Methodology Guide

This suite uses vocabulary the other seven do not. An acquisition runs on a different clock, with a different failure mode, and several of its terms mean something narrower than they appear to. This guide explains the practices; the Glossary defines the terms, and every term below links to its entry. ⚠ For the annotation colors, the point-in-time convention and what is fictional, see How to Read This Suite — this page does not repeat them.

1. Why an acquisition is not a project

A project delivers something that did not exist. An integration takes two organizations that already work and makes them one without either stopping. ACME Health (1,800,000 members) acquired Cumberland Valley Health Plan (420,000) for $1.2B, on a commitment of $85M of Run-rate synergy by the end of Year 3. Every artifact in this suite answers to that number.

The difference that matters is the failure mode. A project fails by not delivering. An integration fails by never finishing — two organizations still running parallel processes eighteen months after close because nobody forced a decision. That is why the disposition matrices are the centerpiece of this suite and why every one of them names a date.

2. You may not integrate before you close

Between signing (February 13, 2023) and closing (September 29, 2023) the two companies remain competitors, and acting as one is Gun-jumping — an antitrust violation rather than a matter of etiquette. Yet the buyer must plan an integration that depends on knowing what it is buying.

The mechanism that resolves it is the Clean Team: engaged by Legal, permitted to look, returning conclusions rather than records. ⚠ Most readers have never heard of one, which is precisely why the protocol is in this suite.

This is the constraint the whole program hangs on. Because member-level data could not lawfully be examined before close, the first cost estimate was produced blind — see section 8. The overrun that follows is a consequence of the law, not of the estimating.

3. Day 1 is not integrated

Day 1 is the first day of common ownership: October 2, 2023. Members keep coverage, providers keep getting paid, regulators see continuity. Integrated is a different state entirely, and on this program the two are separated by roughly fifteen months.

⚠ Conflating them is the classic first-timer error, and it is expensive in a specific way: a plan that treats Day 1 as the finish line leaves no funding, no team and no governance for the fourteen months where the actual integration happens. Day 1 readiness certifies continuity. It certifies nothing about consolidation. Day 100 is a checkpoint on the way, not a gate.

4. A TSA obliges a service, not a capability

A Transition Services Agreement (TSA) keeps the seller supplying named services after close. The sentence that matters: it obliges the seller to provide the service. It does not oblige the seller to make the buyer capable of performing it. Those are different obligations, and only one of them ends the dependency.

So knowledge transfer is contracted as a deliverable with named receivers and Reverse shadowing acceptance — the buyer performs while the provider watches. And because each service can only be switched off once the buyer can run it, TSA exit is the integration. The migrations exist to make the exits possible, not the other way round.

5. Carve-out, and why there is a seller left to exit from

This is a Carve-out: Cumberland Valley Health Plan was a subsidiary of a parent that continues to exist and continues to provide shared services on a meter. In a Whole-company merger the shareholders take cash and the selling entity disappears — there is no continuing counterparty, so there is nothing to transition from.

The carve-out is what makes the TSA coherent, and it also explains the shape of the schedule: functions the parent runs centrally have to move fast, because every month of dependency is billed by a counterparty with no commercial incentive to hurry.

6. Assignment, novation and change of control

Inherited contracts do not transfer by being listed on a matrix. Assignment passes the benefit and can leave the original party liable; Novation replaces the party outright and requires all three to agree. Change-of-control clause terms decide whether the counterparty even has to — 12 inherited agreements carry one here, and 3 give an outright right to walk.

⚠ Business Associate Agreement (BAA) coverage is the Day 1 gate item: 16 are required, and the one most often missed is the divesting parent itself, because nobody files the seller under vendors.

Leverage is set by how long you still need someone, not by how long the relationship has left to run. The vendor being terminated has the most leverage, because the buyer depends on it continuously through cutover. Sweep those contracts first.

7. Insurance entities do not consolidate like companies

Insurance policies are contracts with members, so a book of business cannot be moved the way a company can. Consolidation runs through either a Statutory merger approved by the domiciliary regulator, or Assumption reinsurance under which the acquirer assumes the policies. Both are regulated acts with notice requirements and lead times measured in quarters, not sprints.

8. Why the first estimate was guaranteed to be wrong

An AACE estimate class describes how much an estimate can be trusted, and Progressive elaboration is the principle that it improves as more is known. Ordinarily a Class 5 estimate is early. Here it was compelled: see Class 5, legally imposed.

The consequence is arithmetic, not narrative. The clerical review band was assumed at 8% of records and returned at 26% once the data could lawfully be profiled — see Clerical review band. Nothing about that gap was a failure of effort.

The defensible interview answer. The estimate was not predicted badly; it was structurally guaranteed to be unpredictable. The useful discipline is not “estimate better” but identifying, at charter, which figures are legally unknowable — and funding a re-baseline for them.

9. Two baselines, and the difference between them

Baseline 1 — the deal model was the pre-close estimate. Baseline 2 — the re-baseline was the first produced with the data. The step between them is a Re-baseline, and it is an event, not Change control — change control governs movement away from a baseline and cannot govern the arrival of the first credible one.

From Baseline 2 onward the change log carries the whole reconciliation: $7.6M approved against a final outturn of $59.4M, inside $60.1M authorized. $1.9M of scope was deferred to named owners rather than absorbed into cost.

⚠ A change log that does not reconcile exactly is decoration. The test is whether baseline plus approved changes equals the closeout position. That is what makes an ending auditable rather than merely stated.

10. Margin is not float

Float is a by-product of the network; nobody bought it. Margin was negotiated at signing — the gap between the planned TSA exit and the contractual maximum of March 31, 2025. Spending margin is a Steering Committee decision, not a scheduling adjustment.

This program exited on December 31, 2024 with 90 days still unused — Late but whole. Blowing a contractual maximum is a governance failure; spending a buffer you deliberately bought is a win, and it retroactively justifies having negotiated the term.

⚠ A program that returned every margin day unused would not have been better governed. It would have been over-insured — margin costs something to buy, and buying more than the risk warrants is its own error.

11. Where to go next

If you read one artifact, read the Application Disposition Matrix — it is where the strategy is actually decided, system by system. Then Data Profiling Results for the crisis, and Lessons Learned for why it was structurally guaranteed.

Related: Glossary · How to Read This Suite · Suite index