Build an AI plan for the board in 60 days

A person at a meeting-room table reviewing a single printed page of ranked items with a pen in hand
TL;DR

A credible sixty-day AI plan beats a comprehensive six-month one. The board needs three to five opportunities ranked by business outcome and feasibility, honest timelines, and a clear first move, not every use case scoped. The surprise in the research is that back-office automation tends to pay more than the sales and marketing pilots that attract the most funding, so the prioritising work is choosing the boring answer and being willing to put it in front of the board.

Key takeaways

- A sixty-day plan that ranks three to five opportunities by business outcome, feasibility, time-to-value and investment is more defensible than a comprehensive plan that scopes every use case the board can imagine. - MIT's research finds back-office automation produces the highest returns while sales and marketing pilots show the lowest return despite attracting the most funding, so the highest-value work is often the least exciting to demo. - Avoid AI theatre, the flashy demo with no measurable outcome. Addepar's test is to ask whether the initiative would still matter if it did not use AI at all. - Use a dual-ROI frame, trending ROI as early indicators alongside realised ROI as financial outcomes, so you can show movement without claiming returns that have not arrived. - Set the timeline expectation now, in the plan, not later under pressure. Meaningful ROI commonly takes twelve to twenty-four months, and for a business this size a vendor-led build tends to land better than an internal one.

The listening tour is done. The current-state read is written and honest. Now there are fifteen possible use cases on a whiteboard, a board date in the diary, and a quiet pressure to make the plan look ambitious. The temptation is obvious. Pick the exciting ones, the AI sales assistant, the marketing engine, the thing that will demo well in the room.

The harder instinct is the better one. Ask which of the fifteen can actually be made to pay, keep three to five, and be willing to put the boring answer in front of the board. A credible sixty-day plan beats a comprehensive six-month one, because the board does not need every use case scoped. It needs to see a short, ranked list with honest timelines and a clear first move.

What is the choice you are actually making here?

The real choice is how to rank a long list down to three to five opportunities the business can fund and deliver, and which of those to put first. Keeping all fifteen on the table is the failure mode, not the goal. You are choosing between the work that demos well and the work that pays, and those are rarely the same thing. Get the ranking right and the rest of the plan writes itself.

The method is a value-feasibility score. Take each opportunity and rate it on four things, the business outcome it produces, how feasible it is with the data and skills you have, how long it takes to show value, and what it costs to stand up. Keep the three to five that score highest and park the rest on a visible backlog. A long list of fifteen reads as indecision. A ranked five reads as a decision the board can act on.

The discipline is in what you drop, not what you keep. Every parked use case stays visible on the backlog with a one-line reason it is not yet first, so the board can see you considered it and chose your order deliberately. That single page does more work than a forty-slide deck. It tells the room you have a method, you have applied it, and the five in front of them are the ones the business can actually fund and deliver inside the horizon you are about to set out.

When is the unglamorous back-office work the right first move?

Back-office automation is the right first move more often than the funding patterns suggest. MIT’s research finds that back-office automation produces the highest returns, while sales and marketing pilots show the lowest return despite attracting the most funding. The reason is mechanical. Back-office work tends to be repetitive, rule-bound and measurable, which is precisely where current AI is strong and where outcomes are easy to count.

Think document processing, invoice handling, customer-service triage, reconciliation, the work nobody puts in a board deck because it is dull. These tasks have a clear before and after. You can show hours saved, error rates down, throughput up, in numbers the finance director will accept. The front-office pilot, the AI that writes campaigns or qualifies leads, is harder to attribute, slower to prove, and more likely to stall in what the research calls the pilot-to-scale valley. The dull answer is the one worth funding first because it is the one you can defend.

There is a second reason to start here, and it is about trust as much as return. A back-office tool that reliably saves the operations team three hours a week builds the internal confidence the next, bigger move depends on. Frame the work as freeing people for higher-value tasks rather than replacing them, and adoption follows. Lead instead with a front-office pilot that overpromises and underdelivers, and you spend that goodwill before you have earned it. The boring win is also the safe place to learn how your business actually adopts AI.

When do the front-office and bigger pilots earn their place?

Front-office and larger pilots earn their place once the back-office wins have built credibility and the data is good enough to support them. They are not wrong, they are simply later. A sales or marketing pilot needs clean customer data, a measurable baseline, and a team that trusts the tool, and those conditions usually come after the foundation work is done, not before. Sequenced second, the same pilot has a far better chance.

This is where the phased plan does its work. A foundation phase across the first six months banks the back-office wins and gets the data and governance into shape. An expansion phase, roughly six to eighteen months, is where the front-office pilots and the wider rollout belong, on top of foundations that already pay. A longer horizon beyond that carries the more ambitious bets. The phasing is what lets you say yes to the exciting pilot without putting it first and watching it fail.

What does it cost to get the prioritising wrong?

Getting it wrong costs the budget and your credibility in the same move. Lead with the flashy front-office pilot, watch it stall, and you have spent real money on AI theatre, the demo that impressed the board and then produced no measurable outcome. MIT’s research is blunt, around 95 per cent of AI pilots fail to show measurable P&L impact, and the person handed the mandate carries the blame when they do.

The honest defence is a measurement frame that does not overclaim. Use a dual-ROI approach, trending ROI as the early indicators in the first few months alongside realised ROI as the financial outcomes arrive later. That lets you report genuine movement without pretending returns have landed before they have. Pair it with a buy-versus-build steer, because for a business this size a vendor-led build tends to succeed where an internal one struggles, and an over-ambitious internal build is one of the more expensive ways to get this wrong.

What should you ask before you commit the plan to the board?

Ask three questions of every opportunity that survived the cut, and one of the plan as a whole. For each opportunity, would this still matter if it did not use AI, can we measure the outcome in numbers the board accepts, and can we deliver it with the data and skills we have. Addepar’s test, whether the initiative would still matter without AI, kills off the theatre before it reaches the room.

The question for the whole plan is about time. Meaningful ROI commonly takes twelve to twenty-four months, and many boards expect change far sooner, so set that expectation now, in the document, not later under pressure when a deadline slips. A plan that names the back-office first moves, ranks three to five opportunities with honest timelines, and shows a phased shape with the buy-versus-build call already made is one you can stand behind in the room. If you want a second pair of eyes on the ranking before it goes to the board, book a conversation.

Sources

- MIT, State of AI in Business (2025, Project NANDA). Cited for the finding that around 95 per cent of AI pilots fail to show measurable P&L impact, the backdrop for why ruthless prioritisation matters. https://sranalytics.io/blog/why-95-of-ai-projects-fail/ - MIT research via Schellman (2025). AI implementation failures in real-world deployments. Cited for the pattern that vendor-led projects succeed far more often than internal builds for businesses of this size. https://www.schellman.com/blog/ai-services/ai-implementation-failures-in-real-world-deployments - BridgeView (2025). AI readiness, the five pillars. Cited for the buy-versus-build steer and the success-rate gap between vendor-led and internally built AI projects. https://www.bridgeviewit.com/ai-readiness/ - Addepar (2025). Questions executives should ask before adopting AI. Cited for the AI theatre warning and the test of whether an initiative would still matter without AI. https://addepar.com/blog/questions-executives-should-ask-before-adopting-ai - Propeller (2025). Measuring AI ROI, how to build an AI strategy that captures business value. Cited for the dual-ROI frame, trending ROI alongside realised ROI across time horizons. https://propeller.com/blog/measuring-ai-roi-how-to-build-an-ai-strategy-that-captures-business-value - LogixGuru (2025). The board wants an AI strategy by Tuesday, a CIO's survival guide. Cited for the three-to-five opportunity shape, the foundation and expansion phasing, and the board-deadline framing. https://www.logixguru.com/post/the-board-wants-an-ai-strategy-by-tuesday-a-cios-survival-guide - ESG Dive (2025). Executives fear job loss due to AI. Cited for the twelve-to-twenty-four-month timeline to meaningful ROI against boards expecting immediate change. https://www.esgdive.com/news/execs-fear-job-loss-due-to-AI/818075/ - ScaledAgile (2025). The board questions every CEO should be able to answer about AI. Cited for the board's core questions on alignment, ROI measurement, risk and capability. https://scaledagile.com/blog/the-board-questions-every-ceo-should-be-able-to-answer-about-ai/ - Korn Ferry (2025). Six signs leaders lack AI readiness and how to fix it. Cited for the readiness paradox, where strong operators are handed AI leadership without the AI-specific competencies the task needs. https://www.kornferry.com/insights/featured-topics/gen-ai-in-the-workplace-articles/6-signs-leaders-lack-ai-readiness-and-how-to-fix-it

Frequently asked questions

How many AI opportunities should the plan put in front of the board?

Three to five, ranked by business outcome and feasibility, with the rest parked on a visible backlog. A long list reads as indecision and invites the board to add more. Each opportunity should state the business outcome, feasibility, time-to-value and the investment required, so the board is choosing between scoped options rather than reacting to a wishlist. Resist the pressure to keep more than five.

Why would back-office automation beat a sales or marketing AI pilot?

MIT's research finds back-office automation produces the highest returns while sales and marketing pilots show the lowest return despite attracting the most funding. Back-office work tends to be repetitive, rule-bound and measurable, which is exactly where current AI is strong. The flashy front-office pilot demos well and disappoints on the P&L, so the unglamorous answer is usually the one worth funding first.

What timeline should I promise the board for AI returns?

State plainly that meaningful ROI commonly takes twelve to twenty-four months, against a board that often expects near-immediate change. Setting that expectation in the plan is far easier than defending a missed three-month deadline later. Use a dual-ROI frame so you can report trending ROI, the early indicators, in the first few months, and realised financial ROI as it arrives over the longer horizon.

This post is general information and education only, not legal, regulatory, financial, or other professional advice. Regulations evolve, fee benchmarks shift, and every situation is different, so please take qualified professional advice before acting on anything you read here. See the Terms of Use for the full position.

Ready to talk it through?

Book a free 30 minute conversation. No pitch, no pressure, just a useful chat about where AI fits in your business.

Book a conversation

Related reading

If any of this sounds familiar, let's talk.

The next step is a conversation. No pitch, no pressure. Just an honest discussion about where you are and whether I can help.

Book a conversation