Week 2: Data audit

Reading view of a Markdown deliverable, shown as delivered. Download the Markdown file.

Northwood Capital Climate Data Quality Memo: Active Portfolio

To: Sarah Mitchell (COO)
Copy: David Okafor and Lisa Brennan (Operating Partners)
From: Kith Consulting
Date: October 1, 2026 (v6; first issued September 25, 2026)
Period assessed: FY 2025 climate submissions (carbon collection template and climate questionnaire) for all 10 active portfolio companies
Accompanying: Baseline data requests for all 10 portfolio companies (coverage extended from 4 at no additional cost), and a fund-level request to Northwood
Edition: Portfolio edition: Kith Climate cohort case, fictional client. Written as GP-level analysis under Northwood's Climate Data Collection Protocol.
Status: Final text (v6)


Executive Summary

In one paragraph. Northwood's climate protocol exists on paper but has not run for 2026. All ten portfolio companies fall in the lowest readiness band (26 to 38 out of 100). Meridian, in a live $363.9M to $400.9M exit process, is the most urgent. Four fixes protect the most value: reconciling Meridian's exit data, Harvest Ridge's resubmission (about $34.0M of revenue), Sterling's facility-level traceability ($25.0M to $42.0M of defense RFPs) and Northwood's own LP climate section.

Reading the references. D-numbers (D1 to D45) are entries in the disposition log, one per finding with its evidence. A-numbers (A1 to A29) are entries in the assumption log, one per judgment call with its rationale. NW-numbers are items for Northwood to answer; company codes such as NC-6 are asks in that company's data request. Labels in brackets (Confirmed, Reported, Inferred) say how firmly each statement is sourced.

Northwood's climate protocol exists on paper, but there is no sign in the data room that its 2026 cycle has run. No portfolio company has the Tier 2 climate analysis or board climate materials the protocol specifies. The ESG score tables under the new ESG Risk Assessment Methodology hold header rows only. The Q4 2025 LP letter carries no climate section, although the protocol requires one. Climate appears in board or value-creation materials at only 2 of 10 portcos. (Confirmed for the data room Northwood shared; work held outside it wouldn't show here.) The next hard deadline is the Q4 2026 LP letter. The most valuable one is Meridian's confirmatory diligence, due in Q4 2026 on an enterprise value of $363.9M to $400.9M (as at 31 January 2026).

The question "is our data ready?" has two different answers, depending on who is asking:

Ranked by value at stake rather than by data urgency, four issues call for action now:

  1. Meridian's exit data. Reconcile its two submissions against utility invoices before the LOI, on a live $363.9M to $400.9M process (as at 31 January 2026).
  2. Harvest Ridge's resubmission. One clean submission protects about $34.0M of revenue tied to 2027 customer renewals (as at February 2026), and unblocks its baseline.
  3. Sterling's facility-level traceability. It underpins $25.0M to $42.0M of defense RFPs, with Boeing Defense's requirement mandatory from 2027.
  4. Northwood's own LP climate section. The protocol requires one in the Q4 2026 letter, due in mid-February 2027.

Three more carry high value but need a planned, resourced response rather than a data fix: Harvest Ridge's supplier programme, Pinnacle's refinery clients (about $18.0M of revenue (as at 12 December 2025)) and Ridgeline's sustainability services line ($1.5M to $8.0M a year of revenue, sized by its board). Several further items are cheap to fix. They include an FY 2026 insurance budget gap at Pinnacle and Northwood's collection template, which affects all 10 portcos from January 2027. The next section sets out the full ranking and who owns each item.

Figures followed by an "as at" date are older than 90 days, which Northwood's own data-sources standard requires us to flag. Almost every figure in this memo is that old; the vintage table in Scope and Method shows the source and date behind the headline figures.

Where Northwood's Attention Should Go

This memo ranks the findings in two ways, because they answer different questions:

The two rankings diverge, and that is the point:

How the ranking works.

Act now

IssueValue leverWhat is at stakeClockEffort to fixOwner
Meridian: two CFO-signed submissions disagree; no ESG in exit preparationExit value and deal riskLive exit at $363.9M to $400.9M EV; the risk is diligence friction, not a quantified discountLOI and confirmatory diligence, Q4 2026LowPatricia Wells (Meridian CFO); Mark Sullivan (deal lead)
Harvest Ridge: conflicting submissions block the first baseline customers will seeRevenue protectionAbout $34.0M revenue from 3 of top 10 customers requiring Scope 3 data at renewal; expected to spread to 6 or 72027 contract renewalsMediumHarvest Ridge CFO (Thomas Nguyen)
Sterling: facility-level traceability for defense customersRevenue growth$25.0M to $42.0M of defense RFPs; Boeing Defense carbon data mandatory from 2027Boeing Defense requirement, 2027MediumJanet Chen (CFO) and Derek Williams (COO); David Okafor (Operating Partner)
Fund level: 2026 climate cycle not run; no climate section in the LP letterLP reportingProtocol requires a Q4 climate section; Meridian is Fund I's flagship for Fund III fundraisingQ4 2026 LP letter (45 days after quarter end, mid-February 2027)MediumSarah Mitchell and Jennifer Park (IR); Kith drafts in Week 6

Resource and plan

IssueValue leverWhat is at stakeClockEffort to fixOwner
Harvest Ridge: supplier evidence and a board that rates the requirement Low-MediumRevenue protectionThe same about $34.0M; the board treats it as an administrative compliance item2027 contract renewalsHighHarvest Ridge management and board
Pinnacle: refinery clients sending ESG questionnairesRevenue protectionRefinery clients worth about $18.0M a year; not yet a hard requirementNone statedMediumFrank Morrison (CEO) and Paul Stevens (President); David Okafor (Operating Partner)
Ridgeline: sustainability services line and 2027 REIT requirementRevenue growthBoard sizes $1.5M to $8.0M a year; at about a 30% margin (within a 20% to 40% range) that is about $0.5M to $2.4M EBITDA; REIT client about $4.5M requires emissions reporting by 2027REIT requirement 2027; building performance standards in 12 to 18 monthsHighChristine Albright (CFO) and Ridgeline's board; Sarah Mitchell (boundary decision)

Quick wins

Scheduled

Items for the deal team and Operating Partners, not portco finance teams

Three of the highest-value items are decisions rather than data. They sit with Northwood's deal team or Operating Partners rather than in the data requests:

Scope and Method

One process, applied evenly across the portfolio

All 10 active portcos went through the same process at the same depth:

Data vintage. The newest operating data in Northwood's files is from January 2026 (monthly packages for all 10 portcos), and board materials are from Q4 2025. Events dated after that, such as Q3 2026 renewals and updates, are shown as due, with current status unconfirmed.

Source vintage of the headline figures. Northwood's data-sources standard requires a flag on any figure whose newest source is more than 90 days old. Every date below is more than 90 days before this version's date of October 1, 2026.

Headline figureSourceSource date
Meridian enterprise value, $363.9M to $400.9MMeridian exit strategy (exit-prep folder)2026-01-31
Meridian data-room status (90.0%, 75.0%, EY)Meridian management-package workbook, provenance unclearFebruary 2026
Harvest Ridge, about $34.0M revenue at riskHarvest Ridge strategic initiativesFebruary 2026 (month only, flagged 2026-02)
Pinnacle refinery clients, about $18.0MPinnacle climate questionnaire2025-12-12
Facility energy and fuel figures, all portcosFY 2025 carbon templates and climate questionnairesQuestionnaire dates vary, for example 2025-12-12 (Sterling, Pinnacle), 2026-03-01 (Meridian) and 2026-03-20 (Ridgeline)
Sterling defense RFPs ($25.0M to $42.0M), Ridgeline services line ($1.5M to $8.0M) and the other operating figures in the findingsSource dates not re-verified for this versionThe newest operating data in the data room is January 2026

Number formats. Figures follow Northwood's deliverable standards, with one declared exception: amounts under $0.1M are shown in thousands ($15K, $25K, $47K), because rounding to $0.0M would lose the information.

Coverage extended at no additional cost. The engagement priced four priority portcos for this stage. Because the process is scripted and documented, applying it evenly to all ten active portcos cost little more than applying it to four, so we have extended it at no additional cost. The coverage is even, so the findings are comparable. The same framework re-runs each quarter and extends to further portcos at no added cost, whether run by Kith or by Northwood's own team (Appendix B).

One structural difference is flagged rather than smoothed over. The questionnaires differ in how much detail they give:

Reconciliation levelPortcos
Facility by facilitySterling, Harvest Ridge, Meridian, Patriot; Ridgeline for its HQ and Phoenix office
Portco totals onlyClearview, Keystone, NovaCare, Pinnacle, Trident

A portco reconciled on totals alone can hide offsetting facility errors. Sterling's totals match exactly while every facility conflicts. So traceability scores for the second group are the more generous of the two.

How the score works

Each portco is scored 0 to 4 on eight weighted dimensions, then combined into a composite out of 100.

DimensionWeight (%)What it testsConfidence of the evidence
1. Scope 1/2 activity data completeness25.0Electricity, gas and fuel present for every facilityVerified, source quality unknown
2. Data source traceability15.0Northwood's A/B/C source grade, plus whether the portco's two submissions agreeVerified, source quality unknown
3. Emissions calculation10.0The portco has converted its own data to CO2eSelf-reported
4. Targets10.0Stated, dated reduction targetsSelf-reported
5. Governance10.0A management structure that climate ownership sits inSelf-reported
6. Framework alignment10.0Reporting against CDP, GRI, SASB, TCFD or similarSelf-reported
7. Scope 3 visibility15.0Supplier or customer emissions dataSelf-reported
8. Physical risk documentation5.0Facility risk scores plus documented incidents and costEstimated
BandComposite rangeMeaning
Diligence-ready75 to 100Supports LP disclosure or buyer diligence with minor polish
Partially ready40 to 74A usable baseline exists, with specific, namable gaps
Not diligence-ready0 to 39Foundational gaps; the baseline cannot yet be relied on in diligence

Four points of interpretation:

How this relates to Northwood's ESG Risk Assessment Methodology

Northwood's methodology scores residual ESG risk from 1 (minimal) to 5 (high). This memo scores data readiness from 0 (absent) to 4 (complete and verifiable). They are different measures and run in opposite directions: a higher readiness score is better, and a higher risk score is worse. The readiness score tells Northwood how far to trust a risk score built on the same data.

Northwood ESG dimensionReadiness dimensions that indicate how far to trust it
E1 Physical Climate Risk8 Physical risk documentation
E2 Transition Risk1 Completeness, 2 Traceability, 3 Emissions calculation, 7 Scope 3; with 4 Targets and 6 Frameworks as mitigants
G1 Board & Management Oversight5 Governance (climate only)
E3, S1 to S3, G2Not covered by this audit

Northwood's first scoring cycle has not yet been run. We recommend populating the E1 and E2 rows from the Week 5 climate risk register, with each row carrying its readiness score as the confidence note. Scoring E1 or E2 from today's data would put portcos into Northwood's escalation protocol because of data gaps rather than exposure.

LP-Readiness Scores

Portfolio Company12345678ScoreBandOne fix reaches Partially readyFinancial exposure the data supports
Sterling Precision Manufacturing3200002338Not diligence-readyYes$25.0M to $42.0M in defense RFPs; Boeing Defense carbon data mandatory from 2027
Meridian Business Solutions3300001338Not diligence-readyYesConfirmatory diligence in Q4 2026 on $363.9M to $400.9M EV
Clearview Environmental Services3200101336Not diligence-readyYes$2.0M to $3.0M estimated lost on 5 ESG-scored RFPs
NovaCare Health Partners3200101336Not diligence-readyYesAccuracy of Northwood's LP carbon footprint; limited customer pull
Trident IT Solutions3210001336Not diligence-readyYesESG vendor reviews by clients worth about 42.0% of revenue
Pinnacle Testing & Inspection3200001334Not diligence-readyNoRefinery clients worth about $18.0M sending ESG questionnaires; P&C premium $1.7M, rising 12.0% a year
Ridgeline Property Services2100102330Not diligence-readyNoREIT client worth about $4.5M requiring emissions reporting by 2027; $1.5M to $8.0M a year service line
Keystone Behavioral Health3100001330Not diligence-readyNoAccuracy of Northwood's LP carbon footprint; limited customer pull
Patriot Staffing Solutions3100001229Not diligence-readyNo$0.5M weather-closure revenue deferral in FY 2025; heat-related workers' compensation claims
Harvest Ridge Foods2100002226Not diligence-readyYes (one resubmission)About $34.0M revenue from 3 of top 10 customers requiring Scope 3 supplier data at 2027 renewal

Columns 1 to 8 follow the dimension numbering above. Scores are rounded to whole numbers; exact composites are in data-readiness-composites-v2.txt. "One fix" means a single named follow-up, set out under each company below, lifts the score to 40 or more.

The data quality grade distribution that Northwood's protocol asks LP disclosures to reference is in Appendix D.

What Is Measured and What Is Missing

Data categoryMeasuredMissing or unreliable
Electricity and natural gasReported by all 10 portcos for nearly every facilityTotals conflict with the questionnaire at Harvest Ridge, NovaCare, Keystone, Ridgeline and Patriot. Facility figures conflict at Sterling and Ridgeline. Zeros where the questionnaire reports use at Meridian and Patriot. Sites missing or unexplained at Clearview, Pinnacle, Keystone and Ridgeline.
Site listsA 138-row roster covering all 10 portcosThe roster conflicts with the portco's own records at five portcos. Patriot: 12 of 26 sites are not among its 24 branches, and 10 branches are not in the template. NovaCare: 12 of 30 sites match its 28 clinics, and two sites with FY 2025 figures are 2026 openings. Trident: no Chicago SOC row, and five offices are not in its location table. Keystone: Richmond has FY 2025 figures for a clinic opening in Q3 2026. Ridgeline: the Las Vegas office is missing. Harvest Ridge's sites match. Meridian, Sterling, Clearview and Pinnacle's records don't list sites, so their rosters can't yet be fully tested.
Fleet fuelThe template and questionnaire agree at Sterling, Clearview and Pinnacle. Clearview's reported fleet (85 vehicles) is below the 115 vehicles and 18 rigs in its board deck.Harvest Ridge fleet diesel conflicts by 62.2%. Ridgeline's template holds 39.0% of the fleet fuel its questionnaire reports. Meridian reports fleet fuel while stating it has no fleet. Patriot's volume implies about 9,600 gallons per vehicle.
RefrigerantsPartial at Sterling, NovaCare, Pinnacle, Trident and ClearviewR-404A and R-134a can't be recorded (Harvest Ridge). Charge versus leakage is undefined for every portco. Unknown at landlord-managed sites (Patriot).
Emissions (CO2e)Trident only, self-calculatedTrident's Scope 1 looks understated about 2.8x. Not calculated by the other 9. Kith's Week 3 inventory produces the first baseline.
Targets, governance, frameworksA management structure climate could sit in at Clearview, NovaCare and RidgelineNo targets, board climate oversight or framework reporting anywhere
Scope 3Supplier spend at Sterling ($68.7M) and Harvest Ridge ($94.7M). Ridgeline holds energy data for about 85.0% of the 18M sq ft it manages.No supplier emissions data at any portco
Physical riskHazard scores for every facility; weather incidents documented at all 10; multi-year premium trends at 5Score provider unknown (D15). Scores contradict documented incidents at Harvest Ridge (Bowling Green) and Patriot (Memphis and Jackson flood zones).

Findings by Portfolio Company

Ordered by the financial exposure at stake. Each finding carries its confidence: Confirmed, Reported or Inferred.

Meridian Business Solutions (38, Not diligence-ready)

The most traceable data in the portfolio, entering confirmatory diligence.

Fix that moves it: reconcile against utility invoices, taken from the vendor data room if it already holds them. That lifts traceability to 4, for a score of 41 (Partially ready).
Why it matters: two CFO-signed submissions that disagree are a routine buyer-diligence finding. Resolving them before the LOI keeps the ESG section of diligence clean on Fund I's flagship realization, and the exit process may be further along than the exit plan alone suggested: advisor selection was on the March 12, 2026 board agenda, a date that has now passed, so the current status needs confirming.
If it isn't fixed before confirmatory diligence: the likely consequences are a list of buyer questions to answer under time pressure, a slower ESG workstream, and possibly a request for a specific representation on the environmental data. A price effect is unlikely for an office-based business with a small footprint. The cost is management time and credibility at the point in the process where both are scarcest. (Inferred)

Harvest Ridge Foods (26, Not diligence-ready)

The weakest data, attached to the most time-bound revenue exposure.

Fix that moves it: one clean resubmission that settles the period, the conflicting totals and the refrigerants. That gives a score of 40 (Partially ready).
Why it matters: a Week 3 baseline built on the wrong submission would have to be restated in front of the customers holding about $34.0M of revenue. The planned H2 2026 customer Scope 3 readiness programme is the natural owner of the fix.

Sterling Precision Manufacturing (38, Not diligence-ready)

Totals reconcile exactly; the facility detail underneath does not.

Fix that moves it: reconcile the facility split against the three utilities' invoices. As with Meridian, an invoice-checked facility reconciliation lifts traceability to 4, for a score of 45 (Partially ready).
Why it matters: defense primes assess supplier carbon data at facility level, and Boeing Defense's requirement is mandatory from 2027.

Ridgeline Property Services (30, Not diligence-ready)

Ridgeline's climate data problem is also a revenue opportunity.

Fix that moves it: no single fix; it is 10 points away. First step: a corrected template covering all 7 offices and all 40 vehicles.
Why it matters: the service line is revenue creation, and it needs Ridgeline's own baseline first.

Pinnacle Testing & Inspection (34, Not diligence-ready)

The most financially documented physical risk in the portfolio.

Fix that moves it: no single fix; it is about 6 points away. First step: settle the roster and source basis.
Why it matters: a likely FY 2026 budget shortfall driven by Gulf Coast climate pricing is an EBITDA item for the next board meeting, not a data item.

Clearview Environmental Services (36, Not diligence-ready)

Internally consistent totals on an unconfirmed facility roster.

Fix that moves it: confirm the roster with facility-level utility bills, and confirm the fleet count. That gives a score of 40 (Partially ready).
Why it matters: the roster sets Scope 2 grid factors and physical-risk coverage. Unexplained sites undermine the rest of the dataset.

Trident IT Solutions (36, Not diligence-ready)

The only portco that has calculated its own emissions, and the calculation looks wrong.

Fix that moves it: methodology documentation for the Grade B sites, plus reconciled electricity. That gives a score of 40 (Partially ready).
Why it matters: clients worth about 42.0% of revenue run ESG vendor reviews, and an understated footprint in one of them is a reputational risk.

Patriot Staffing Solutions (29, Not diligence-ready)

Climate is already moving Patriot's cost base through heat.

Fix that moves it: no single fix; it is about 11 points away. First step: carry the two branches' figures into the template and confirm the vehicle count.
Why it matters: the heat and workers' compensation link is the one climate item already moving Patriot's EBITDA.

NovaCare Health Partners (36, Not diligence-ready)

Fix that moves it: correct the two rows and explain Baltimore, with facility bills. That gives a score of 40 (Partially ready).
Why it matters: uncorrected, these rows would carry about 156,000 therms of error into Northwood's portfolio footprint.

Keystone Behavioral Health (30, Not diligence-ready)

Fix that moves it: no single fix; it is 10 points away. First step: the CFO's review, with utility or landlord statements replacing the six estimates.
Why it matters: Grade C data has to be flagged in Keystone's board materials under the protocol.

Northwood's Collection Template

Seven fixes to the template Northwood sends every January 15 would improve all 10 submissions at once. They are listed in Appendix E and are a Quick win above.

Items for Northwood to Confirm

  1. Where Meridian's sale process stands now, and whether the ESG section was needed for the March 12, 2026 advisor selection. That date has passed. Meridian's February 2026 management-package workbook reports the financial data room 90.0% complete, the legal data room 75.0% and an EY quality-of-earnings engagement under way; the Q4 2025 board deck says the quality-of-earnings work launches in Q2 2026. The two disagree, and the workbook's origin is unclear (item 7). The live question for Mark Sullivan: what is the current stage, and when does the ESG section need to be clean.
  2. Classification of fuel-card records. We treat them as Grade A-equivalent because they record measured gallons, but the protocol names only utility bills as Grade A.
  3. Physical-risk scores. Who provided them, and when they were retrieved.
  4. The fund-level inputs requested at kickoff. Status of the LP reporting templates, the IC reporting cadence, and any prior ESG consulting output.
  5. Ridgeline's reporting boundary, before Week 3 starts. Should the buildings Ridgeline operates for clients sit inside its footprint, or be reported separately? The answer changes Ridgeline's figures by orders of magnitude, and the Week 3 baseline needs it on Monday. Kith recommends Ridgeline's own offices and fleet as its FY 2025 boundary. The buildings it operates for clients should be recognised as under its operational management, but reported and flagged separately, so that about 285M kWh of client-owned consumption doesn't distort Ridgeline's or the portfolio's picture. Ridgeline's own property register supports this: on the 30.0% of the managed portfolio it covers, 93.8% of the square footage is REIT-, PE fund- or family office-owned, and only 6.2% is owner-occupied. This is subject to a review of the management contracts.
  6. Whether the six additional portco requests go out on Monday. They cover Keystone, NovaCare, Patriot, Pinnacle, Ridgeline and Trident. They could also follow later, or wait for the January 2027 collection cycle.
  7. Where the Meridian and Patriot management-package workbooks came from. Their cover sheets read "Ridgepoint Capital Fund I" and "Ridgepoint Capital Fund II", and the Meridian workbook disagrees with the Q4 2025 board deck on four figures (Meridian findings). Until the origin is explained, Kith does not rely on either workbook; the data-room percentages and the EY engagement stay Reported, on a single source.
  8. Which dates and financials to use for Harvest Ridge and Ridgeline. The firm overview gives entry in January 2025 and March 2025 (revenue $185M and $92M); the closing summaries and data spine give April 1, 2025 and July 1, 2025 (revenue $152M and $95M). Kith uses the closing summaries, which makes ownership 9 and 6 months of 2025; please confirm.

Resolved without needing your input: the Harvest Ridge and Patriot climate questionnaires are both signed "Thomas Nguyen, CFO". The data room's personnel records show two different people with that name, one at each company, so both requests are addressed correctly as drafted.

Data Requests Accompanying This Memo

There is one request per portfolio company, all issued through Northwood. Each contains only that company's own data, with no scores or comparisons, per the protocol's Tier 1 access rule.

Every ask is placed in one of three tiers by a single rule, so portco teams know what to do first:

TierTestDue
1. Blocks the Week 3 baselineWithout it, a baseline figure would rest on a known conflict or gap of 5.0% or more of the company's electricity, gas or fleet fuel, or on a figure implausible on its face at 5.0% or more of the footprint; or it blocks issuing a request5 business days
2. Baseline proceeds provisionallyResolving it changes figures by less than 5.0%, or changes source grade, traceability or scope classification10 business days
3. Needed after Week 3No effect on Week 3 figures; informs materiality, risk and disclosure work in Weeks 4 to 610 business days; partial answers accepted

The tiers set sequence, not importance. Sterling has no Tier 1 asks, because its facility split doesn't change its total, yet its Tier 2 invoices matter most for its defense customers. Harvest Ridge's supplier evidence is Tier 3, though about $34.0M of revenue depends on it.

This timetable depends on the memo being reviewed and the requests issued on Monday, September 28, 2026. Tier 1 responses are then due Monday, October 5, after quarter-end close; Tiers 2 and 3 by Monday, October 12.

The ten requests, their contacts and their ask counts by tier are listed in the accompanying fund-level request (data-request-northwood-fund-level-2026-09-v3-draft.md). The full register of asks, each with its tier, the finding it traces to, the evidence that would satisfy it and the reason for its tier, is finding-criticality-register-v3.csv (79 asks).

Where Tier 1 answers are most at risk.

For all three, the requests offer a partial-answer route, and the Week 3 baseline proceeds on the rule below.

If responses are late, the Week 3 baseline still covers all 10 portcos, on a stated provisional rule.

Every such figure carries a methodology note stating what was assumed and what would close the gap.

Caught During Reconciliation: Meridian's Exit Plan Table

While reading Meridian's exit plan to date its diligence window, we noticed that the sensitivity table in Meridian's exit plan (January 31, 2026) has mislabelled rows. It isn't a climate item, but we thought you'd want to know. At 10.0x and 10.5x it shows equity value under the enterprise value label: at 10.0x on $38.3M LTM EBITDA, EV is $383.0M, but the table shows $341.0M. The 8.5x and 9.0x rows are each $12.0M below the correct EV. The 9.5x base case and the MOIC figures are correct. A buyer or advisor would catch this, so it is worth correcting before materials circulate.


Appendix A: Methodology and Alignment with Northwood Standards

Scoring choiceNorthwood standardWhere Northwood applies it
Traceability graded A/B/CClimate Data Collection Protocol, Data quality hierarchyQA Check 4: grade appears in emissions baseline, board materials and GP dashboard; LP disclosures reference the distribution
A portco's two submissions must agreeProtocol QA Check 1 (every figure traced to a source) and QA Check 2 (structural consistency)All climate deliverables, including LP reporting
Grade B requires methodology documentationProtocol, Grade B definitionSame hierarchy
A Grade C majority with conflicting totals caps traceability at 1Protocol: Grade C is a last resort and must be flagged downstreamSame hierarchy
Risk scores need a provider and retrieval dateProtocol QA Check 1Tier 2 climate risk assessments
Risk scores tested against documented incidentsProtocol QA Check 3, adversarial reviewAll climate deliverables
Findings tied to financial impactProtocol purpose: "Standalone ESG scores without financial context do not leave this firm"All GP-level analysis

Two elements go beyond Northwood's standard and are named as such. The first is the cross-submission consistency test, which is justified by QA Check 1. The second is the electricity-weighted grade view, a stand-in until Week 3 produces the emissions-weighted grade.

Two routes to a higher score are excluded. Kith's Week 3 inventory does not count toward any portco's emissions calculation score. And Clearview's Scope 3 score won't rise from supplier data we chose not to request, because supply chain isn't sector-material for Clearview.

Appendix B: How to Re-Run Next Quarter

Every step is scripted or logged. The scripts use only the Python standard library. The year, file paths, contacts and dates are set in one file, pipeline-config-v1.json. Judgment inputs live in data files, not code.

  1. Place new submissions in the protocol locations: the carbon template under _data/carbon/ and each portco's questionnaire under climate/_submitted/. Set the new year in pipeline-config-v1.json.
  2. Record an inventory of the data room and its data vintage, and register thresholds before reviewing any data (methodology v2, Stages 0 and 0b).
  3. Run verify-climate-data-v2.py for automated checks across all facilities.
  4. Transcribe each questionnaire's figures into the questionnaire extract, with a citation per row. Run a second, independent extraction and compare the two with crosscheck-questionnaire-extracts-v1.py.
  5. Run reconcile-carbon-vs-questionnaire-v3.py. It reconciles the template against the questionnaires for every portco and recovers malformed rows by a stated rule.
  6. Sweep each portco's board, budget and value-creation documents for climate-relevant content, and read the relevant passages in full.
  7. Log every flagged item in the disposition log, with evidence. Log every judgment call in the assumption log.
  8. Score each dimension per climate-data-readiness-rubric-v1.md in the scores file, with a citation for every score. Run score-composite-v2.py and grade-distribution-v3.py.
  9. Update finding-criticality-register-v3.csv and prioritisation-inputs-v1.csv. Run prioritise-findings-v2.py, then build-data-requests-v2.py to generate the requests (company text is in request-narratives-v1.md).
  10. Update this memo, then run verify-pipeline-v2.py. It re-runs every script and checks the memo's figures against the outputs; any failure blocks issue.

Emission factors, with source and retrieval date, are in _reference/emission-factors/. The process is set out in _reference/audit-methodology-v2.md.

Known limits: questionnaire extraction is still a manual transcription, checked by a second extraction. The playbook still lives in Kith's working folder; moving it to a Northwood-owned location is part of the Week 6 handover.

Appendix C: Source Log

SourceUsed for
_data/carbon/carbon-collection-template-2025-submitted.csvEnergy, fuel and source-grade figures
portfolio/{portco}/climate/_submitted/climate-questionnaire-2025.md (10 files)Questionnaire figures and statements
portfolio/{portco}/ board decks, budgets, value-creation plans, KPI dashboards (10 portcos)Governance, commercial and EBITDA links
portfolio/meridian-business-solutions/exit-prep/Exit timeline and valuation
_data/climate-risk/physical-risk-scores.csvFacility hazard scores
portfolio/{sterling, harvest-ridge-foods}/suppliers/supplier-network.csvSupplier spend
_reference/climate-data-collection-protocol.md, esg-risk-assessment-methodology.md, deliverable-standards.mdNorthwood standards
esg/scores/, fund/lp-communications/q4-2025-lp-letter.mdFund-level status
EPA GHG Emission Factors Hub 2025 (eGRID2023), retrieved September 25, 2026; _reference/emission-factors/emission-factors-us-v1.csvTrident Scope 1 check (53.06 kg CO2 per MMBtu); indicative emission shares
Kith working files: disposition log (D1 to D45), assumption log (A1 to A29), data-readiness-scores-v2.csv, reconciliation, cross-check and grade outputsEvery finding and score above
ridgeline-property-services/managed-properties/property-register.csv, ridgeline-property-services/board-materials/2025-q4-board-deck.md, meridian-business-solutions/reporting/meridian-management-package-q4-2025-v1.xlsx, harvest-ridge-foods/management/key-personnel.md, patriot-staffing-solutions/management/key-personnel.mdStage 0 closeout (D41 to D45): Ridgeline boundary evidence, Meridian exit stage, Thomas Nguyen contact resolution

Appendix D: Data Quality Grade Distribution

Northwood's protocol (QA Check 4) requires each portco's data quality grade to appear in its emissions baseline and board materials, and LP disclosures to reference the portfolio distribution. The formal grade is emissions-weighted and needs Week 3 emission factors. The interim view below weights by electricity and shows each label as submitted. Where a portco's own evidence contradicts its label, that share is shown as disputed.

Portfolio CompanyGrade A (%)Grade A-equivalent, fuel card (%)Grade B (%)Grade C (%)Label disputed (% of kWh)Basis of dispute
Sterling Precision Manufacturing100.00.00.00.0100.0Facility figures conflict with the questionnaire despite utility-bill labels (D6)
Harvest Ridge Foods100.00.00.00.0100.0Two submissions conflict on every total (D4)
Ridgeline Property Services100.00.00.00.059.8Regional offices labelled utility bills; questionnaire says square-footage estimates (D23)
Clearview Environmental Services89.510.50.00.00.0None
Meridian Business Solutions82.80.017.20.05.7Boston labelled a utility bill but is a 60.0% allocation (D12)
NovaCare Health Partners75.20.00.024.80.0None
Trident IT Solutions58.70.041.30.00.0None
Keystone Behavioral Health46.50.00.053.50.0None
Pinnacle Testing & Inspection26.55.168.50.00.0Questionnaire cites utility bills; template grades most as B (D30)
Patriot Staffing Solutions12.34.70.083.10.0None

Read this table alongside the scores. The three portcos showing 100.0% Grade A are not the three strongest. Their labels are exactly what their own evidence disputes. The fuel-card share is also unreliable for electricity, because the template allows one source label per row.

Appendix E: Fixes to Northwood's Collection Template

These sit in the template Northwood sends every January 15. Fixing them once improves all 10 submissions.

IssueEffectRecommended fix
Refrigerant fields cover R-22 and R-410A onlyHigh-GWP refrigerants can't be reported; fugitive emissions understatedAdd a refrigerant-type field
Refrigerant fields don't say whether they record charge or leakageBlocks Week 3 fugitive emissions for every portco reporting refrigerantsDefine as annual recharge; collect charge separately
One source label per row, while rows mix fuelsThe per-data-point grade the protocol requires can't be recordedOne source field per fuel
No period-basis or partial-year fieldAnnualized, trailing-12-month and part-year figures mix unannounced (Sterling, Harvest Ridge, Pinnacle, Keystone, Trident)Add period basis and months of data
No row-structure validationMalformed rows at Harvest Ridge and Ridgeline shift values into the wrong columnsValidate field count on receipt
Physical-risk scores carry no provider or retrieval dateThey can't be cited in board or LP materials under QA Check 1Record provider and vintage
Portco identifiers differ from the data spine (Sterling, Pinnacle)Automated joins drop facilities unless aliasedUse spine identifiers

Version notes

What changed from v5 (October 1, 2026): a source-level check of the Stage 0 findings led to six corrections. (1) Meridian's data-room percentages and EY engagement rest on one management-package workbook whose origin is unclear, so they are labelled Reported, not Confirmed (Meridian findings; Items to Confirm). (2) A site-list check found that the carbon template's sites don't match company records at Patriot, NovaCare, Trident, Keystone and Ridgeline; this is now stated in the findings and asked in the data requests (What Is Measured and Missing; company findings). (3) Clearview's fleet is no longer described as fully reported: its board deck lists 115 vehicles and 18 rigs against 85 reported. (4) NovaCare's two rows are described as clinic rows, and the 90.5% is stated as the gap between the template and the questionnaire. (5) Headline figures carry Northwood's staleness flag, and a vintage table shows the source and date behind each (Scope and Method). (6) The March 12, 2026 board date is treated as passed, and the request timetable is flagged as needing a reset. Not changed in v6, and logged for a later version: the Ridgeline square-footage estimate and one pronoun. The data requests are regenerated as v3 with 9 new asks (79 in total).

What changed from v4 (September 27, 2026): a Stage 0 closeout opened five files never read in the original build. One update to Meridian (its exit process is further along than the exit plan alone showed), one to Ridgeline (its own property register supports the boundary recommendation with real ownership data, and softens the "12 to 14M sq ft" self-estimate), and one resolved item removed from what Northwood needs to confirm (the two portco CFOs named Thomas Nguyen are two different people). Everything else is unchanged from v4. Full account: methodology/stage-0-closeout-v1.md.


Portfolio edition — Kith Climate cohort case, fictional client. Prepared by Daniel Vetsko for Kith Consulting.