A private equity model can be precise to the second decimal place and still produce a vague operating plan.
That gap used to be survivable. Multiple expansion, cheap debt and shorter holding periods could absorb a surprising amount of operational ambiguity. The current return environment is less forgiving.
Bain’s 2026 Global Private Equity Report calls the shift “12 is the new 5.” Its analysis says a typical deal in the 2010s needed about 5% annual EBITDA growth to produce a 2.5x multiple of invested capital over five years. Today, the same outcome can require roughly 10% to 12% annual EBITDA growth because leverage is lower, borrowing costs are higher and purchase multiples remain elevated.[1]
That is not a small adjustment to the 100-day plan. It changes the operating system for ownership.
If your post-close plan says “optimize pricing,” “professionalize sales,” “capture procurement savings” and “accelerate AI,” you do not have a value-creation system. You have a list of verbs.
Here’s what works: reverse-engineer the underwriting case into a Weekly EBITDA Driver Tree that reaches the portfolio company’s operating meeting. Every material value lever gets a baseline, an owner, a leading indicator, a dependency, a downside case and an evidence rule.
That is how the thesis stops living in the deal model and starts running the company.
The 100-day plan has an altitude problem
Most 100-day plans are built at the wrong level.
The investment committee approves a return case with detailed assumptions. Management then receives a collection of initiatives. The board sees monthly or quarterly financial results. Between those layers, nobody has a reliable view of whether the mechanisms behind the thesis are working soon enough to intervene.
EBITDA is a lagging outcome. By the time a miss appears in the monthly close, the operational failure may be eight or twelve weeks old.
A pricing initiative can fail because salespeople keep discounting. A retention plan can fail because onboarding defects are creating avoidable churn. A procurement target can fail because contracts renew before negotiations start. An AI productivity case can fail because saved hours never become removed cost, added capacity or incremental revenue.
All four may still look “on track” in a red-amber-green slide deck.
I have executed more than 15 acquisitions and built a software and infrastructure business to €240 million in ARR before a €1.5 billion exit. The lesson is consistent: a synergy belongs in the base case only when it has an owner, a systems path, a delivery date and evidence that survives the board pack.
Everything else is optimism wearing a spreadsheet.
The Weekly EBITDA Driver Tree
The framework has four levels. Each level answers a different operating question.
Level 1: Required value creation
Start with the outcome the deal must produce, not the initiatives people already want to run.
Document the five-year EBITDA bridge required by the underwriting case. Separate the contribution expected from revenue growth, gross-margin expansion, operating efficiency, add-on integration and any exit assumptions.
Do not hide a weak operating case behind multiple expansion. Keep exit multiple, deleveraging and operational improvement visible as separate components of value.
The first question is simple: How much EBITDA must exist, by when, for the return case to hold without heroic assumptions?
Level 2: Explicit value levers
Break the required EBITDA growth into named mechanisms:
- pricing and discount discipline;
- gross retention and expansion;
- sales productivity and conversion;
- product or service mix;
- procurement and vendor consolidation;
- workflow automation and capacity release;
- shared services;
- add-on revenue or cost synergies.
Each lever needs a baseline in currency and operating units. “Improve retention” is not a lever. “Reduce gross revenue churn from 11% to 8%, protecting €1.2 million of annual recurring revenue at a 78% gross margin” is a lever.
This is where bad assumptions should become uncomfortable. If the baseline cannot be reproduced, the value claim is not ready for the plan.
Level 3: Operating mechanisms
For every lever, identify the mechanism expected to move it.
Pricing may require a new approval workflow, contract segmentation, renewal timing and sales compensation changes. Automation may require process redesign, structured data, exception handling, human approval and adoption by the people doing the work. Add-on synergies may depend on CRM consolidation, product packaging or customer consent.
Assign one accountable executive. Contributors can be many; accountability cannot.
Then tag each mechanism as either standalone or integration-dependent. This distinction matters. A portfolio company can change its discount policy immediately. A cross-company upsell motion may depend on unified account data, aligned incentives and a common commercial process. Those do not deserve the same probability or timeline.
Level 4: Weekly proof
Attach one leading indicator to each material mechanism.
Not ten metrics. One signal that tells the operating team whether the mechanism is beginning to work.
For example:
- pricing: percentage of quotes outside the approved discount corridor;
- retention: number and value of at-risk renewals without an agreed rescue plan;
- sales productivity: qualified pipeline created per fully ramped seller;
- procurement: spend entering renewal within 90 days without a named negotiation owner;
- automation: accepted output per human review hour, including exception handling;
- integration: percentage of target accounts mapped to a verified cross-sell owner and offer.
Weekly indicators are not mini financial statements. They are early-warning instruments. Their job is to expose a broken mechanism while there is still time to repair it.
The seven fields every material lever needs
A useful driver tree is not a pretty diagram. It is a control system. Every material lever needs seven fields.
1. Baseline. The current performance, source system, measurement date and finance owner who can reproduce it.
2. EBITDA contribution. The gross benefit, implementation cost, recurring cost, timing and confidence-adjusted net contribution.
3. Accountable owner. One executive with authority to change the mechanism, not merely report on it.
4. Weekly leading indicator. A signal that moves before EBITDA and has a defined source and update cadence.
5. Dependency. The system, hire, contract, integration, customer approval or management decision that can block delivery.
6. Downside case. What the contribution becomes if timing slips, adoption stalls or the mechanism produces only half the expected result.
7. Evidence rule. The proof required before the lever can be counted as delivered: finance-reconciled savings, realized price, retained revenue, paid invoices or demonstrably redeployed capacity.
This last field is where many AI cases collapse. Hours saved are not EBITDA. A faster workflow only creates value when the business removes cost, absorbs more volume with the same team, improves conversion or increases retention. If finance cannot reproduce the bridge, the claim stays outside the realized case.
Build the tree from the board room down—and from operations up
There are two bad ways to build this system.
The first is top-down only. The deal team divides the EBITDA target across initiatives, assigns owners and calls the plan complete. The numbers add up, but the operating mechanisms do not.
The second is bottom-up only. Functional leaders submit projects they already wanted, attach benefits and hope the sum resembles the investment thesis. The projects may be sensible, but they are not necessarily sufficient.
The driver tree forces both directions to meet.
From the top, the return case defines the required contribution and timing. From the bottom, operators define what can physically change, which dependencies exist and what evidence will appear first. Any gap between the two is a decision, not an invitation to invent another initiative.
You either increase the ambition of a credible mechanism, fund a new one, extend the time horizon, change the capital structure or revise the thesis.
That conversation is uncomfortable. It is also far cheaper in month one than in year three.
Where AI fits—and where it does not
AI can make this operating system faster, but it cannot rescue weak ownership.
Here’s where it works:
- unify financial, CRM, ERP, ticketing and procurement data into a governed metric layer;
- draft weekly variance commentary with links back to source evidence;
- flag stale baselines, missing owners and dependencies approaching a deadline;
- detect changes in leading indicators before the monthly board pack;
- maintain a portfolio pattern library showing which mechanisms worked, in which context and at what cost;
- generate exception queues for management rather than another dashboard to browse.
The objective is not an autonomous board member. It is a tighter evidence loop.
Humans still choose the thesis, challenge assumptions and allocate resources. AI reduces the cost of collecting, reconciling and explaining the evidence. The best system gives the operating partner and management team superpowers; it does not pretend accountability can be automated.
Ownership matters too. The driver tree contains sensitive deal assumptions, portfolio performance and management judgments. Build the data and workflow layer so the fund and portfolio company own the logic, history and evidence—not a black-box SaaS vendor.
A 30-day proof path
Do not launch this across the portfolio. Prove it on one company and one material part of the thesis.
Days 1–5: Rebuild the bridge
Select the three largest operational contributors in the underwriting case. Reconcile each baseline with finance. Name the accountable owner. Separate standalone value from integration-dependent value.
Kill any lever that has no reproducible baseline or operating mechanism. It can return when the evidence exists.
Days 6–10: Define weekly signals
Choose one leading indicator for each lever. Document the source, update cadence, acceptable range and escalation threshold. Make sure management can influence it weekly.
If the signal cannot trigger a decision, it is reporting noise.
Days 11–20: Build the evidence loop
Connect the minimum required data sources. Automate collection where stable; use controlled manual inputs where they are faster for the proof. Create one exception queue with an owner and due date for every variance.
Do not spend the month building a perfect data platform. The proof is whether the operating meeting gets a faster, more accurate decision.
Days 21–30: Run four operating cycles
Review the indicators weekly. Record decisions, owners and expected effects. At day 30, compare the original thesis, actual mechanism performance and finance-reconciled value.
Then make one of three decisions: scale, redesign or stop.
Success is not a polished dashboard. Success is finding one mechanism that is working and deserves more resources—or exposing one that was never real before it consumes another quarter.
The new standard for a credible ownership plan
Bain’s “12 is the new 5” is a useful shorthand, but the operator implication is bigger than the headline. Double-digit EBITDA growth cannot be managed as a quarterly narrative.
It needs a weekly system that connects return requirements to operating mechanisms and operating mechanisms to evidence.
That system should make weak assumptions visible, distinguish standalone improvement from integration-dependent synergy, and force every material lever to earn its place in the thesis.
A 100-day plan can still be the starting point. It just cannot remain a list of initiatives.
Turn it into a driver tree. Put it in the operating meeting. Demand 30 days to proof.
If you want to build the first version around a live portfolio thesis, Book a 30-minute strategy call.
