How to write an AI and data value creation plan for a board pack

Most PE-backed businesses have a slide about AI in their board pack. It says something like "exploring opportunities in AI" or "AI strategy under development." It is, in effect, a placeholder. It tells the board nothing about where value will come from, on what timeline, or what it will cost to get there.

The operating partners and investment professionals reading this know the problem. The business knows it needs to do something with AI and data. The management team is not equipped to write the plan. External consultants produce decks that are long on vision and short on financial specificity. And the clock is ticking - whether you are 18 months into a hold or starting to prepare for exit.

This article will show you how to structure an AI and data value creation plan that belongs in a board pack. One that is specific, financially grounded and defensible under scrutiny. Not a strategy document. A value creation plan.

What a value creation plan is not

Before the structure, the distinction that matters most.

A data strategy is an IT document. It describes architecture, governance and tooling. It belongs in a technical workstream, not a board pack. An AI roadmap is a project plan. It describes what will be built and when. Neither of these is a value creation plan.

A value creation plan makes a specific claim: if we invest £X in capability Y, we will realise £Z in EBITDA impact by date D. Everything else is context in service of that claim.

The mistake most management teams make when asked to produce this document is to answer the wrong question. They describe what they will do rather than what it will be worth. The result is a document that demonstrates activity but cannot be challenged, validated or used to hold anyone accountable.

If your current AI and data section cannot be interrogated with the same rigour as a capex proposal, it is not a value creation plan. It is a holding position dressed up as strategy.

Start with the value, not the technology

The plan should be built backwards from commercial outcomes - not forwards from technical capability.

That means beginning with the levers. For most mid-market businesses, AI and data value creation concentrates in four areas: revenue growth through better customer targeting or pricing decisions, gross margin improvement through operational efficiency, cost reduction in back-office or support functions, and risk reduction through better forecasting or compliance automation.

Take a distribution business with £600m revenue. The CFO suspects they are leaving margin on the table through inconsistent pricing across their sales team. They have the transactional data - 18 months of orders, customer segments, product mix. The question is not "should we use AI for pricing?" The question is: "What is the revenue per transaction variance across equivalent customer cohorts, and what does closing 30% of that gap mean for EBITDA at our current volume?" That is the starting point for a value creation plan.

For each lever you identify, you need three things before it goes in the pack: a quantified baseline, a realistic improvement assumption and a time-to-value estimate. If you cannot produce all three, the lever is not ready for the board pack. It belongs in a pre-work diagnostic.

The four sections every plan needs

Once you have identified two or three financially credible levers, the plan itself needs to be structured around four sections.

1. Current state assessment

This is not a technology audit. It is a commercial data audit. What decisions are currently being made on gut or lagging data? Where does the business lack visibility that a competitor with better analytics would exploit? A one-page summary of three to five named decision-failure points is enough. For example: "Churn is only identified after it has happened because the CRM is not integrated with product usage data. The business is losing an estimated £4m in renewals annually that earlier signals would allow them to recover."

2. Prioritised initiatives with financial cases

List no more than four initiatives. For each, provide: what the initiative is, what data or AI capability it requires, the cost to build or deploy, the expected EBITDA impact, the timeline to first value, and the owner. A table format is appropriate here. If an initiative cannot be costed and timed, it should not be on this list.

3. Capability and dependency assessment

What does the business need to have in place for the initiatives to succeed? This is where you address data quality, integration requirements and team capability honestly. Boards dislike surprises. If the pricing initiative depends on clean CRM data that does not currently exist, say so - and cost the remediation. Hiding dependencies is how value creation plans fail.

4. Governance and review cadence

Who owns delivery? How often does the board see progress against the financial case? What are the milestones at which investment continues or stops? This section is short - half a page - but it signals that the plan is being run with the same commercial discipline as any other growth initiative.

Making the numbers credible

The biggest credibility problem with AI and data value creation plans is that the financial cases are untethered. Someone has extrapolated a benefit from an industry benchmark or a vendor case study with no connection to the actual business.

Boards - and certainly experienced investment professionals - will see through this. The number does not need to be large to be credible. It needs to be derived.

A useful test: can you show the arithmetic? If you are claiming a pricing initiative will add £2.1m EBITDA, the board should be able to see: current average transaction value, variance across customer segments, the percentage of transactions in scope, the assumed improvement rate, and the revenue and margin impact at current volume. Five lines. If you cannot show those five lines, the number is not credible.

One practical note on assumptions: use management's own numbers wherever possible. If the commercial director has already said that 15% of accounts are underpriced relative to comparable accounts, use that figure. It is harder to challenge an assumption that came from inside the business.

For businesses using or evaluating business intelligence tooling, this is where a platform like Quantsole adds direct value in the pack preparation process - not by generating the strategy, but by pulling the actual variance analysis from live transactional data rather than estimates.

Calibrating ambition to hold period

The plan should reflect where you are in the hold. This sounds obvious. It is frequently ignored.

If you are 12 months from a planned exit, a multi-year data transformation programme is the wrong plan. The board pack should show quick, demonstrable wins that improve EBITDA in the near term and create a credible AI and data story for the incoming buyer. Think pricing analytics, automated reporting, churn prediction with a direct commercial intervention attached.

If you are 12 months into a five-year hold, you have more room to invest in foundational capability - data infrastructure, team building, more complex model development - provided the near-term initiatives are funding the investment through their own returns.

The plan should contain a simple timeline view that maps initiatives against the expected hold period and shows when EBITDA impact accrues. This is the clearest signal that the plan was written by people who understand the investment, not by people who understand technology.

The cost of a placeholder strategy

Every quarter that the AI and data section of a board pack says "under development" is a quarter of value creation lost. More importantly, it is a quarter in which a buyer's diligence team - or a competitor - is building a picture of a business that has not kept pace.

AI and data capability is increasingly a valuation question, not just an operational one. According to analysis from McKinsey Global Institute, companies that are leaders in data and analytics are twice as likely to be in the top quartile of financial performance in their industries. The gap between businesses that have embedded these capabilities and those that are still exploring is widening.

If your portfolio company's board pack contains a placeholder, the first step is a structured diagnostic - typically a four to six week engagement that produces the current state assessment, identifies two or three financially grounded initiatives and builds the outline case for the full plan.

Rodan runs these diagnostics as paid, fixed-scope engagements. They are designed specifically for PE-backed businesses that need commercial rigour, not another technology roadmap.

If you want to talk through what that would look like for a specific portfolio company, book a diagnostic conversation with the Rodan team.