled
Fleet scale ceiling	When subscription + triage + upkeep dominates—`Automation stop rules`
Book placement	`Farm accounting baseline`

Assumptions¶

Assumption	Rationale
Pilot before fleet	One water level + one pump signal beats ten half-debugged channels.
Triage is real labor	Alerts without resolution SOP increase labor.
Remote site = asymmetric trip value	Trip reduction can justify instrument where distance binds—not automatic.
Security and patching	Charge hours to automation burden.
No double-count	If milestone model already credits family hours saved, do not also book as cash unless hired replacement.

Placeholders (do not fabricate)¶

Symbol	Meaning
`T0`	Baseline period length
`N_TRIP0`	Baseline unnecessary trips to `SITE_FARM` per month
`N_TRIP1`	Post-instrument trips (pilot)
`C_MI`	Fully loaded cost per mile
`H_TRIP`	Hours per round trip
`R_SHADOW`	Shadow $/hr for family time
`CAPEX_I`	Total capital for pilot install
`OPEX_I_YR`	Annual recurring for this instrument cluster
`FP_RATE_MAX`, `T_FP_EVAL`, `H_TRIAGE_BUDGET_WK`, `T_PATCH_MAX`	Policy ceilings — fill

ROI worksheet — per instrument (template)¶

Field	Instrument I___	Notes
Baseline trips / mo		manual log
Pilot trips / mo		after same season conditions
Trips saved / mo	`= max(0, baseline − pilot)`
$/trip saved	`miles × C_MI + H_TRIP × R_SHADOW` (optional components)
Annual trip $ benefit	`trips saved × 12 × $/trip`	placeholder
Annual OPEX		subs + LTE slice + spares
`H_TRIAGE_MO`, `FP_RATE`, `FP_COST_YR`		triage + false positive
`UPKEEP_YR`		patch + field + admin + spares
Net annual benefit (labor-inclusive)	trip $ − OPEX − FP_COST_YR − upkeep $ equivalent − triage $	triage at R_SHADOW
Simple payback (years)	`CAPEX_I / net annual`	if net > 0

Leverage vs cost — explicit balance sheet¶

Automation effect	Cost side	Leverage side
Remote visibility	CAPEX, OPEX, triage, FP cost, upkeep	May reduce trips; may speed fault detection—prove
Alerts	Triage labor; alert fatigue	Risk reduction if SOP exists
Actuators / control	Liability, maintenance; complexity	Labor change only with training + proof
Data platform (farmOS, etc.)	Hosting, integration, backup	Single source of truth for units and inventory

Pilot vs fleet economics¶

Stage	What to document	Gate to advance
Pilot	Before/after trip log; incident count; `FP_RATE`, `H_TRIAGE_WK`	Net hours ≥ 0 or documented kill (DC-* )
Fleet	Marginal OPEX + marginal triage per added node	Sublinear curve or stop (`stop rules`)

Enterprise mix uncertainty¶

When primary enterprise is not fixed, instrument I may only apply to subset of revenue (e.g. stock water only if grazing path active). Add column “Applies if scenario S___” to the ROI table.