Introduction: Why AI Changes the 100-Day Playbook
You've closed the acquisition. The integration team is mobilised. The traditional 100-day plan is already on the wall: baseline the financials, identify quick wins, stabilise revenue, optimise working capital. It's solid. It's proven. But it's also incomplete.
The operating partners who move fastest on AI in the first 100 days after close don't just capture incremental EBITDA improvements—they unlock structural value creation that compounds through the hold period. This isn't about pilot projects or proof-of-concept exercises. This is about shipping production-ready AI systems that move the needle on unit economics, operational efficiency, and customer experience within 90 to 100 days of acquisition close.
The difference between a PE firm that treats AI as a 2025 initiative and one that treats it as a Day 1 value lever is roughly 300–500 basis points of EBITDA uplift across a 5–7 year hold. That's not hyperbole. That's the gap between firms deploying agentic automation, intelligent process optimisation, and AI-driven revenue acceleration in the first quarter versus those waiting for the traditional strategic planning cycle.
This playbook is written for operating partners, CFOs, and transformation leads who need to move fast, measure precisely, and avoid the pilot trap that kills 70% of enterprise AI initiatives before they reach production. We'll walk through the specific decision framework, the architecture patterns that actually work at scale, the governance structures that enable speed without chaos, and the sequencing that lets you stack wins.
The 100-Day AI Value Creation Framework
Traditional 100-day plans focus on four levers: revenue protection, margin expansion, working capital optimisation, and synergy capture. AI doesn't replace these—it amplifies them. But it requires a different sequencing and a different set of decisions.
The AI-native 100-day framework has five parallel workstreams:
Workstream 1: Diagnostic and Opportunity Mapping (Days 1–20)
Before you architect anything, you need to know what you're solving for. This isn't a six-month discovery phase. It's a focused 20-day sprint to identify the top 3–5 value creation levers where AI can move the needle in the next 80 days.
Start with the financials. Where is the business bleeding cash or leaving money on the table? In a logistics company, it's route optimisation and last-mile delivery costs. In healthcare, it's clinical operations scheduling, billing accuracy, and patient no-shows. In hospitality, it's dynamic pricing, housekeeping efficiency, and guest complaint resolution. In financial services, it's compliance review cycles, fraud detection, and customer onboarding.
The diagnostic should answer these specific questions:
- What processes consume the most labour hours without requiring deep domain expertise?
- Where is decision-making slow, inconsistent, or based on incomplete information?
- Which customer-facing workflows have the highest friction or abandonment rates?
- What data exists but isn't being leveraged for decision-making?
- Which revenue or cost levers move the P&L by more than 50 basis points when optimised?
During this phase, you're also assessing the technical readiness of the organisation. Can they integrate APIs? Do they have data governance frameworks? Is there a single source of truth for customer, operational, or financial data? These aren't blockers—they're inputs into your sequencing.
Workstream 2: Architecture and Model Selection (Days 15–35)
While the diagnostic is running, your engineering team should be locking in the technical architecture. This is where many PE-backed firms go wrong. They assume all AI problems require custom fine-tuned models or proprietary datasets. Most don't.
The fastest path to production AI in 100 days uses frontier models—Claude Opus 4, GPT-4 Turbo, or Gemini 2.0—with retrieval-augmented generation (RAG) for domain-specific knowledge, deterministic validation layers for high-stakes decisions, and integration points that sit on top of existing systems rather than replacing them.
For a financial services firm, this might mean deploying an AI agent that reads compliance documents, flags exceptions, and routes them to the right reviewer—not replacing the reviewer, but cutting review time by 60%. For a healthcare system, it's an AI workflow that schedules clinical staff based on patient acuity, bed availability, and staff certifications—reducing scheduling time from 4 hours to 15 minutes and cutting overtime by 25%.
The key decision: agentic architecture versus supervised learning. Agentic systems (AI agents that take actions, call APIs, make decisions within guardrails) are faster to production and easier to measure. Supervised learning models require more data, more time to train, and more complex deployment infrastructure. Unless you're optimising a high-frequency decision with millions of examples, you want agentic workflows.
During this phase, you're also locking in the governance layer. This means defining what decisions the AI can make autonomously, what requires human review, what audit trails are required, and what happens when the system encounters an edge case. This isn't bureaucracy—it's the difference between a system that runs for 90 days and one that runs for 5 years.
Workstream 3: Data Preparation and Integration (Days 20–50)
Data is the constraint. Not model capability. Not infrastructure. Data.
You need to know: What data exists? Where is it stored? What's the quality? What's missing? Can you access it programmatically? What's the latency requirement?
For most PE-backed companies, the answer is scattered. Customer data in Salesforce. Operational data in a legacy ERP. Financial data in a disconnected accounting system. Clinical data in multiple EMRs. Guest data in a property management system that hasn't been updated since 2015.
Your integration strategy should assume you're not rebuilding the data architecture in 100 days. You're building connectors and transformation layers that sit on top of what exists. This means APIs, ETL pipelines, and middleware that can pull data from multiple sources, standardise it, and serve it to your AI systems.
Data quality is the second constraint. If your customer data has 40% missing phone numbers or your clinical data has inconsistent terminology, your AI system will learn those inconsistencies. The 100-day approach is to accept some data quality issues and build validation rules into the AI system itself—flagging uncertain decisions, requiring human review for edge cases, and continuously improving the underlying data as you go.
Workstream 4: Pilot Deployment and Measurement (Days 40–80)
This is where most organisations fail. They pilot with a small team, see promising results, and then spend 6 months trying to scale. Instead, you should pilot with a specific, measurable outcome in mind and scale immediately if the outcome is positive.
The pilot should answer: Does this system deliver the promised outcome? Can it be integrated into existing workflows? What's the actual latency, accuracy, and cost? What breaks?
For a 100-day deployment, your pilot should be 2–4 weeks, not 8–12 weeks. You're not optimising for perfection. You're optimising for "good enough to move the needle" and "safe enough to scale". This means accepting 85–90% accuracy on non-critical decisions, requiring human review for edge cases, and iterating rapidly based on real-world feedback.
Measurement is non-negotiable. Define your KPIs before you deploy: time saved per transaction, error rate reduction, cost per transaction, customer satisfaction, revenue impact. Measure them daily. If you're not seeing movement in 2 weeks, kill the pilot and move to the next lever.
Workstream 5: Scaling and Governance (Days 70–100)
If the pilot works, scaling should be fast. You've already built the integration, you've already trained the team, you've already locked in the governance model. Scaling is about expanding the scope—more users, more transactions, more workflows—while maintaining the same quality and safety standards.
This is also where you're building the operating model for ongoing improvement. Who owns the AI system? How do you handle model updates? What's the process for adding new use cases? How do you handle customer or employee feedback?
The best PE-backed AI deployments treat the AI system like a product. There's a product owner, there's a backlog, there are sprints, there are metrics. It's not a one-time project. It's an ongoing capability that evolves with the business.
Value Creation Levers: Where AI Moves the P&L
Not all AI opportunities are created equal. Some deliver 50 basis points of EBITDA improvement. Others deliver 500. Here's where to focus in the first 100 days.
Revenue Acceleration (200–400 bps EBITDA impact)
AI can move revenue through three mechanisms: reducing customer acquisition cost, improving conversion rates, and increasing customer lifetime value.
For a SaaS business, this might mean deploying an AI agent that qualifies leads, answers objections, and books demos—cutting sales cycle time by 30% and increasing conversion rates by 15%. For a hotel group, it's dynamic pricing that optimises room rates based on demand, competitor pricing, and booking patterns—increasing revenue per available room (RevPAR) by 8–12%. For a healthcare system, it's an AI workflow that identifies patients at risk of no-show and proactively contacts them—reducing no-show rates from 20% to 8% and improving patient throughput by 15%.
The common thread: you're using AI to make better decisions faster, at scale, with less human intervention. You're not replacing salespeople or revenue managers—you're giving them better information and automating the low-value parts of their job.
Cost Reduction (150–300 bps EBITDA impact)
AI can reduce costs through labour automation, process optimisation, and resource allocation.
Labour automation is the most obvious. If a process takes 10 FTEs and can be 80% automated with AI, you're looking at 8 FTE reduction, which at $80k all-in cost per FTE is $640k of annual savings. For a $50m EBITDA business, that's 130 basis points. But labour automation is also the slowest to realise because it requires change management and often has employment considerations.
Process optimisation is faster. Reducing the time to process a claim from 2 hours to 20 minutes, or reducing the time to schedule clinical staff from 4 hours to 15 minutes, doesn't require headcount reduction—it just means your existing team can handle more volume or move to higher-value work. For a 20% productivity improvement across a cost centre, you're looking at 50–100 bps of EBITDA improvement in 100 days.
Resource allocation is the most underrated. AI can optimise which assets are deployed to which tasks. In logistics, it's route optimisation and vehicle utilisation. In healthcare, it's OR scheduling and bed management. In hospitality, it's housekeeping allocation and maintenance prioritisation. A 5–10% improvement in asset utilisation is 100–200 bps of EBITDA improvement.
Risk and Compliance (50–150 bps EBITDA impact)
This is where PE firms often miss the opportunity. Compliance, fraud detection, and risk management are typically viewed as cost centres. But they're also where AI can unlock value by reducing losses and avoiding penalties.
For a financial services firm, an AI system that flags suspicious transactions with 95% accuracy can reduce fraud losses by 30–40% and reduce the false positive rate that clogs your compliance team. For an insurance company, an AI system that assesses claim legitimacy can reduce claims leakage by 15–20%. For a healthcare system, an AI system that flags coding errors before submission can reduce audit losses by 25–30%.
The key is measuring the impact accurately. Many firms assume compliance AI will reduce headcount. It won't. It will reduce losses and allow your existing team to handle higher volumes or more complex cases.
Customer Experience (100–200 bps EBITDA impact through retention and NPS)
This is the hardest to quantify but often the most valuable. AI can improve customer experience through faster response times, more personalised interactions, and proactive problem-solving.
For a SaaS business, an AI chatbot that resolves 60% of support tickets without human intervention reduces support costs by 40% and improves customer satisfaction because customers get answers in seconds instead of hours. For a hotel group, an AI concierge that personalises recommendations, handles room requests, and resolves complaints improves guest satisfaction and increases repeat bookings. For a healthcare system, an AI workflow that answers patient questions, schedules appointments, and sends appointment reminders improves patient experience and reduces no-shows.
The impact flows through to retention, which flows through to LTV, which flows through to valuation multiples. A 5–10% improvement in retention is worth 200–400 bps of EBITDA improvement over a 5-year hold.
Architecture Patterns That Work at Scale
The difference between an AI system that works in a pilot and one that works at scale is architecture. Here are the patterns that deliver production-ready AI in 100 days.
Pattern 1: Agentic Workflows with Deterministic Validation
This is the fastest path to production. You deploy an AI agent (Claude Opus 4, GPT-4 Turbo, or Gemini 2.0) that performs a specific workflow: read input, gather context, make a decision, take an action. But you wrap that with a deterministic validation layer that catches errors, flags edge cases, and ensures consistency.
Example: An AI agent that processes insurance claims. The agent reads the claim, retrieves policy details, checks for fraud indicators, estimates payout, and recommends approval or denial. But before the recommendation goes to the claims processor, a deterministic rule engine checks: Is the payout within policy limits? Is there a fraud flag? Is there a prior claim? If anything is off, the system flags it for human review instead of making the decision autonomously.
This pattern is fast because you don't need to train a model. You don't need a massive labelled dataset. You just need to integrate the AI with your existing systems and define the validation rules. It's safe because humans stay in the loop for edge cases. It's scalable because the AI can handle millions of routine decisions while your team focuses on the complex ones.
Pattern 2: RAG (Retrieval-Augmented Generation) for Domain Knowledge
Your organisation has domain knowledge scattered across documents, policies, procedures, and the heads of experienced employees. RAG lets you encode that knowledge so the AI can access it.
Instead of fine-tuning a model on your specific domain (which requires thousands of labelled examples and 8–12 weeks), you build a retrieval system that searches your knowledge base and feeds relevant context to the AI. The AI then answers questions or makes decisions based on that context.
Example: A compliance officer asks the AI agent, "Is this customer transaction compliant with AML regulations?" The system retrieves relevant AML policies, recent regulatory updates, and similar historical cases, feeds them to the AI, and the AI makes a recommendation. The AI isn't hallucinating or guessing—it's reasoning based on your actual policies and precedents.
RAG is fast to build (2–4 weeks), accurate (because it's grounded in your actual knowledge), and easy to update (you just add new documents to the retrieval system).
Pattern 3: Human-in-the-Loop with Feedback Loops
No AI system is perfect. But every decision the AI makes (especially the ones humans review) is a training signal. You capture that signal and use it to improve the system.
Example: An AI system recommends which customers should be contacted for upsell. A salesperson reviews the recommendation, decides whether to act on it, and whether it resulted in a deal. You capture that feedback and use it to retrain the recommendation engine. Over time, the system gets better at identifying high-value opportunities.
This pattern is powerful because it means your AI system improves with every decision. You don't need to wait for a scheduled model retraining. You're continuously learning from real-world feedback.
Pattern 4: Modular Integration with Existing Systems
Your company has existing systems: CRM, ERP, accounting, HR, etc. You don't want to rip and replace. You want to build AI on top.
This means using APIs to pull data from existing systems, using webhooks to trigger AI workflows, and using standard data formats to integrate results back into existing systems. The AI sits on top of your existing stack, not underneath it.
Example: Your CRM has customer data. Your accounting system has transaction data. Your AI system pulls data from both, makes a decision (which customers to contact, what to offer, when), and pushes the result back to your CRM as a task or recommendation. The sales team still uses the CRM. The AI is just making the CRM smarter.
This pattern is fast because you're not rebuilding systems. It's safe because you're not touching existing workflows. It's scalable because you can add more AI workflows without touching the underlying infrastructure.
Governance: Speed Without Chaos
This is where most organisations stumble. They want to move fast but also want perfect governance. You can't have both. You have to choose what to govern and what to move fast on.
The Governance Hierarchy
Tier 1: High-stakes decisions (financial, legal, safety, healthcare)
These require human review, audit trails, and governance approval. An AI system can assist (gathering information, making recommendations), but the final decision stays with a human. Deployment timeline: 60–100 days because you need to define the governance model, get approvals, and implement audit trails.
Tier 2: Medium-stakes decisions (customer experience, operational efficiency)
These can be AI-driven with human review for edge cases or exceptions. An AI system makes the decision autonomously for 80–90% of cases and flags the rest for human review. Deployment timeline: 40–60 days because you need to define the validation rules and flag criteria.
Tier 3: Low-stakes decisions (recommendations, optimisations, predictions)
These can be fully autonomous. An AI system makes the decision, logs it, and humans review the results periodically. Deployment timeline: 20–30 days because you just need to integrate and monitor.
The Governance Operating Model
Who owns the AI system? Not IT. Not data science. A business owner. Someone who understands the business impact, owns the P&L, and can make fast decisions about scope, priorities, and trade-offs.
How do you handle model updates? Through a sprint-based process. Every 2 weeks, you review performance, identify improvements, and deploy updates. This isn't a big annual retraining. It's continuous iteration.
What's the escalation path? If the AI system encounters an edge case, makes an error, or encounters a regulatory issue, who decides what to do? Define this in advance. It should be fast (24–48 hours) because delays kill momentum.
The 100-Day Sequencing: Stacking Wins
Timing matters. You want to sequence initiatives so that early wins build momentum, early learnings improve later initiatives, and you're not overloading the organisation.
Days 1–20: Diagnostic and Quick Wins
Identify the top 3–5 value levers. Start with the one that's easiest to measure and has the fastest payback. This is your momentum builder. You want to see results by Day 60.
Examples of quick wins:
- Deploying an AI chatbot that handles 40% of support tickets (2–3 week deployment, 20% cost reduction)
- Implementing dynamic pricing in a hotel or airline (2–3 week deployment, 5–10% revenue uplift)
- Building an AI workflow that flags compliance exceptions (2–3 week deployment, 10–15% audit efficiency improvement)
Days 20–40: Architecture and Data
While the quick win is being piloted, you're locking in the architecture for initiatives 2 and 3. You're also assessing data quality and building integration pipelines. This work runs in parallel with the quick win pilot.
Days 40–60: Pilot Initiatives 2 and 3
If the quick win is working, you pilot the next two initiatives. You're now running three parallel workstreams: scaling the quick win, piloting initiative 2, and piloting initiative 3.
Days 60–80: Scale and Measure
You're scaling the quick win to the full organisation. You're scaling initiatives 2 and 3 if they're working. You're measuring everything. By Day 80, you should have 3–5 AI systems in production, delivering measurable value.
Days 80–100: Optimise and Plan for Year 2
You're optimising the systems you've deployed. You're also planning the next wave of AI initiatives for Year 2. You've proven the capability. Now you're scaling it.
Implementation: How Brightlume Executes This Playbook
This isn't theoretical. This is how Brightlume executes the 100-day AI plan for PE-backed companies.
Brightlume's approach is built on three principles: ship production AI in 90 days, use frontier models and agentic architecture, and measure ruthlessly.
The diagnostic phase is intensive but short. Brightlume's team spends 2–3 weeks embedded with the operating partner, CFO, and business leaders. They're not building a 100-page strategy document. They're identifying the 3–5 highest-impact opportunities, assessing technical readiness, and locking in the architecture.
The deployment is parallel. While the quick win is being built (2–3 weeks), the architecture and data integration for initiatives 2 and 3 are being planned. This means by Week 4, you have one system in pilot and two more systems in development.
The measurement is continuous. Brightlume builds dashboards that track every metric in real time. By Day 60, you know exactly what's working, what's not, and what to double down on.
The governance is pragmatic. Brightlume doesn't build perfect systems. They build systems that are good enough to move the needle, safe enough to scale, and fast enough to hit the 100-day timeline. They define the governance model upfront, lock in the approval process, and then move.
The result: Brightlume's clients deploy production AI systems in 90 days with an 85%+ pilot-to-production rate. That's not a pilot success rate. That's a production deployment success rate. Most of their clients see measurable value—cost reduction, revenue uplift, or operational improvement—within 100 days of close.
Real-World Example: Financial Services Firm
Let's walk through a real example. A PE firm acquires a mid-market financial services company with $50m EBITDA. The company has 200 FTEs, processes $2b in transactions annually, and has a compliance team of 8 people.
Days 1–20: Diagnostic
The operating partner and CFO identify the top value levers:
- Compliance review automation (reduce review time from 2 hours to 20 minutes per transaction)
- Fraud detection (reduce fraud losses by 30%)
- Customer onboarding (reduce onboarding time from 5 days to 1 day)
- Claims processing (reduce processing time from 1 hour to 10 minutes)
The diagnostic also reveals: the company has a data warehouse with 5 years of transaction history, a modern API-first architecture, and a compliance team that's drowning in manual review work.
Days 20–35: Architecture
The engineering team locks in the architecture:
- Compliance review: AI agent that reads transaction details, retrieves relevant policies and precedents, flags exceptions, and recommends approval or rejection. Deterministic validation layer checks for edge cases.
- Fraud detection: AI system that scores transactions for fraud risk based on historical patterns and real-time signals. Flags high-risk transactions for manual review.
- Customer onboarding: AI agent that collects customer information, validates against regulatory requirements, and flags missing or inconsistent data.
- Claims processing: AI agent that reads claim details, checks policy coverage, estimates payout, and recommends approval or denial.
Days 20–50: Data Integration
The team builds APIs to pull transaction data, customer data, and policy data from existing systems. They build a retrieval system that indexes historical transactions, policies, and regulatory updates. They define the validation rules for each AI system.
Days 40–60: Pilot Compliance Review
The compliance team pilots the AI system on a sample of 500 transactions. The system reviews each transaction, flags exceptions, and makes a recommendation. The compliance team reviews the flagged transactions and the AI recommendations.
Result: The AI system correctly identifies 95% of compliant transactions and flags 100% of non-compliant transactions. The compliance team spends 10 minutes on the flagged transactions instead of 2 hours on each transaction. Time savings: 90%.
Days 60–80: Scale and Pilot Fraud Detection
The compliance team scales the AI system to all 10,000 transactions per month. They also pilot the fraud detection system on a sample of transactions.
Result: The fraud detection system identifies 92% of fraudulent transactions and 8% false positive rate. The team investigates the flagged transactions. True fraud detection rate: 88%. False positive rate: 8%. The system catches an average of 2–3 fraudulent transactions per week that would have slipped through manual review.
Days 80–100: Scale and Measure
Both systems are scaled to production. The compliance team is now handling 10,000 transactions per month with 2 people instead of 8. The fraud detection system is catching 50–60 fraudulent transactions per month, saving the company $200k–$300k per month in fraud losses.
P&L impact:
- Compliance automation: 6 FTE reduction × $80k = $480k annual savings (100 bps EBITDA)
- Fraud detection: $200k–$300k monthly savings = $2.4m–$3.6m annual savings (50–70 bps EBITDA)
- Customer onboarding (piloted in parallel): 30% faster onboarding = 10% increase in customer throughput = $2m–$3m annual revenue uplift (40–60 bps EBITDA)
Total 100-day impact: 190–230 bps of EBITDA improvement. At a 10x EBITDA multiple, that's a 19–23% value increase from a 100-day AI deployment.
Avoiding the Pilot Trap
The most common failure mode for enterprise AI is the pilot trap: you pilot for 6 months, prove the concept, and then spend another 6 months trying to scale. By the time you reach production, you've lost momentum, the business has moved on, and the project dies.
The 100-day approach avoids this by treating pilots as mini-productions. You pilot with a specific, measurable outcome. You pilot with real data and real users. You pilot with the same governance and monitoring as production. If it works, you scale immediately. If it doesn't work, you kill it and move to the next opportunity.
The key is defining success upfront. Before you pilot, you should know: What's the success metric? What's the threshold? How long do we pilot? If we hit the threshold, do we scale immediately or do we need additional approval?
For most of Brightlume's clients, the answer is: if the pilot hits the KPI, we scale in the next sprint. No additional approvals. No additional pilots. Just scale and measure.
The Operating Partner's Role
The operating partner's job in the 100-day AI plan is not to build the AI systems. It's to:
- Set the strategic direction: Which value levers are we pursuing? What's the target EBITDA improvement?
- Unblock the team: Remove obstacles. Get access to data. Get executive alignment. Get budget approval.
- Hold the timeline: Keep the project moving. No six-month discovery phases. No endless pilots. 100 days.
- Measure ruthlessly: Track every metric. Know exactly what's working and what's not. Make fast decisions based on data.
- Build the operating model: Define who owns the AI system. Define the governance model. Define the escalation path. Make sure the system is sustainable beyond the 100 days.
The operating partner is the CEO of the AI transformation, not the CTO. Your job is to make sure it happens, not to build it.
The Next 400 Days: Scaling Beyond 100
The 100-day plan is not the end. It's the beginning. Once you've proven the capability and built the operating model, you're ready to scale.
Days 100–200: You're expanding the scope of your initial AI systems (more workflows, more users, more transactions) and piloting the next wave of AI initiatives. You're also building the internal capability to own and operate these systems.
Days 200–400: You're scaling to the full organisation. You're building the data infrastructure to support more complex AI systems. You're also exploring more advanced use cases: predictive analytics, generative workflows, autonomous decision-making.
By Day 400 (roughly 13 months post-close), a well-executed AI transformation should have delivered 300–500 bps of EBITDA improvement, built internal AI capability, and positioned the company for continued improvement through the hold period.
This is where you see the true value of the 100-day approach. You're not just capturing one-time improvements. You're building a capability that compounds through the hold period. Every quarter, you're adding new AI workflows, optimising existing ones, and capturing incremental value.
Conclusion: The 100-Day Decision
The choice is simple. You can treat AI as a 2025 initiative and hope you catch up to competitors who moved in 2024. Or you can treat it as a Day 1 value lever and capture 200–500 bps of EBITDA improvement in the first 100 days.
The 100-day AI plan is not a nice-to-have. It's a must-have for PE firms that want to outperform. It's the difference between a 1.5x return and a 2.5x return over a 5-year hold.
The playbook is proven. The architecture patterns work. The governance models are battle-tested. The only question is: Are you going to move fast enough to capture the value?
If you're a PE operating partner, a CFO, or a transformation lead tasked with driving AI value creation post-close, the time to start is now. Not in 100 days. Not in 30 days. Today.
For PE firms ready to move on AI immediately post-close, Brightlume specialises in shipping production-ready AI systems in 90 days. The firm's approach is purpose-built for operating partners: diagnostic in 2–3 weeks, architecture locked in by Week 4, first system in pilot by Week 5, and production deployment by Week 12. With an 85%+ pilot-to-production rate and measurable value delivered in the first 100 days, Brightlume helps PE firms capture the AI value creation lever that most competitors are still debating.
The 100-day AI plan is not about being first. It's about moving faster than the competition and compounding that advantage through the hold period. The firms that execute this playbook will see 200–500 bps of EBITDA improvement. The firms that wait will see their multiples compress as competitors capture the value first.
The choice is yours. But the clock is already running.