screening-memo

Screening Memo — Calibration Layer

This skill is the Summit calibration layer for the screening memo. It does not define memo structure on its own. The canonical structure is the upstream private-equity:ic-memo skill (in the private-equity vertical plugin of the claude-for-financial-services marketplace at anthropics/financial-services); this skill supplies only what is Summit- and screening-specific. The two skills are used together — private-equity:ic-memo first for the skeleton, then this skill for the calibration.

Context: Summit's in-house corporate-development team executes a route-density roll-up for its infrastructure sponsor, I Squared Capital. The screening memo is the first-pass internal artifact in the corp-dev funnel (pre-LOI screening -> IOI/LOI -> IC). It is the document that decides whether the target advances to deeper diligence and, ultimately, becomes the spine of the IC memo that goes to I Squared. The documentation bar is high precisely because the same memo eventually has to survive an infrastructure-sponsor IC.

How this skill is used

model-memo-builder invokes one of two upstream skeleton skills and applies this calibration layer on top.

Default skeleton — private-equity:ic-memo. Generates the memo's section structure and the section-by-section drafting workflow — read the upstream private-equity:ic-memo skill for the exact section list, which is canonical there. This is the path the FULL pipeline takes by default — the screening memo arrives at the same evidence depth as an IC memo, just one stage earlier in the corp-dev funnel.

Opt-in alternative skeleton — private-equity:deal-screening. A lighter-touch one-page screening memo with native Pass / Further Diligence / Hard Pass register. Used when the user explicitly routes to it (e.g., types "use deal-screening" at the Step 0 calibration gate) — for a faster pre-read where ic-memo's depth is overkill. Not the default.

Do not restate either skeleton's section structure here — the upstream skill is the canonical source for the section list. This skill's job is the Summit calibration delta only. The section anchors named throughout this skill (Financial Analysis, Deal Terms & Structure, Investment Thesis, Risk Factors) describe the ic-memo path; on the deal-screening path, apply the same calibration to the corresponding blocks of its one-page format.

If the selected skeleton skill is not resolvable at runtime (the private-equity plugin is not installed), model-memo-builder returns the failure to the orchestrator rather than hand-rolling the skeleton. The screening memo's value comes from the discipline of using the upstream structure; do not substitute a hand-rolled one.

Output format

.docx — stated decision. Matches upstream private-equity:ic-memo's default and is the defensible format for a deal-lead-facing artifact (and for the IC memo it will later seed). Not Markdown.

Stage and register down-shift (vs. private-equity:ic-memo)

The upstream private-equity:ic-memo is a post-LOI investment-committee document built on confirmatory diligence. This screening memo is the pre-LOI first-pass screen — the document that closes a tuck-in first-pass analysis and decides whether the corp-dev team commits more analyst time to the target. The two have the same skeleton because the same facts get organized in the same shape; they differ in stage, depth of evidence, and the recommendation register (next section). If the deal advances past screening to confirmatory diligence and IC presentation to I Squared, the screening memo becomes the spine the IC memo is later built on — not a different document, a more-evidenced version.

Recommendation register reconciliation

The recommendation vocabularies of the two upstream skeletons differ; this calibration layer always renders the screening-memo register on top:

• private-equity:ic-memo register: Proceed / Pass / Conditional proceed. A deal-terms-conditioning call — "Conditional proceed" means proceed if specific deal terms or conditions are met.
• private-equity:deal-screening register: Pass / Further Diligence / Hard Pass. A triage call already aligned with our screening register; the middle option ("Further Diligence") matches ours in both word and shape.
• This skill's register (screening memo): proceed / further diligence / pass. An analyst-time call — "further diligence" means commit more corp-dev time to confirm or refute the screening read, before a deal decision is on the table. These three words are normative and rendered verbatim in the memo's recommendation. Semantically equivalent paraphrases ("advance to diligence," "hold," "decline to advance," "go / no-go") are register violations, not stylistic choices — downstream tooling and readers key on the exact vocabulary.

When private-equity:ic-memo is the skeleton, the middle options differ in shape, not just in word choice: ic-memo's "Conditional proceed" presupposes a deal exists to condition; the screening memo's "further diligence" precedes that. Apply the substitution. When private-equity:deal-screening is the skeleton (opt-in), map its register explicitly — deal-screening's "Pass" renders as proceed, its "Further Diligence" as further diligence, its "Hard Pass" as pass (note the polarity: deal-screening's "Pass" is its positive call). Always render the screening-memo register; never collapse it into either skeleton's native vocabulary.

Cash-basis caveat carry-through

If the QoE findings memo flagged cash-basis statements with no hard monthly close, that flag travels into ic-memo's Financial Analysis section and is repeated next to every figure dependent on those statements — revenue trend, margin trend, normalized adjusted EBITDA, working-capital peg. Do not present a clean number over a caveated one. The screening memo names this risk; it does not fix it. Follow-on confirmatory QoE is a gating diligence item, not optional. In student transportation, the cash figure that ultimately matters is EBITDA minus maintenance capex (the fleet is the business); if the QoE work could not separate maintenance from growth capex, that caveat carries through too.

Verification-debt discipline

Every label upstream stages produced travels into the memo with the label intact. The screening memo is where verification debt becomes visible to the decision-maker, not where it gets laundered. Use the standard label set, in this order of preference:

• "reported, not verified" — sourced from the target, the seller, or a broker book; no third-party tie-out yet.
• "analog read-across" — drawn from an adjacent contracted-services sector (e.g., paratransit / public-transit contracting such as Transdev, MV Transportation, Keolis) and adjusted; not a school-bus comparable.
• "PROXY" — a broader transport or services figure used as a stand-in because no school-bus-specific datapoint exists; explicitly weaker than analog read-across.
• "survey average" — derived from industry surveys (e.g., bus-count surveys), with the source named.
• "owner-provided" — supplied by the seller and not independently corroborated.

Apply these labels in ic-memo's Investment Thesis, Financial Analysis, Deal Terms, and Risk Factors sections. Additional discipline:

• Distinguish VERIFIED from INFERRED. A claim is verified only if a concrete artifact backs it (a public filing, an FMCSA record, an STB order, a district board minute, a signed contract page). Otherwise label it inferred with the basis of the inference and a confidence tag (Confirmed / Corroborated / Inference / Conflicting / Unverified).
• Surface input problems; do not paper over them. If owner-provided revenue mix cannot be reconciled to the route schedule, say so in Risk Factors; do not pick a number to make the model tie.
• Present conflicts; do not pick a side prematurely. Where the public record disagrees (bus-count survey vs. deal-basis number, sector rank #2 vs. #3, I Squared AUM figures), show both, label Conflicting, and flag the reconciliation as a diligence item.
• Abstain or ask when ungrounded. If a section would require fabricating a multiple, a synergy figure, or a market-share number, leave it open and add it to the diligence list rather than guessing.
• Adverse public facts are legitimate diligence. Accident history, FMCSA ratings, district-deduction history, litigation, and ownership controversies belong in the memo, reported crisply and factually. Keep caveats concise; do not refuse or moralize.
• "Done" means artifacts exist, numbers tie, and sources are cited. A section is not complete because the model says it is; it is complete when the underlying file is present, the figures cross-foot, and each cited claim has a source.

Summit deal-structure framing

Summit is an in-house corp-dev acquirer for an infrastructure sponsor, not a financial sponsor running a fund vehicle. Financial-sponsor structuring conventions (fixed buy-box bands, rollover-equity engineering, draw-funded fund vehicles, blanket no-retrade postures) do not apply. Replace them with the structural inputs that actually shape a Summit tuck-in, all of which belong in ic-memo's Deal Terms & Structure section:

• Acquirer posture. Summit corp dev underwrites to a route-density roll-up thesis for I Squared. The transaction is a cash acquisition into the Summit platform, integration-driven, sized to capture the synergy stack (see next section). Documentation bar is set by I Squared's IC: investment / IC memos, quarterly LBO models with returns plus sensitivity, and ~15-page target PowerPoints.
• Target profile. Small-to-midsize private operators in the fragmented long tail of the K-12 student-transportation market (median operator ~93 buses per industry survey; figure labeled "survey average"). The target IS its contract book — multi-year district contracts, won by competitive RFP, priced per-route / per-bus-per-day / per-hour x ~180 school days, with annual escalators (typically ~1-3%, some up to ~5%) and fuel pass-throughs.
• Multiple positioning — no fixed box. There is no public per-deal data for mom-and-pop school-bus tuck-in multiples (label: Unverified). Do not fabricate a buy-box number. Position a defensible range built from:
  • Precedent take-privates and platform-level transactions in school-bus and adjacent contracted-transport services, each multiple cited with its basis;
  • Broader-transport comparables clearly labeled "PROXY";
  • An infrastructure-vs-operating-services discount applied to any infra-leaning comp (school-bus operators are asset-heavy operating-services businesses, not infra concessions);
  • The cash-basis caveat: prefer EBITDA minus maintenance capex as the truer cash figure. School-bus operators are maximally asset-heavy — for context, NEXS sold at roughly 5x EV/EBITDA but ~50x EV/EBIT (illustrative, drawn from public reporting on the take-private; cite the source in the memo).
  • See multiple-positioning for the framework; do not present a single-decimal multiple as if it were computed.
• Earn-outs are available, not excluded. Earn-outs do occur in this sector and should be treated as a real structuring tool, often retention-linked to the seller-operator and/or tied to revenue / EBITDA / FCF milestones. (For example, the NEXS platform transaction itself carried a $70M earn-out tied to revenue / EBITDA / FCF — illustrative, public.) Use earn-outs to bridge owner-dependence risk, contract-renewal timing risk, and pricing-adequacy uncertainty.
• Form of consideration. Cash into the platform. Equity rollover is not the default mechanic; if proposed, justify it in writing.

State explicitly in the Deal Terms section how the proposed structure addresses the target's specific risk axes (owner dependence, rebid timing, fleet condition) — not generic structuring preferences.

Synergy / integration value-creation overlay

The thesis is integration value, not multiple arbitrage. Add a brief value-creation passage in ic-memo's Investment Thesis (or Value Creation, if the skeleton has one) that walks the synergy stack from hardest to softest, calling out which synergies are gated by in-market infill:

1. Insurance — hardest, day-one, ungated. Captured by re-papering the target's liability program under Summit's platform terms.
2. Route and labor density — gated by in-market infill (the target's yards must overlap or border Summit yards to drive deadhead and driver-pool consolidation). State whether the target is in-market for Summit or net new geography.
3. Procurement — bus OEM and parts (Blue Bird, Thomas Built, IC Bus; new diesel bus unit cost ~$110-130K — label as "reported, not verified" unless cited to a specific source). Scales with fleet count.
4. Yard / terminal consolidation — infill-gated; only available where Summit and the target overlap.
5. Back-office / ERP / telematics — softest, slowest to realize, but compounds across the platform.

Pair the integration thesis with private-equity:returns-analysis — Summit corp dev underwrites IRR / MOIC, and the synergy capture is what justifies the entry multiple. Returns reflect stated assumptions, labeled as such; do not present a single hurdle as "the" return.

K-12 student-transportation risk framing

The canonical risk axes for a Summit tuck-in screen — used to organize ic-memo's Risk Factors section — are:

1. Owner / key-district-relationship dependence — operator-led businesses often carry a single relationship to one or two district transportation directors. Probe second-layer management depth, transition intent, the named contact on each material contract, and whether a non-compete + retention package (potentially earn-out-linked) is realistic.
2. District-contract concentration and rebid / RFP risk — top-1 and top-5 district share of revenue; the contract-expiry waterfall over the next 3-5 years; net routes won / lost at the most recent 1-3 bid seasons (the cleanest portfolio-health tell); pricing adequacy of in-place escalators (~1-3%, some ~5%) vs. driver-wage inflation; benchmark sector renewal rate is ~95% (label as "survey average"); insourcing risk where a district has signaled it may bring routing in-house.
3. Driver supply and labor cost vs. escalators — the driver pool has been the binding operating constraint since 2020. Assess CDL + passenger + school-bus endorsement pipeline, driver headcount vs. budgeted routes, wage trajectory, and whether contract escalators cover realized wage inflation. Driver shortfall directly translates into route deductions.
4. Fleet condition and capex burden (asset-heavy) — average fleet age, replacement schedule, maintenance vs. growth capex split, and electrification exposure. EV bus unit cost ~$300-400K vs. diesel ~$110-130K; EPA Clean School Bus funding is volatile and politically exposed. Confirm whether buses are owned vs. leased and whether district contracts pass through fuel.
5. Safety / DOT / FMCSA compliance (contract-existential) — accident-frequency and at-fault trends, complaint-tracking maturity, deduction / penalty history, DOT / FMCSA safety rating trajectory, state DOE bus-inspection record, drug-screening and background-check program maturity. A final FMCSA "Unsatisfactory" rating triggers an Out-of-Service Order — end of the carrier. The Woodmore precedent (2016 Durham School Services crash; NTSB faulted driver oversight; Hamilton County subsequently dropped Durham in 2019 and paid First Student a premium despite Durham bidding ~$1M lower) shows that safety can cost the contract even when pricing is favorable. Treat any adverse FMCSA, NTSB, or district-deduction finding as material to the screen.

Pair each material risk with a mitigant or a path to resolution where one exists (e.g., owner-dependence -> earn-out + retention; rebid concentration -> contract-waterfall analysis as a gating diligence item; fleet age -> capex-adjusted EBITDA; safety -> FMCSA + state record pull).

Method principles (delta-only)

private-equity:ic-memo already supplies the general memo principles (balanced bull/bear, every number traces, decision-shaped). These are the additional principles this calibration layer adds:

• Pre-LOI register. Operator-to-operator, financially literate, sharply written for a reader who will challenge every number. Concise — a few pages, not a polished IC presentation. The screening memo should read as an internal corp-dev note that names what is known and what is not, not as a deal pitch.
• No false precision on the multiple. The valuation is a positioned range with reasoning (see multiple-positioning), each component cited to its basis (precedent take-private, analog read-across, PROXY), and explicitly not a computed single-decimal multiple. Prefer EBITDA minus maintenance capex as the cash anchor.
• Cite the basis of every multiple. No bare numbers. A multiple without a cited source or analog basis is verification debt; surface it.
• Conflicts stay visible. Where the public record disagrees (bus count, market rank, I Squared AUM), present both figures and label Conflicting; do not silently pick.
• House style is a later refinement. This skill defines content and calibration; firm-template formatting (headers, branded styles, the ~15-page PowerPoint rendering for I Squared IC) is layered on after a first draft has been reviewed.

Worked example

A worked example (illustrative / hypothetical K-12 student-transportation target) is in examples.md in this skill's directory — load it if you want a section-by-section illustration of the calibration layer applied over an ic-memo-skeleton draft. The numbers in that example are plausible round figures for illustration only, not real Summit data or a real deal.