How to Find the Highest-Value AI Opportunities Across Your Portfolio

Operating partners are sitting on a list of twenty or more possible AI use cases across the portfolio, with no reliable way to rank them.

By some estimates cited by RAND, more than 80% of AI projects fail, roughly twice the rate of IT projects that don't involve AI. The most common causes RAND identified show up before anything gets built: leaders misunderstand the problem AI is supposed to solve, or the company doesn't have the data to support it. In a portfolio, that usually means the wrong use case was chosen to begin with.

This post lays out a repeatable way to map, score, and sequence AI opportunities across a portfolio, so the highest-value bets get resourced first.

Not All AI Opportunities Are Equal

McKinsey studied AI maturity across 471 PE-backed companies in 31 industries and found four distinct levels. At level one, AI is absent or limited to isolated pilots. At level two, it's embedded in daily workflows for speed and better decisions. At level three, it's built into the product, and at level four, it creates entirely new revenue streams.

The valuation data shows where the difference shows up. Level one and level two companies trade at nearly the same median revenue multiple, 13x and 14x, which suggests the market doesn't reward productivity gains on their own. The multiple rises once AI becomes part of what the company sells: 20x at level three and 31x at level four, where software businesses reach 33x. Level four companies also post 52% higher revenue per employee than level three.

That doesn't make productivity work a waste. McKinsey recommends running level one and two improvements across the portfolio, since they apply to almost every company and pay back quickly, and pursuing levels three and four selectively. The distinction matters when you rank opportunities. Productivity wins make a company more efficient, but they rarely move its multiple. Companies with strong proprietary data and a realistic path to putting AI into the product deserve a separate, longer look, because that's where the valuation upside sits.

Map Readiness Before You Score Opportunities

Before ranking opportunities, you need to know which portfolio companies can actually execute on them. AlixPartners recommends auditing every company across five dimensions: data infrastructure, technology stack, leadership alignment, talent and capabilities, and risk and compliance. Score each on a simple three-point scale and you get a heat map across the whole portfolio, not just a gut feeling about which company seems "ready."

Skipping this step is how firms end up chasing an ambitious use case in a company that doesn't have the data to support it, or funding a pilot that stalls because no one senior actually owns it.

Doing this once per company misses the pattern. Doing it across the whole portfolio at once shows you where the strongest starting points already exist, and where a company needs groundwork before it's worth a pilot at all. Source Capital ran exactly this kind of assessment across its own portfolio and turned it into a repeatable program instead of a one-off exercise.

To see where a single company stands, our free two-minute AI Audit scores readiness using The ADOPT Method™.

Score Opportunities by Impact and Feasibility

Once you know what each company can execute, score the specific opportunities inside it. AlixPartners frames this as an impact-versus-feasibility matrix, sorting use cases into quick wins and strategic bets rather than a single ranked list.

The six criteria from Firm-Level AI or Portfolio-Level AI map onto those two axes. Value and urgency determine impact. Readiness, ownership, measurability, and speed to value determine feasibility. They're the same criteria used to choose between firm-level and portfolio-level work, applied one level down to individual opportunities.

Quick wins score high on both axes and need little lift. Strategic bets score high on impact but lower on feasibility, usually because of readiness or a longer road to results, and most level three and four opportunities fall here. They're worth funding only when someone senior owns the outcome and there's a real way to measure it once it lands.

Where the Highest-Value Opportunities Usually Hide

Deloitte's research into AI-driven value creation in PE portfolios identifies five levers. Three of them are where most operating partners will find opportunities worth scoring.

Revenue growth comes from predictive analytics, lead scoring, and AI-driven sales outreach. McKinsey describes a PE-owned industrial materials supplier that used an AI sales agent to re-engage inactive customer accounts and reached a 30% engagement rate during the pilot. In Deloitte's research, a B2B distributor raised sales productivity 15% after adding AI-enabled lead scoring.

Margin expansion comes from automating manual, repetitive processes. Siemens, not a portfolio company but a useful benchmark for what's achievable, automated accounts payable across more than a million invoices a year, cutting manual effort by 60% and processing costs by 40%.

Differentiation comes from turning data a company already collects into something customers value. SkyChefs, an Aurelius portfolio company, used AI sensors to optimize inflight menus, improving meal profitability and cutting costs by 25%. The same data let SkyChefs show clients how its service improved customer satisfaction, which strengthened its position going into contract renewals.

The Trap That Keeps Portfolios Stuck at Level One

Most portfolios stall at levels one and two, and the reason is rarely the technology itself. Deloitte found only 22% of organizations feel highly prepared to handle the talent side of generative AI adoption, and only about a quarter feel highly prepared on governance and risk.

Without a plan for either, pilots stay isolated. A team automates one report, calls it a win, and never builds the case for anything bigger. The opportunity map only pays off if someone owns turning early wins into the next round of bets, instead of letting each pilot live and die on its own.

A Four-Step Process to Run Across Your Portfolio

Bring the pieces above together into one cycle you can repeat every couple of quarters.

1. Map. Audit every portfolio company against the same five readiness dimensions.
2. Score. Rank every opportunity that surfaces by impact and feasibility.
3. Sequence. Fund quick wins immediately. Stage strategic bets behind a named owner and a measurement plan.
4. Prove. Use the first results to justify the next round of investment, then run the cycle again as portfolio companies mature.

This mirrors what AlixPartners describes as opportunity mapping, roadmap development, execution, and value scaling, condensed into a cycle a portfolio can run on repeat instead of a one-time exercise.

Start With the Opportunities Your Portfolio Can Actually Execute

The pilots that succeed are usually the ones where the value, the data, and the ownership were in place before the first dollar was spent. Mapping readiness first and scoring opportunities second is how you find them, and funding the ones that clear both gives you the results you need to justify the next round.

Book a Strategy Call to map the highest-value opportunities across your portfolio.

Frequently Asked Questions

What's the difference between an AI opportunity and an AI use case?

For practical purposes, they're the same thing. Both describe a specific, scoped application of AI inside a business function, like automated lead scoring or invoice processing, as opposed to a broad company-wide AI strategy.

How many AI opportunities should a portfolio company pilot at once?

We generally recommend starting with one or two, each with a named owner. Running more than a company can staff properly is how pilots stall and never get evaluated.

Do all portfolio companies need the same readiness assessment?

Yes. The same five dimensions apply everywhere: data infrastructure, technology stack, leadership alignment, talent, and risk and compliance. What differs is the score each company gets, not the criteria themselves.

What's a quick win versus a strategic bet?

A quick win scores high on both impact and feasibility and needs little lift. A strategic bet scores high on impact but takes longer or carries more risk, so it needs a senior owner and a clear way to measure the result before it gets funded.

Why do so many portfolio companies get stuck automating small tasks instead of pursuing bigger wins?

Usually because of a talent or governance gap, not the technology. Without a leader who owns turning early wins into bigger bets, pilots stay isolated instead of building toward the next round of investment.