All Insights

Diligence

The Art of the LP Reference Call: What to Ask, Who to Call, and How to Read Between the Lines

A GP-approved reference list is a sales tool. Great LPs treat it as a starting point, not the destination. Here is how to run reference calls that actually change your conviction.

Matthew S. · Founder, Allocator Desk · July 2026 · 9 min read

Every LP has sat through a reference call that felt like a hostage video. The reference is friendly, articulate, and suspiciously well prepared. They love the GP. They cannot think of a single weakness. They have to jump in ten minutes for another call, which is a shame, because they had so much more to say.

You end the call, thank them, and write "positive reference" in your notes. And you have learned exactly nothing.

Reference calls are one of the highest leverage steps in diligence and one of the most poorly executed. Done well, they surface the things a data room cannot: how a GP behaves when a deal breaks, whether the team actually likes each other, and whether the culture of the firm matches what the partners say on stage. Done badly, they are a checkbox that produces a false sense of confidence.

This is a practical guide to running reference calls that are worth the hour on your calendar.

Why the standard approach fails

The default reference process looks like this. The GP sends a list of six names. You take the calls. Everyone says nice things. You proceed.

There are three problems with that flow, and they compound.

First, the GP curates the list. That is not a scandal, it is human nature. But the list you receive is the list of people who will not hurt the raise. It is not a random sample of the GP's counterparties.

Second, the questions LPs typically ask are the ones the reference is most prepared for. "How is the team to work with?" "How did they perform during the pandemic?" These are softballs that the reference has heard four times this quarter.

Third, the calls tend to be short. Thirty minutes gets you pleasantries plus one substantive topic. It does not get you the story of how a deal actually unfolded.

The good news is that all three problems have straightforward fixes. You just have to be willing to work a bit harder than most of your peers, which is exactly the kind of edge that compounds over a career.

Building a reference universe

The GP list is your starting point. Your job is to expand it to something like a real sample.

Start with the ones they gave you. Take those calls, but treat them as a way to get the official narrative on tape. You will hear the version of events the GP wants you to hear. That is not useless. It is your baseline.

Then build an off-list universe. There are five buckets worth mining, and most LPs stop at one or two.

Portfolio company management. LinkedIn and PitchBook will show you every CEO who has worked with this GP. Prioritize the ones who exited, the ones who were replaced, and the ones whose companies underperformed. That last group is where the interesting information lives. A GP who behaved well with a struggling founder tells you something a smooth win never will.

Co-investors and syndicate partners. Every deal has a lead and a support cast. Call the support cast. They saw the same deal, sat in the same board meetings, and often have a franker view of how the lead behaved when things got tense.

Advisors who were paid. Investment bankers, legal counsel on transactions, quality of earnings providers, industry consultants. These people were paid regardless of outcome, which means they can be unusually candid. They also have opinions on how the GP negotiates, which is diagnostic.

Former employees. This is the highest-yield and most misused bucket. LPs sometimes avoid ex-employees because they worry about bias. That is backwards. Some bias is fine when you are triangulating. What you want is enough voices that the signal shows through the noise. Two disgruntled ex-partners telling different stories is data. Two disgruntled ex-partners telling the same story is a finding.

Other LPs in prior funds. Especially the ones who did not re-up. A quiet coffee with an LP who passed on Fund IV is worth more than a glossy hour with an LP who is the anchor of Fund V.

You do not need twenty calls. You need eight to twelve, weighted toward the off-list universe, with enough coverage across funds and vintages that patterns can emerge.

Questions that actually work

There is a body of questions that consistently produce information, and another body that consistently produces marketing. Learn the difference.

Retire these. "What are their strengths?" "How would you describe their culture?" "Would you invest with them again?" These questions cue the reference to reach for their prepared answer. You will get the tagline from the pitch deck read back to you with slightly different word choice.

Use these instead.

"Walk me through a deal you did together from first meeting to exit." The narrative form is diagnostic. It is very hard to bluff a story. Watch for where they slow down, where they get vague, and where they name specific people. A reference who talks in first names about the GP's junior team knows the firm. A reference who only mentions the two founding partners has met the firm.

"What was the moment you were most nervous during that deal, and what did the GP do?" This surfaces behavior under pressure, which is what you actually care about. Almost every deal has a wobble. Any reference who cannot remember one either was not involved or is not being straight with you.

"Who else at the firm did you work with day to day?" The answer tells you whether the deal partner is a solo act with a support team, or a genuine institution. Both can be fine, but they price differently. A firm that sells "twelve investment professionals" and delivers one is a firm you should discount.

"What would you do differently if you were doing this deal again?" Even a friendly reference will usually cough up one honest reflection here, and their answer often includes an implicit critique of the GP that would never survive a direct question.

"Who else should I talk to?" Asked at the end of every call, once. This is how the off-list universe grows organically. Some of the best references in a diligence process are people the GP never mentioned and never would.

Layered on top of those, you need the specifics that matter to your process. Attribution on wins and losses. Whether the GP renegotiated fees or terms mid-hold. How the GP handled founder transitions, board disagreements, follow-on rounds, and down rounds. Whether the firm made good on side letters and MFN commitments. Whether reporting arrived on time and matched what the GP said on the phone. If you have not read the side letters and MFN guide, those are the questions that most naturally connect back to it.

Listening for the tell

The content of the call matters. The delivery matters more.

There is a small library of tells worth internalizing.

The pause before a positive. If a reference hesitates before saying something nice, ask what they were about to say instead. Nine times out of ten there is a hedge under the compliment.

The absent negative. A reference who cannot name a single weakness is either untruthful or uninformed. Both are red flags. If you get this, gently press with "if you had to nominate one thing the firm needs to work on, what would it be?" Watch how long they take.

Overreach. If a reference is claiming credit for the GP's returns or, worse, if the GP is claiming credit for the reference's returns, you have found a status game. Status games predict misaligned behavior later.

Team specificity. Great references talk about a team by name. Weak references only know the fundraiser. If you are hearing about the same one or two partners in every call and the firm markets itself as a deep bench, that is a datum.

Language drift. When several references use identical phrases, you are hearing coaching. GPs sometimes prep their references. That is not a disqualifier, but it should be noted. Ask an open-ended question, listen for the scripted answer, and then move to something the script did not cover.

None of these are conclusive on their own. They are patterns. Diligence is pattern recognition run on enough surface area to be trustworthy.

Choreographing the process

The mechanics matter almost as much as the content.

Do the off-list calls first if you can. You will hear the story in its most unfiltered form before the GP's chosen references have a chance to set the narrative. Even ten minutes of prior context makes the on-list call much more useful, because you know what you are testing.

Time-box appropriately. Forty-five minutes is the sweet spot. Thirty is too short for a real deal walkthrough. Sixty invites throat clearing. If the reference is generous, extend it. If they are guarded, do not stretch.

Take notes in the same structure across calls. You are going to be triangulating twelve calls, and unless you have a consistent template, patterns will slip past you. A simple grid with the reference, the deals discussed, attribution, behavior under stress, culture signals, and their unprompted comments works well. This is exactly the kind of thing a structured [operational due diligence framework] is designed to catch.

Debrief while it is fresh. Ten minutes right after the call to write your read is worth more than an hour a week later. Include your gut. "Sounded coached" is a legitimate note. Your future self will thank you for the honesty.

Loop back. If three calls in you notice a pattern, use it in the next call. "A couple of other people I have spoken to mentioned that the firm's junior team turns over faster than they would like. Is that consistent with what you saw?" You are not laundering hearsay, you are testing a hypothesis with someone who can confirm or deny.

What reference calls should change

Reference calls do not usually flip a decision from yes to no. If your diligence memo depends on the reference calls to save it, you were probably going to invest anyway and the memo is doing the work of an alibi.

What good reference calls do is calibrate conviction. They move a soft yes to a firm yes, or a firm yes to a conditional yes with a side letter, or a firm yes to an actual no. They also sharpen the questions you take into the re-up decision three or five years from now. The line "I am tracking this concern from Fund V diligence" is one of the most useful sentences an LP can write in a monitoring note. It is also how institutional memory is built, one honest note at a time.

The compounding value of reference calls is not the yes or no on this fund. It is the network you build over a decade of thoughtful conversations with founders, co-investors, advisors, ex-partners, and other LPs. Those relationships pay dividends every fundraising cycle. They also make you a better allocator, because you stop taking the pitch at face value.

A short and honest checklist

Before you close out the reference process on a fund, ask yourself a few questions.

Did you talk to at least three people who were not on the GP's list. Did you talk to at least one person who had a bad outcome with the firm. Did you get a specific deal walkthrough from at least half your references. Did any two references, on separate calls, tell you the same unprompted concern. Did anyone tell you something the GP would not want you to hear.

If the answer to most of these is no, you did not do reference calls. You did a formality.

The GPs you most want to back will respect an LP who does the real work. The ones who do not are telling you something too.

Topics

• Diligence
• LP Operations
• GP Evaluation