All Insights

Portfolio Construction

Co-Investments for LPs: How to Build a Disciplined Program That Actually Adds Value

Co-investments promise lower fees and concentrated alpha, but the fee savings only matter if the program is built with the same discipline as primary fund selection. A working framework for sourcing, underwriting, sizing, and measuring a co-investment book.

Matthew S. · Founder, Allocator Desk · June 2026 · 11 min read

Co-investments are one of the most misunderstood tools in an LP's portfolio. They get pitched as free alpha - lower fees, concentrated exposure to a GP's best ideas, faster capital deployment - and many LP programs are built on that pitch. The reality is more complicated. Co-investments can absolutely improve net returns, but only if the program is built with the same discipline as primary fund selection. Without that discipline, co-investments quietly add concentration risk, adverse selection, and operational burden that the fee savings never recover.

This is a working framework for building a co-investment program that holds up across cycles - what to underwrite, how to size, what to say no to, and how to measure whether the program is actually adding value.

Why co-investments look better than they are

The standard pitch is fee economics. A primary commitment carries 2% management fee on commitments and 20% carry above an 8% hurdle. A co-investment is typically zero fee, zero carry, or close to it. On a single deal, that fee differential is worth roughly 300 to 500 basis points of gross IRR depending on hold period. Compounded across a program, the math is genuinely compelling.

The problem is that fee savings are a deterministic benefit applied to a probabilistic outcome. You only collect the fee savings if the underlying deal performs. And the deals that GPs syndicate as co-investments are not a random sample of their portfolio - they are the deals where the GP needs more equity than the fund can write alone. That selection is not always negative, but it is never neutral. Understanding why a specific deal is being offered is the single most important diligence question, and the one most often skipped.

The adverse selection question

Co-investment offers come from three places, and the source matters more than most LPs admit.

The first is capacity-driven syndication. The fund found a deal that fits the strategy perfectly but is too large for the concentration limits in the LPA. The GP needs to syndicate the excess equity. These are the cleanest offers. The GP wants the deal, the fund wants the deal, and the co-investment is just a structural workaround. Performance tends to track the underlying fund.

The second is conviction-driven syndication. The fund could write the full check but the GP wants to deepen exposure to a name they believe is asymmetric. They offer it as a co-invest to anchor LPs to reinforce the relationship and signal conviction. These can be excellent, but they require the LP to underwrite the GP's conviction, not just the deal economics.

The third is risk-management syndication. The deal is in the strategy zone but the GP is uncomfortable holding the full position - sector concentration, single-name risk, leverage profile, or a thesis that depends on factors outside the GP's normal underwriting. They syndicate to share the downside. These are the dangerous offers, and they look identical to the first two in the deck. The only way to distinguish them is direct conversation with the deal partner and a careful read of the fund's existing concentration in the same sector or theme.

A simple discipline: before any co-investment underwriting begins, ask the GP in writing why this specific deal is being syndicated and what proportion of the fund's available capital is going into it. If the answer is vague or the fund check is small relative to total commitment, the deal needs heavier scrutiny, not less.

Building the underwriting framework

A co-investment underwrite is not a mini fund underwrite. The questions are different because the holding period, exit dynamics, and recourse to the GP are different.

The core underwriting dimensions:

Deal-level conviction. Is the company itself something the LP would want to own through a public market vehicle if it were available? If the answer requires the GP's value-add to be true, the bar is higher. LPs who lean heavily on "the GP will fix it" tend to underperform on co-investments because the GP's attention is split across the whole fund and the co-invest is not where they spend the marginal hour.

Entry valuation and structure. Co-investments are typically offered at the same valuation as the fund's primary investment. That sounds fair until you remember the fund got fee load and management economics on the deal that the co-invest does not. The economic equivalence is not actually equivalent. Whether that matters depends on the deal, but it should be acknowledged in the model.

Concentration impact. A single co-investment is rarely large enough to move portfolio concentration. A program of ten to twenty co-investments over a vintage absolutely can. LPs need a sector and geography concentration view that includes co-investments alongside fund look-through exposure, refreshed at least quarterly.

Exit alignment. Co-investment exits follow the fund's exit decision. If the fund is in year eight and the LPA forces a sale, the co-invest is sold too - even if the LP would rather hold. There is no independent exit option in most co-investment structures.

Information rights. What does the LP actually get? Quarterly company updates? Board observer rights? Access to the GP's internal valuation memos? Most co-investments offer thin reporting because the GP is not contracted to provide it. The LP needs to know going in what they will and will not see for the next five to seven years.

Sizing discipline

This is where most co-investment programs fail. The mechanical version - "we will allocate 1% to 2% of the portfolio to any deal that clears underwriting" - sounds disciplined but produces a portfolio that is concentrated in the deals offered fastest, not the deals with the best risk-adjusted return.

A better approach uses two sizing dimensions:

Conviction tiers. Define three or four conviction levels with explicit check size ranges before the offer arrives. A tier-one conviction deal might be 2% to 3% of the portfolio. A tier-three deal might be 0.5%. The tier is set by the underwriting committee, not negotiated with the GP. This forces conviction to be expressed in check size, which is the only language deals understand.

Annual budget. Cap total co-investment deployment per vintage as a percentage of total commitments that year - typically 20% to 40% for established programs, less for new ones. This prevents the program from absorbing every offer that comes in and forces the team to say no to mediocre deals because the budget is finite.

The combination forces a real comparison across the deal pipeline rather than first-come-first-served deployment.

What to say no to

Saying no to co-investments is hard because each individual no looks like a missed opportunity and the relationship cost with the GP is real. A few categories of deal that almost always justify a no:

• **Deals offered with less than three weeks of diligence time.** The compressed timeline is a structural disadvantage and the deals that are offered with no time are disproportionately the ones the GP is rushing for a reason.
• **Deals that move the portfolio above a stated concentration limit on sector, geography, or single name.** The limit exists for a reason. Override it once and it stops being a limit.
• **Deals where the GP partner cannot articulate the downside scenario in a fifteen-minute call.** This is not a test of the partner's intelligence. It is a test of whether the deal has been stress-tested.
• **Deals where the lead GP is not putting incremental equity from the fund alongside the syndication.** If the fund is at its position limit, that is one thing. If the fund chose not to fully fund a deal it is also syndicating, that is signal.

A healthy program declines roughly half the deals it sees. Programs that take everything offered are not programs - they are passive recipients of GP capacity needs.

Measuring whether the program is actually working

The reporting most co-investment programs produce - "we deployed $X across N deals at Y average check size" - tells you nothing about whether the program is adding value. The actual measurements that matter:

Co-investment IRR and MOIC versus the underlying fund's IRR and MOIC for the same deals. This is the only true apples-to-apples test of the fee benefit. If the co-investment is delivering meaningfully better net returns than the same dollar in the fund would have, the program is working. If it is delivering similar or worse, the fee savings are being eaten by something - usually adverse selection or concentration drag.

Co-investment IRR versus a passive private equity benchmark for the same vintage. This tests whether the program is adding value relative to just committing the same dollars to more primary funds. Co-investments should beat this benchmark by at least 200 to 300 basis points net to justify the operational cost.

Deal-level attribution. What percentage of co-investment returns come from the top two or three deals? If the answer is above 70%, the program is closer to a venture portfolio than a diversified co-investment book and needs to be discussed as such with the investment committee.

Decline ratio. What percentage of offered deals did the program decline, and how have the declined deals performed? This is uncomfortable to measure but it is the only honest test of the underwriting discipline.

Building the operational layer

Co-investment programs fail operationally more often than they fail on underwriting. The deals close fast, the documentation is bespoke, the reporting is irregular, and the cash flows do not fit the standard fund waterfall. A few operational disciplines that are not optional:

A central document repository per deal with the SPA, side letters, capital call notices, distribution notices, and any reporting the GP provides. The repository is built at deal close, not retrofitted later when someone asks a question.

A consolidated co-investment ledger that tracks commitment, called capital, distributions, current fair value, and the corresponding fund commitment so cash flows can be reconciled.

A quarterly valuation refresh. The GP will provide a number; the LP should have an independent view, even if it is just a check that the GP's number is internally consistent with the fund's reported NAV trajectory.

A pre-IC checklist that gets completed for every co-investment, not just the large ones. The checklist is the institutional memory of why each deal was approved, and it is the only artifact that lets the program learn from its own decisions over time.

The bottom line

Co-investments can meaningfully improve net returns for LPs who treat them as a discipline rather than an accommodation. The fee economics are real, but they are not the source of edge. The edge comes from honest underwriting of why each deal is being offered, sizing discipline that resists the offered-fastest bias, and operational rigor that keeps the program coherent across vintages.

Programs that get this right add hundreds of basis points of portfolio-level return over a cycle. Programs that do not get this right add concentration risk and operational drag in exchange for fee savings that never quite materialize. The difference is not the deals - it is the program around the deals.

Topics

• Co-Investments
• Portfolio Construction
• LP Operations
• Manager Selection
• Diligence