News

Automation is no longer simply an efficiency initiative for private equity operations; it is increasingly becoming a strategic and value creation imperative. For years, PE sponsors approached automation through a reactive lens by replacing aging equipment, while investing to improve throughput, expand capacity, or upskill labor as new needs arose.

However, as PE sponsors look to optimize automation solutions in the years ahead, upgrading machinery only on a reactive basis can leave value on the table. There are growing opportunities to take a front-footed approach and implement a broader catalogue of automation solutions within a portfolio company investment, especially as it pertains to customized machinery, vision systems, safety upgrades, and more recently AI-driven portfolio company analytics.

When automation is properly deployed, it can create sustained gains in throughput, labor productivity, quality,
visibility, and operational consistency.

One of the most underappreciated attributes of automation as a PE value creation lever is its durability. Many operational improvements implemented during a hold period depend on management teams maintaining processes and systems. Automation behaves differently. Once embedded into production, automated systems tend to persist. The cost savings and efficiency gains these systems deliver relative to manual processes in many cases outweigh the ongoing cost of maintaining and upgrading the automation. That durability can meaningfully influence the operating profile of the investment.

When automation is properly deployed, it can create sustained gains in throughput, labor productivity, quality, visibility, and operational consistency. As AI capabilities increasingly integrate into industrial workflows, from predictive maintenance to intelligent production planning, the impact becomes even more significant. For firms managing portfolios of manufacturing businesses, these differentiators have the potential to compound across the portfolio over time.

Automation at Work, Delivering Results
MiddleGround has applied this philosophy across a number of portfolio companies, using automation not simply to reduce costs, but to unlock operational improvements that persist long after implementation. At Race Winning Brands, a manufacturer of automotive and powersports components, we completed the first phase of an automated forging-press installation at the company’s Ohio facility, a project expected to generate approximately $9 million in equity value creation. The forging press had long constrained production due to lengthy die changeovers. Automating the process increased throughput while also improving ergonomics by reducing heavy lifting for operators and lowering temperatures on the production floor. The project represents the first phase of a broader modernization effort designed to make the forging operation safer and more efficient.

We applied a similar approach at the company’s Detroit-area facility, where MiddleGround engineers developed an automated system combining a robotic arm and hopper capable of processing hundreds of parts at a time. Previously, employees manually handled push rods during a heat-treatment process, holding them in induction heaters before transferring them to cooling stations, a repetitive task that added little value while exposing workers to unnecessary safety risks. Automation removed employees from a hazardous process, redeployed them to higher-value work and increased production throughput.

At exit, an automation roadmap signals that value creation is systematic rather than opportunistic, with a clear runway for the next owner to continue scaling.

Another area where we see a lot of promise is in gathering and applying data. At MiddleGround, we’ve worked on an AI application designed to compile operational data from a portfolio company’s systems that can be used to make better-informed business decisions. This system aims to produce data on different aspects of their business – providing visibility into key business trends and where there might be pockets of opportunity, allowing portfolio company leadership teams to spend more time on business strategy.

Automation as a Repeatable Value Creation Strategy
A critical component that determines success in planning and implementing automation systems is the utilization of a roadmap-based approach. Although it sounds simple, a longer-range view gives sponsors visibility across the entire hold period and encourages focus on initiatives that can be scaled over time. And if built properly, meaning that the end date extends beyond just your ownership, it creates a more durable operating framework and positions the business for continued success after exit. Building these capabilities at the firm level requires changes across the entire investment lifecycle.

First, automation must become part of the investment thesis at deal entry. Expected automation gains should be evaluated alongside procurement savings, pricing initiatives, and other operational improvements as part of the core value-creation plan. For many sponsors, this requires formalizing a capability that historically sat at the margins of the operating playbook.

Second, automation and process improvement teams must operate in parallel, rather than sequentially. In practice, that means conducting diligence together, optimizing the manufacturing process and automation solutions simultaneously, and implementing both through a coordinated effort. When lean manufacturing initiatives and automation programs are pursued independently, firms often end up automating broken processes. When integrated, the benefits can compound.

When sponsors empower portfolio companies to integrate their data systems with AI tools, it presents an opportunity to reduce one of the largest friction points in portfolio operations: reporting.

Third, firms should develop repeatable automation playbooks that can be deployed across a wide range of portfolio companies. Vision systems, in particular, represent a rapidly expanding opportunity set. A camera mounted above a production line, for example, can be trained to detect minute surface defects such as hairline cracks or small dents, before finished products leave the facility. Compared with major capital equipment upgrades, these systems require relatively modest investment while addressing one of the more costly failure modes in manufacturing: quality issues that go undetected until products reach customers. As technology has matured, applications across industrial environments have broadened significantly, enabling more advanced quality control, throughput optimization, predictive maintenance, and production monitoring capabilities than were previously available.

At exit, an automation roadmap supported by documented operational improvements and a pipeline of future initiatives may differentiate a business in a competitive sale process. It signals that value creation is systematic rather than opportunistic, with a clear runway for the next owner to continue scaling.

AI as the Operating Layer
Artificial intelligence is increasingly reshaping how sponsors monitor and manage portfolio company performance.

Historically, portfolio monitoring was labor-intensive and heavily dependent on management teams manually preparing reports. By the time operational trends emerge through monthly or quarterly reporting cycles, the window for early intervention is limited.

When sponsors empower portfolio companies to integrate their data systems with AI tools, it presents an opportunity to reduce one of the largest friction points in portfolio operations: reporting. Metrics, including labor efficiency, equipment utilization, scrap rates, and production throughput, no longer need to wait for month-end reporting cycles, and anomalies, such as labor inefficiency, utilization deterioration, or inventory dislocation, could be flagged automatically. Management teams are then able to spend less time assembling reports and more time thinking ahead, implementing strategic initiatives, and operating the business.

The objective is not to replace boots-on-the-ground operational engagement, but to make it more precise and effective. When sponsors and management teams operate from the same real-time data environment, they can align on priorities faster, identify root causes earlier, and work to intervene before operational issues materially impact performance.

What makes this especially powerful is the ability to identify trends across an entire portfolio simultaneously. PE sponsors can help detect emerging operational issues before they appear in quarterly reporting cycles, benchmark performance across facilities, and arrive at site visits with a far more targeted agenda.

The Window Is Narrowing
The conditions for embedding automation and AI capabilities into industrial PE operating models have never been more urgent.

Labor costs across manufacturing remain elevated, while skilled labor shortages continue to constrain throughput and capacity expansion. At the same time, automation technologies have matured, and implementation costs have become more accessible for middle market businesses. Meanwhile, AI-enabled operational monitoring has evolved from a theoretical concept into a practical tool with the potential to meaningfully improve portfolio oversight and decision-making.

Most sponsors are still in the early stages of treating automation as a true operational discipline. Firms that invest now in building centralized automation expertise, developing repeatable implementation playbooks, and integrating AI-driven operational monitoring may be advantaged as these capabilities become baseline expectations across PE.

For private equity firms managing industrial portfolios, the path forward is clear: automation must move from a line-item capital decision to a repeatable operational discipline, embedded into diligence from day one and deployed systematically across the portfolio.

About the Author
John Stewart is the Founding and Managing Partner of MiddleGround, which he established in 2018, where he oversees the overall management of the firm and serves on its investment committee.

John began his career as an hourly line worker at Toyota Motor Corporation, holding numerous management and executive positions over an 18-year career. He moved into private equity in 2007, joining Monomoy Capital Partners as a principal and head of the firm’s operating group. Over the next decade he was steadily promoted, becoming a partner in 2016. Across his career, John has worked on numerous transactions and served on the boards of more than 25 businesses, spanning both middle-market and Fortune 100 companies.

John is frequently sought after by fellow investors, limited partners, and business leaders for his vision and leadership in industrial manufacturing. His “blue collar” roots are the DNA of MiddleGround and set the tone for the firm’s culture.

This article is provided for informational purposes only and reflects the views of the author as of the date of publication. It does not constitute investment advice, an offer to sell, or a solicitation of an offer to buy any securities or investment products. MiddleGround Capital is a registered investment adviser. The operational improvements and strategies discussed herein are based on general industry observations and the firm’s experience; past operational results are not indicative of future performance. Any references to specific technologies, tools, or capabilities reflect the firm’s current understanding and are subject to change. This material should not be relied upon as a guarantee of any particular outcome.

Private equity investors are asking harder questions than they were a few years ago. Knowing that a portfolio beat public markets is no longer enough. Boards, CIOs, and investment committees now want to know what produced the result and whether it can be repeated. Was the return earned through manager selection? Strategy allocation? The timing of when capital went to work? Or did the market simply do the heavy lifting?

Those questions matter more now because private markets have moved from a niche allocation to a core holding. The investor here is the allocator — a pension plan, endowment, or fund-of-funds committing across dozens of managers — and the portfolio in question is that book of fund commitments, not the operating companies held inside any single fund. Investors still pay premium fees for access, and in exchange they want a clearer picture of what that access is buying them. The difficulty is that the standard benchmarking tools were never built to answer the question.

Benchmarking tells you what happened, not why
Most private equity benchmarking still measures one fund at a time. Quartile rankings, peer-group comparisons, and pooled return data all provide context, but none of them explain how a portfolio’s decisions combined to produce its overall result.

Most measurement tools stop at the fund level and struggle to explain how a collection of decisions
added up to total portfolio performance.

That gap matters, because a portfolio full of top-quartile funds can still disappoint. Capital may have been committed unevenly across vintage years. The portfolio may have leaned into the wrong strategy at the wrong time. Allocation choices may have shut it out of the market’s strongest-performing segments. Traditional benchmarking confirms the outcome; it rarely explains the drivers behind it.

Construction adds a further wrinkle. The same underlying data can yield materially different benchmark results depending on how the benchmark is built — which raises a fair question for any investor: if the yardstick itself can move, how much weight should the answer carry?

The missing piece: portfolio-level attribution
The problem grows at the portfolio level, where results are shaped by a series of decisions that compound over time. How much capital should be committed? When should it be deployed? Which managers should receive it? How should it be spread across buyout, growth, venture, secondaries, and other strategies? Each choice moves the final number.

Yet most measurement tools stop at the fund level. They can show how an individual manager performed, but they struggle to explain how a collection of decisions added up to total portfolio performance. That limitation used to be manageable, when portfolios were simpler and consisted largely of traditional closed-end funds. Today’s portfolios — customized mandates, separately managed accounts, evergreen structures, co-investments, secondaries, and wider global exposure — carry far more flexibility and far more complexity. As portfolios have grown more sophisticated, so have the questions investors ask of them.

A different way to look at performance
Answering those questions calls for a different method. HarbourVest’s approach isolates the major decisions behind a portfolio’s return. Drawing on a transparent, investable universe, it builds thousands of simulated portfolios and changes one variable at a time — manager selection, strategy allocation, or commitment timing — to estimate how much each decision contributed to the result.

Telling a genuinely strong manager from a lucky one — or a shrewd allocation from a fortunate market — takes a record deep
enough to know how thousands of other funds fared over the same years.

In practice, the analysis is designed to show whether performance came from backing the right managers, investing in the right parts of the market, deploying capital at the right time, or some mix of the three. The framework rests on a proprietary database spanning nearly four decades and covering more than 23,000 funds, 5,000 general partners, and roughly 73,000 holdings. Rather than measure a portfolio against a single benchmark, it measures the portfolio against a broad range of realistic alternatives drawn from that same universe. The output is less a ranking than a diagnostic.

The idea will feel familiar to many investors. Private equity firms have long decomposed a company’s return into its sources of value — revenue growth, margin expansion, leverage, add-on acquisitions, and multiple change. HarbourVest applies the same logic one level up, at the portfolio. The goal is not to confirm that a portfolio performed well, but to understand what made it perform.

From Theory to Practice: Two investors, One Market
The method is easiest to follow through an example. Consider two institutional investors with access to the same universe of private equity opportunities over the same period, one earning 1.52x and the other 1.89x MOIC — multiple of invested capital, or the dollars returned for every dollar put to work. Traditional benchmarking would note that the second portfolio won and stop there. Attribution analysis starts from a shared reference point — the 1.66x a typical portfolio drawn from that same universe would have earned — and asks how each investor’s decisions moved them above or below it.

How to read the charts below
The market baseline (1.66x) is the return a typical portfolio drawn from the same universe would have earned. Each factor below shows how much one decision moved an investor above or below that baseline; the four add up to the gap between the market and the investor’s result.

(1) Manager selection — which private equity firms — the general partners, or GPs — and which of their funds the allocator commits capital to. It measures whether the specific funds chosen beat or lagged the typical fund of their type and is usually the largest single swing.

(2) Vintage timing — when capital is committed. A fund’s vintage is the year it begins investing, and returns vary by vintage; steady pacing spreads the bet across years, while concentrated pacing wagers on timing.

(3) Stage & size — where in the market capital is weighted — large versus mid- versus small-market buyout, and, more broadly, buyout versus growth versus venture.

(4) Geography — how commitments split across regions such as North America, Europe, and Asia, and whether that regional mix helped or hurt versus a typical portfolio.

Investor 1 finished modestly below the market, at 1.52x. A handful of decisions explain the 0.14x shortfall. Weaker manager selection was the largest drag, costing 0.08x; less consistent deployment across vintage years cost a further 0.04x; and a heavy tilt toward large-buyout strategies — where the portfolio underweighted the small- and mid-market segments that led during the period — cost another 0.02x. Geography was broadly neutral. The portfolio committed more to managers that went on to underperform and missed some of the market’s stronger vintages by deploying unevenly. Through a traditional lens, it simply lagged; attribution names the specific choices that put it there.

Investor 2 finished well ahead of the market, at 1.89x, and for reasons the analysis can point to. Stronger manager selection added 0.19x — by far the largest single contributor. More consistent deployment across vintage years added another 0.04x, and a tilt toward the small- and mid-market segments that led during the period added 0.01x, with geography a marginal 0.01x drag. The allocation is the mirror image of Investor 1: where the first portfolio put 96% of its capital into large buyout, Investor 2 committed only about a fifth there and the balance to mid- and small-market funds — the parts of the market that outperformed. The conclusion is not merely that Investor 2 outperformed; it is that identifiable construction decisions, not luck, did the work.

Analysis
Read side by side, the two portfolios show the same lesson from opposite directions: manager selection was the swing factor, worth nearly 0.28x of MOIC between them, while consistent timing and diversification quietly added to or subtracted from the edge.

Strip away the decimals and the story is simple. Both investors fished the same pond, yet one came back with far more. The biggest reason was people: Investor 2 was simply better at picking managers. Handed the same roster of firms out raising funds, it backed the teams that went on to deliver and steered clear of the ones that stumbled — the private-markets version of hiring the right people before the rest of the market catches on. That judgment alone explains most of the gap. Investor 2 also read the opportunity better, leaning into the small- and mid-market funds that led the period while Investor 1 crowded into large buyout, and it committed capital at a steadier pace rather than betting on a couple of vintage years. None of it was luck or a rising tide — both faced the identical market. It came down to a handful of good decisions, and the best of them was knowing which managers to trust.

The standard is changing
Investor expectations are moving. High-level performance reporting no longer satisfies many boards and investment committees; they want evidence for how a return was generated and which decisions helped or hurt. For managers and allocators alike, that means performance has to be explainable, not merely reportable.

Doing that well is harder than it looks. Telling a genuinely strong manager from a lucky one — or a shrewd allocation from a fortunate market — takes a record deep enough to know how thousands of other funds fared over the same years. It is that asset-level detail, accumulated over nearly four decades, that lets HarbourVest segment where a portfolio’s returns actually came from.

As private markets portfolios keep growing in size and complexity, attribution is likely to become a standard part of fiduciary oversight. Investors need to separate returns handed to them by market conditions from returns earned through manager selection, portfolio construction, and capital deployment. In private markets — where capital is committed and locked up for years, and where performance dispersion across managers runs far wider than in public markets — those construction choices carry real weight. Knowing whether a portfolio outperformed will always matter. Understanding why may matter more.

About the Author
Sofia Gertsberg joined HarbourVest in 2016 to lead the firm’s Quantitative Investment Science team. The team’s objectives are to utilize HarbourVest’s proprietary data set to develop quantitative tools and models to enhance fundamentally-driven investment and portfolio construction process across the firm.

Sofia joined the firm from State Street Global Advisors, where she was the global head of fixed income and money market risk leading the investment risk oversight of assets across active, smart beta, and passive strategies for fixed income, cash, and currency portfolios. Her responsibilities included leading a global team of investment risk managers and analysts, designing and implementing investment risk monitoring framework, setting internal risk limits, and conducting risk management due diligence for outside managers. Sofia was also a voting member of SSGA’s Sub-advisor Oversight and Valuation Committees. Prior to that, she was a market risk manager at State Street Global Markets. She also held a director of analytics role at Debt Exchange.

Sofia received a BS in Economics from the University of Latvia in 2000 and an MBA from Boston University in 2003. She speaks fluent Russian and Latvian.

Data and Illustration Notes

As of June 30, 2025. Source: HarbourVest proprietary data set comprised of information aggregated from multiple data sources, including HarbourVest and third-party data providers. Net of underlying management fees and carried interest. Gross of HarbourVest management fees and carried interest. Other expenses borne by investors in the HarbourVest managed funds / accounts may reduce returns. Performance in USD. For illustrative purposes only. Past performance is not a reliable indicator of future results. Not representative of any HarbourVest fund, account, or experience.

HarbourVest Partners, LLC is a registered investment adviser under the Investment Advisers Act of 1940. This material is solely for informational purposes and should not be viewed as a current or past recommendation or an offer to sell or the solicitation to buy securities or adopt any investment strategy. The opinions expressed herein represent the current, good faith views of the author(s) at the time of publication, are not definitive investment advice, and should not be relied upon as such. This material has been developed internally and/or obtained from sources believed to be reliable; however, HarbourVest does not guarantee the accuracy, adequacy or completeness of such information. There is no assurance that any events or projections will occur, and outcomes may be significantly different than the opinions shown here. This information, including any projections concerning financial market performance, is based on current market conditions, which will fluctuate and may be superseded by subsequent market events or for other reasons. The information contained herein must be kept strictly confidential and may not be reproduced or redistributed in any format without the express written approval of HarbourVest.

Nothing herein should be construed as a solicitation, offer, recommendation, representation of suitability, legal advice, tax advice, or endorsement of any security or investment and should not be relied upon by you in evaluating the merits of investing in HarbourVest funds or in any other investment decision.

Market analysis is not representative of any HarbourVest product. This presentation reflects quantitative analysis of the global private equity industry derived from HarbourVest’s proprietary Quant Database, a compilation of private equity partnership and transactional data drawn from internal and external sources. The proprietary Quant Database has been developed internally based on information obtained from sources believed to be reliable; however, HarbourVest does not guarantee the accuracy or completeness of such information. This proprietary database is intended to be representative of the broader private equity market and does not reflect the investment performance of any HarbourVest investment or the experience of any investor in any HarbourVest fund.

Artificial intelligence has moved from experimentation to infrastructure in M&A. Dealmakers are increasingly relying on AI to source opportunities, analyze targets, and identify risks, but when it comes time to sign a deal, they still want a human holding the pen.

That tension sits at the center of The New Deal Team, a new report from Datasite and FT Longitude based on a March 2026 survey of 1,000 senior dealmakers across 27 countries. All respondents had lead decision-making responsibility on at least three transactions during the previous 24 months. Participants included professionals from private equity, corporate development, law, accounting, and advisory firms across the Americas, EMEA, and APAC regions. While 62% of respondents believe human-only decision-making is no longer defensible in complex transactions, 45% say the decision to proceed to signing should always remain exclusively human.

The survey shows AI is most deeply embedded in the most data-intensive phases of dealmaking. Due diligence leads adoption, with 50% of respondents reporting regular or fully embedded use and only 4% reporting no use at all. Sourcing and screening follow, with 43% reporting regular or embedded use and 96% either using or exploring AI in some capacity. Strategy and ideation, deal marketing, and deal preparation also showed meaningful adoption as firms increasingly use AI to accelerate analysis and workflow management.

As AI takes on more of the heavy lifting in M&A, human judgment becomes more valuable, not less.

Yet adoption drops sharply as transactions move toward closing. Thirty-one percent of respondents report no AI use at closing, and only 16% use it heavily. Board reporting and governance show similar patterns, with 27% reporting no AI use at all. When asked where responsibility should reside for proceeding to signing, 45% said the decision should be made entirely by humans, while another 33% favored a human decision informed by AI. Just 7% said they would proceed based on an AI recommendation without human review.

“AI can automate analysis, but it can’t own accountability,” said Rusty Wiley, president and CEO of Datasite. “As AI takes on more of the heavy lifting in M&A, human judgment becomes more valuable, not less. The firms that outperform will be those that combine the speed and scale of AI with the trust, experience and oversight that only people can provide.”

The data also indicates that advisors are moving faster than corporate acquirers and financial sponsors in embedding AI into transaction workflows. The largest gap appears in due diligence, where 53% of advisors report regular or fully embedded AI use compared with 45% of clients. If that trend continues, AI capabilities could increasingly become a source of differentiation among investment banks, accounting firms, consultants, and legal advisors competing for transaction mandates.

AI is becoming indispensable in dealmaking and investing, but adoption alone isn’t the advantage.

The findings also suggest AI is influencing transaction outcomes rather than simply improving efficiency. Twenty-four percent of respondents said AI helped them complete a transaction that otherwise would have been missed, while 43% said AI is already making better deal decisions than humans in certain situations. Two-thirds of respondents believe using AI across the deal lifecycle is an effective way to reduce transaction risk.

The impact is beginning to extend beyond deal execution and into firm operations. Twenty-six percent of respondents said they had delayed or canceled hiring for a role because AI could perform the work as effectively. Nearly half of respondents said they would accept AI input in dealmaking if it was 80% as accurate as a human, and 71% believe firms that fail to adopt AI today will struggle to compete within five years.

“AI is becoming indispensable in dealmaking and investing, but adoption alone isn’t the advantage,” said Raj Bakhru, general manager of Blueflame AI, a provider of AI-powered workflow and research tools for private equity and investment banking professionals that was acquired by Datasite in 2025. “The real challenge is ensuring AI outputs are accurate, secure and trusted. Strong governance, transparent workflows and human oversight are what determine whether AI creates value or introduces risk.”

The survey’s governance findings reinforce that view. Accuracy and security ranked as the two most important attributes dealmakers require from AI tools, cited by 71% and 70% of respondents, respectively. Human review and validation were identified as the leading mechanism for building trust in AI-generated outputs, followed by the use of purpose-built applications and deployment within secure environments.

As firms increasingly incorporate AI into sourcing, diligence, portfolio monitoring, and reporting processes, questions surrounding oversight, accountability, and fiduciary responsibility are moving beyond IT departments and into investment committees and boardrooms. Security and compliance concerns remain the largest barriers to broader adoption, cited by 44% of respondents, while 35% pointed to a lack of internal expertise.

The report ultimately suggests that AI is changing the nature of dealmaking rather than replacing dealmakers.

Regional differences were pronounced. Asia-Pacific respondents reported the highest levels of AI adoption across most phases of the deal process and were generally the most willing to delegate decision-making authority to AI. Europe, Middle East, and Africa respondents were the most cautious, particularly around security and regulatory concerns. Dealmakers in the Americas were the most concerned about competitive pressure, with 74% saying firms that fail to embrace AI today will struggle to compete within five years.

Looking toward 2030, respondents expect AI’s benefits to vary by stage of the deal process. In due diligence, the most commonly cited benefit is improved identification and mitigation of risks. In sourcing and screening, dealmakers expect AI to help manage a greater volume of opportunities. In strategy and ideation, respondents anticipate better decision-making, while users expect AI to improve board reporting and governance through faster access to more accurate information and insights.

The report ultimately suggests that AI is changing the nature of dealmaking rather than replacing dealmakers. As technology assumes more responsibility for sourcing, diligence, and transaction preparation, the skills that remain uniquely human—negotiation, relationship management, strategic judgment, assessing trust and intent, and accountability—become increasingly valuable.

For private equity firms, the more interesting question may no longer be whether AI will become part of the investment process, but how quickly AI proficiency becomes a baseline expectation. Virtual data rooms, CRM systems, and digital diligence tools were once differentiators before becoming standard infrastructure. The survey suggests AI may be following the same path, with the competitive advantage shifting from access to the technology to how effectively firms integrate it into their investment processes.

Minneapolis-headquartered Datasite, led by President and CEO Rusty Wiley, provides software used to manage information flow across private market transactions. Its product portfolio includes virtual data rooms, deal-sourcing platform Grata, agentic AI platform Blueflame AI, and governance software provider Sherpany.

The survey was conducted by FT Longitude, the research and thought leadership division of the Financial Times Group, on behalf of Datasite in March 2026.

Click HERE to download a copy of Datasite’s The New Deal Team.