We connected each company's footprint to its financials, sized the exposures in dollars, ranked them, and costed the response. This is what it says.
The ten companies carry $13.5M a year of expected loss: $7.9M on earnings they already have and $5.6M on revenue not yet won. Three of them (Sterling, Harvest Ridge, Pinnacle) hold about 75% of it. 5 of 46 exposures (the 10 carbon-price scenarios are left out) are large enough to act on; the rest need watching.
Each piece reads from the one before it, so a number on any page traces back to the data room or to a labelled Kith estimate.
Expected loss a year at everything lost (pale) and half lost (solid). The priority four hold 71% of the total.
Each dot is one exposure at one company: how likely over the remaining hold, and how big against its EBITDA. Dashed lines are Northwood's cut-offs (15% of EBITDA, 35% chance).
Hover any dot for its detail. The numbered dots are the large exposures, listed on the next page.
These are the only points in the upper half of the matrix. 2 are likely: spend on them. 3 are unlikely but would hurt: transfer the risk.
Mitigate now
$11.3M
customer data requirement: 49% of EBITDA, 50% chance over the hold. Expected loss $2.2M a year.
Hedge or insure
$8.4M
customer data requirement: 42% of EBITDA, 22% chance over the hold. Expected loss $749K a year.
Mitigate now
$10.9M
customer data requirement: 36% of EBITDA, 50% chance over the hold. Expected loss $5.4M a year. This is revenue not yet won, not earnings in hand.
Hedge or insure
$4.1M
sole-source supplier: 18% of EBITDA, 5% chance over the hold. Expected loss $65K a year.
Hedge or insure
$4.6M
sole-source supplier: 15% of EBITDA, 5% chance over the hold. Expected loss $230K a year.
4 companies have a sized response to customers' data requests (Meridian and Clearview carry the exposure without one). Answering all of them costs about $117K in total and reduces expected loss by up to $8.7M a year (chance-weighted, if the exposed revenue were lost; $4.4M at half lost, Kith's illustration); $5.4M of that is on revenue not yet won. Only revenue the company earns today is counted toward exit value.
| Company | Revenue | Cost to respond | Payback | Protects a year of profit, all lost | Protects a year of profit, half lost |
|---|---|---|---|---|---|
| Pinnacle | Earns today | $9K† | 9 to 18 days† | $749K | $375K |
| Harvest Ridge★ | Earns today | $54K† | 18 to 33 days† | $2.2M | $1.1M |
| Ridgeline | Earns today | $45K† | 90 to 161 days† | $363K | $182K |
| Sterling★ | Revenue not yet won | $9K† | 1 to 2 days† | $5.4M | $2.7M |
Bars show the payback range; the orange line is the company's modelled exit. 10 of 14 measures reach payback after it, so a buyer would have to credit the saving.
Rated on the low end of savings and the high end of cost.
Quickest payback first, with revenue earned today ahead of revenue not yet won. The full ranked list is in the matrix and initiatives page. “Exit value effect” is the low-end annual saving times the company’s exit multiple; whether a buyer pays that multiple is a judgment.
This is Kith's model, the report date plus the remaining hold, not a Northwood plan. Where a hold is a floor of one year it is the earliest modelled date. Where Northwood's own plan differs, use that.
All ten are modelled on the same rules, so the six others can be read the same way.
$9.6M
expected loss a year (71% of the total; figures are rounded, so parts may not add exactly). 12 of the 18 sized initiatives. Sterling, Harvest Ridge, Meridian, Clearview.
$4.0M
expected loss a year. 6 sized initiatives. Keystone, NovaCare, Patriot, Pinnacle, Ridgeline, Trident.
Of 56 exposure lines with a dollar figure, 51 rest on a Kith estimate (†), 5 on a company-reported figure and 0 on official data; 13 of the Kith estimates were built from published benchmarks where the data room was silent. 6 more lines have no figure because the company reported no loss or no exposure; they are listed, not estimated.
Official published data, or an actual from a system of record.
A figure the company reported, an accounting estimate, or a value derived from A and B inputs.
Judgment, a proxy, undated or simulated data. A Kith estimate: no outside source or company has confirmed it. Marked †.
| Estimate | How it was made | What it rests on | Cross-check | Result |
|---|---|---|---|---|
| Insurance premium (four companies with no premium figure; Keystone uses its own $285,000 premium as the base; its $20K† line is the estimated increase) | Median of the five portfolio companies that state a 2025 premium, as a share of LTM revenue, times the US casualty rate change, applied to the whole premium as an upper bound | 0.29% of revenue (range 0.13% to 1.44%, n=5); casualty rate +7% (Marsh, Q2 2026) | RIMS total cost of risk is 1.0% of revenue, 3.4 times the portfolio median | consistent as an upper bound only The RIMS figure includes retained losses and administration, so it overstates the premium; the observed portfolio ratio is used and RIMS is kept as the ceiling. |
| Supplier concentration (eight companies without a supplier file) | McKinsey Global Institute: more than 40% of a year's profit lost per decade over a disruption every 3.7 years, scaled by a dependence factor of 0.25 (people and facility services) or 0.5 (equipment and consumables) | 14.8% of one year's EBITDA per disruption before scaling; resulting impacts 3.7% to 7.4% of EBITDA | Company-specific sole-source impacts from supplier files: sterling 15.2% of EBITDA; harvest 18.1% of EBITDA | consistent in size, not independent The two goods makers with their own supplier files sit at 15 to 18% of EBITDA per disruption, at or just above the 14.8% benchmark. Both are Kith's own one-lead-time calculation, so this is a consistency check, not independent evidence; the service companies are scaled down by judgment, so confidence is low. |
| Efficiency measures without a stated cost or saving (solar, HVAC saving, lighting, sub-metering) | Median cost and payback quartiles of the matching recommendation code in the DOE industrial assessment database | lighting 1.1 to 4.0 years (n=19,471); HVAC 2.3 to 7.0 (n=1,044); solar 6.2 to 14.9 (n=1,590); sub-metering 0.1 to 0.9 (n=28) | Paybacks computed by Kith from the lighting costs and savings two companies state: sterling 1.5 years; harvest 1.0 to 1.5 years | 2 of 2 conservative company paybacks sit inside the lighting range The comparison covers lighting only, because no company states a solar or HVAC saving; the benchmark is untested on those. |
| Water price (Harvest Ridge) | Median supply price of TN and KY plants in the DOE industrial assessment database | $4.16 per 1,000 gallons (n=186) | Context: Nashville raised rates 3% in 2026, the sixth straight annual rise; US typical combined bill +4.6% in 2024 (Bluefield) | small in any case Even doubling the price leaves the line under half a percent of EBITDA; sewer charges are excluded. |
| Patriot electric vehicles | Eight vehicles at the national average mileage, the signed 2025 gasoline price, EPA mpg and EV energy use; cost eight times the EV price gap | 260,780 miles per vehicle would be implied by the reported 76,700 gallons | FHWA average: 11,026 miles a year for cars, 11,360 for SUVs, pickups and vans | reported fuel does not fit eight vehicles The reported fuel is not used; the first move says to check the fleet's mileage first. |
| Reference input: DOE industrial assessment database, recommendation code 2.9114 | Quartiles of implementation cost over annual saving across the recommendations with both stated; median cost | n=1,590; median cost $344,510; payback 6.21 / 9.43 / 14.90 years (25th / median / 75th) | DOE industrial assessment database (pinned reference file), recommendations sheet: recommendation code, implementation cost, annual saving | computed from the pinned file Recomputed on every run; nothing typed. |
| Reference input: DOE industrial assessment database, recommendation code 2.7232 | Quartiles of implementation cost over annual saving across the recommendations with both stated; median cost | n=1,044; median cost $21,464; payback 2.28 / 4.29 / 7.00 years (25th / median / 75th) | DOE industrial assessment database (pinned reference file), recommendations sheet: recommendation code, implementation cost, annual saving | computed from the pinned file Recomputed on every run; nothing typed. |
| Reference input: DOE industrial assessment database, recommendation code 2.8117 | Quartiles of implementation cost over annual saving across the recommendations with both stated; median cost | n=28; median cost $3,000; payback 0.14 / 0.54 / 0.90 years (25th / median / 75th) | DOE industrial assessment database (pinned reference file), recommendations sheet: recommendation code, implementation cost, annual saving | computed from the pinned file Recomputed on every run; nothing typed. |
| Reference input: DOE industrial assessment database, recommendation code 2.7142 | Quartiles of implementation cost over annual saving across the recommendations with both stated; median cost | n=19,471; median cost $4,082; payback 1.10 / 2.26 / 4.02 years (25th / median / 75th) | DOE industrial assessment database (pinned reference file), recommendations sheet: recommendation code, implementation cost, annual saving | computed from the pinned file Recomputed on every run; nothing typed. |
| Reference input: industrial water price, Tennessee and Kentucky | Median of water consumption cost over usage across plants in the DOE database | $4.16 per 1,000 gallons; n=186; assessments FY2016 to FY2026 | DOE industrial assessment database (pinned reference file), assessments sheet: state, water consumption cost, water usage | computed from the pinned file Supply price only; sewer charges are not included, so the line is a floor. |
| Reference input: regular gasoline price | The signed 2025 weekly average of the EIA series 'Weekly U.S. Regular All Formulations Retail Gasoline Prices' (52 weeks) | $3.097 per gallon | the signed assumptions table (assumption G2-A8) and the pinned EIA weekly gasoline file | read from the signed assumption The same price the fleet-fuel measure at Clearview uses. |
| Reference input: commercial electricity price | The U.S. Total row of the EIA commercial price table | 14.53 cents per kWh | the pinned EIA average retail price table, U.S. Total row | read from the pinned file Used for the EV electricity cost. |
| Customer data requirement, Clearview | The company's own estimate of $2-3M of contract value forgone in 2025, valued at gross profit, recurrence Likely | $2.5M midpoint | Company states 40-50% of revenue comes from clients with environmental commitments: not used as a loss, only as context | company-stated Based on an observed loss, so more grounded than a share-of-revenue what-if. |
| Customer data requirement, Meridian | The two requesting clients at the average customer's size | 145 customers on $225M revenue; top customer 11% | Upper bound: two clients at the top-customer share = $50M | derived, low confidence The company states no figure, so the base is derived; the range is shown on the line. |
The work is complete as it stands; these are the limits of what it can tell you.
Every figure on these pages is in the exposure model and the matrix and initiatives page, with its grade; the explorer lets you test them.
This note is for whoever audits or maintains the work. It shows what each part of the process does and how the parts link, without opening a single script. Counts are the record as at 11 October 2026, before the final (G7) freeze.
The charter's 16 rules say what must be true: units, scenarios, bands, missing data, ranking, simulated data and the customer data requirement. Nothing on a page departs from them unless a dated decision says so.
24 run-order steps (87 script versions are kept) build every file from the data room and the reference data. They stop with a plain message if an input, header or period no longer matches, and never overwrite.
Gate checks re-derive figures from source and fail on a defect; each is proved with seeded failures before it is trusted (33 test files). A check that passes is stronger evidence than a reviewer's opinion.
Reviewer agents, each read-only and given one question, work from the same packet: Evidence review, Audience read, Presentation review, Method review, Process review. They raise findings; they cannot change a file.
180 decisions are logged with the date and channel, and the owner's words wherever the owner decided. A builder's proposal is marked proposed until the owner ratifies it.
289 entries are linked to one another, so an earlier entry cannot be altered without breaking the chain. 259 journal entries say what was done, why, and what was left open.
Each gate froze the files it covered. Where a check failed or a reviewer raised a must-fix finding, the work went back, was versioned and re-run. Dates are when each gate was frozen; findings are those raised at that gate.
| Gate | Closed | Checks | Reviewers | Must-fix raised | Should-fix raised | Files frozen |
|---|---|---|---|---|---|---|
| G0 Foundations | 2026-10-09 | 10 | Process review | 22 | 43 | 78 |
| G1 Data and models | 2026-10-09 | 25 | Method review, Evidence review | 4 | 9 | 100 |
| G2 Outputs | 2026-10-09 | 20 | Evidence review, Presentation review, Audience read | 5 | 26 | 104 |
| G3 Close | 2026-10-09 | 14 | Process review | 8 | 6 | 103 |
| G4 Rebuild | 2026-10-10 | 22 | Evidence review, Method review, Process review | 9 | 24 | 250 |
| G5 Explorer | 2026-10-11 | 10 | Evidence review, Audience read | 3 | 17 | 176 |
| G6 Handover | 2026-10-11 | 12 | Process review | 3 | 6 | 175 |
| G7 Revision and final delivery | signed and frozen 11 October 2026 | 13 | Method review, Evidence review, Audience read, Process review | 38 findings, all fixed | 52 | |
Gate G7 covers the revision of this delivery set after G6: the decisions logged after G6, a delivery check with seeded failures, and a reviewer round on method, evidence, audience and process before it was frozen. What was not reviewed: there was no Presentation-lane read of the final pages; the counts in this note are the record at the date shown and are not recomputed; and fixes made after the G7 round were re-run through the delivery check and the re-run test, not re-reviewed.
Reviewers worked from the same packet, read-only, with one question each. These are examples of what they caught, in plain terms.
Caught: A five-year forward hold was applied to every company, but Sterling is already 5.7 years in and Meridian exits in the first quarter of 2027.
Done: Hold years are now set per company from its documented exit plan.
Caught: Clearview's weather events were counted over one year when the company lists three, which overstated how likely they are.
Done: The window is now the three years the company reports.
Caught: Ridgeline's growth-plan investment was stated in its own plan, but the model said it was not stated.
Done: Quantified from the plan's own investment table and shown as a company plan, not capitalised.
Caught: Ridgeline's weather impact had been taken as a property owner's cost, not Ridgeline's own.
Done: Carried as Ridgeline's own small event; the owner-borne losses are shown as context.
Caught: Pinnacle's largest event had combined two storms, lifting its severity band.
Done: Apportioned by closure days; the combined figure is shown only as an upper bound.
Caught: Exit dates were labelled 'planned exit' though they are Kith's model, not a Northwood plan.
Done: Relabelled 'modelled exit', with floors marked as the earliest modelled date.
Caught: Paybacks in days and exit values read as sized benefits though they rest on a what-if.
Done: Labelled 'if the half-lost case happens' and 'what-if, not likelihood-weighted'.
Caught: The decision list shipped to the client was missing most decisions its own files cite.
Done: Rebuilt from the record so every cited decision is present.