Insights

How do I show my board that AI is improving EBITDA?

By Maya Moufarek, founder of Forward Operating Partners · 11 November 2026

Six in ten companies using AI have not yet seen company-wide EBIT impact from it, according to McKinsey's latest survey, and 82% of company directors rate the information linking AI to business performance as fair, poor or missing (McKinsey, 2026) (PwC, 2026).

Most advice on closing that gap assumes a profitable company with a finance team, where AI savings show up as costs cut. Many scale-ups don't fit that picture. They are still investing for growth, and AI shows up in their numbers as hires they never had to make.

How do you show a board that AI is improving EBITDA?

To show a board that AI is improving EBITDA, report AI in the number that board already reads: an EBITDA bridge with each AI item net of its cost for a profitable, often PE-backed company, and revenue or output per head against plan for a VC-backed scale-up still investing for growth.

Pick the number before the work starts, with the board, and take the baseline then. A private-equity owner values the business on earnings, so the bridge from last year's EBITDA to this year's, with AI items shown as separate bars and their running costs taken off, is the format that board expects. A venture board steering a loss-making company reads efficiency per person instead.

How do you show AI's impact if your scale-up isn't profitable yet?

A scale-up that isn't profitable yet should show AI's impact as revenue or output per head against the plan the board approved, with the AI cost included, because ARR per employee is the efficiency number its investors already compare.

At $20 million to $50 million of ARR, top-quartile ARR per employee rose 42% in a year to $350,000 (High Alpha). Across ICONIQ's wider software set, median ARR per employee rose from $182,000 to $237,000 over five years while operating expense per employee stayed flat at $220,000 to $230,000 (SaaStr on ICONIQ). Both benchmarks come from venture investors.

What AI metrics do VC investors look at?

VC investors look at revenue per employee and go-to-market output per head when they judge a scale-up's AI story: in ICONIQ's 2026 data, companies that adopted AI heavily generated about $640,000 of net new ARR per go-to-market employee, against $370,000 for the rest (ICONIQ, 2026).

ICONIQ surveyed more than 150 go-to-market leaders and is itself an investor, so treat the gap as a benchmark to aim at. Its most AI-forward companies ran go-to-market teams 20% to 30% leaner.

How do you show AI savings when you haven't cut headcount?

Show AI savings without headcount cuts by comparing the hiring plan with actual headcount and output: in a growing company, AI savings usually arrive as roles that were planned and never needed, and that gap, net of the AI cost, is the number to put in front of the board.

Heavy AI adopters are still hiring. Ramp and Revelio Labs linked spending data to workforce records for 21,559 US firms and found that companies investing heavily in AI grew headcount 10.2% over the two years after adoption (Ramp, Summer 2026). That link is correlational. Scale-ups expect to keep hiring too: 73% of those Deloitte surveyed across Europe, the Middle East and Africa expect headcount to grow, and in the UK 69% expect growth while 24% plan a hiring freeze (Deloitte, 2026).

To take an illustrative example: if the plan approved in January had four more people in support by October, and support is handling the volume with the same team plus an agent, those four salaries minus the agent's cost are the saving. Hours freed count once they turn into one of those things: a hire not made, or more output from the same people.

What should an AI board slide include?

An AI board slide should show one number agreed before the work started, its baseline and latest value, the full AI cost including the people who run it, realised results kept apart from capacity freed, and the initiatives that were stopped.

I measure the cost side as revenue over people plus tokens, so the agents, the tools and the time of the people who build and check them all sit in the same sum. List the initiatives you stopped as well, with the reason.

Why do boards distrust AI numbers?

Boards distrust AI numbers because most companies report activity rather than a measured financial result: 82% of directors rate the information they get linking AI to business performance as fair, poor or missing.

Private companies find it hard as well. 59% say they struggle to communicate AI's value to stakeholders (KPMG, 2026), and only 13% of SaaS companies track specific KPIs for AI impact (High Alpha). McKinsey finds that chatbots, copilots and summarisers help people work faster but "rarely change a P&L", while end-to-end changes to a workflow do (McKinsey, 2026). When I surveyed founders across one VC firm's portfolio in April, measuring whether AI was working came last on their list of priorities, among the roughly 10 who replied.

As CMO, I took Pharmacy2U through a private-equity exit. At Forward Operating Partners we set the measure before the work starts and report it on a line the board already reviews, before and after, with the cost included.

What will your next board pack say AI did to the numbers?