- Give the same data room to two associates and you get different evidence, not just different conclusions — and both memos will read as complete.
- Diligence variance has three sources: the standard is tacit, the standard is partner-specific, and the standard erodes under deal pressure.
- Lengthening the checklist makes variance worse, because a 300-line list produces skimming with a paper trail suggesting otherwise.
- A usable standard is written at the level of the question rather than the topic: 'customer concentration' is a topic, and a list of what must be established for each top-ten customer is a standard.
- Scoring every deal against the same framework buys comparability and an audit trail on absence; it does not buy a decision, which stays with the investment committee.
- A firm that has not written its standard down reconstructs its diligence method from scratch roughly every four years as the bench turns over.
Give the same data room to two second-year associates and you get two different memos. Not different conclusions from the same evidence, which would be healthy. Different evidence. One will have built out the customer cohort analysis and skated the supplier side. The other will have found the supplier concentration and taken the retention number at face value. Both memos will read as complete. Neither reader will know what the other one looked at and did not find.
This is the most expensive quiet problem in a mid-market fund, and it is not a training problem. It is a specification problem. Nobody wrote down what a complete first-pass diligence looks like at this firm, so each associate reconstructs it from whatever they absorbed on the last deal they staffed.
Where the variance actually comes from
Three sources, in descending order of size.
The standard is tacit. Most firms have an investment thesis document and a memo template. Neither is a diligence standard. The memo template tells you what sections the output has. It does not tell you what has to be true before you write the section. A partner reading a thin market section cannot tell whether the associate looked and found nothing or did not look.
The standard is partner-specific. In practice the firm does not have one standard. It has four, one per partner, and the associate learns which one applies from who is on the deal. This is often defensible — partners have real, earned, differing views about what matters in their sectors. It becomes a problem when the deal is staffed by whoever is free rather than by sector, and when the partner who cared most about a particular failure mode is not in the room.
The standard erodes under deal pressure. Every firm's diligence is more thorough in January than in the fourth week of a competitive process. The workstreams that get cut when the clock tightens are not the least important ones. They are the ones with the least visible output, which is usually the ones where finding nothing looks identical to not looking.
What the fix is not
It is not a longer checklist. Firms that respond to this problem by lengthening the diligence checklist reliably make it worse. A 300-line list produces skimming, and skimming produces the same variance with a paper trail suggesting otherwise. The checkbox gets ticked because the item was considered, and "considered" covers a range from a three-hour analysis to a glance.
It is also not more review. Adding a partner review gate catches errors in what was written. It does not catch the absence of what was not, because the reviewer is reading the memo, not the road not taken. A reviewer can only find a missing workstream if she happens to think of it, which returns you to the tacit standard.
What the fix is
The standard has to be explicit, scoped, and applied by something that does not get tired in week four.
Explicit means written at the level of the question, not the topic. "Customer concentration" is a topic. "For each of the top ten customers: contract start date, renewal mechanism, notice period, whether pricing is contractual or list, and whether the relationship predates current management" is a standard. The first can be satisfied by a paragraph. The second can be audited.
Scoped means the standard varies by what you are buying. The version that applies to a founder-run services business with no CFO is not the version that applies to a sponsor-owned platform on its third add-on. Firms that maintain one universal standard end up with one that is either too thin for the hard deals or too heavy for the easy ones, and in both cases people stop using it.
Applied consistently is where most of the value is and where firms have historically had no good option. A standard that requires an associate to work through 140 questions on every deal is a standard that gets abandoned in a live process. The questions still need asking. What changed recently is that the first pass through them no longer has to be done by hand — which means the associate's time moves from establishing what the room says to judging what it means, and the standard stops being the thing that gets cut when the clock tightens.
What this looks like in practice
One prospect described the goal precisely, without prompting: they wanted their investment principles document — the real one, the one that governs IC — ingested and applied as a scorecard to every opportunity, so that every analysis came back scored against the same criteria in the same order.
That is the right ask, and it is worth being clear about what it gets you and what it does not.
It gets you comparability. When every first pass is scored against the same framework, the twelfth deal of the year can be read against the first eleven, and a partner can ask why this one scored a 3 on supplier resilience when the last three in the sector scored 4s. That question is unavailable in a world of bespoke memos.
It gets you an audit trail on absence. A scorecard that returns "cannot assess — the room does not support this" is more useful than one that returns a guess, and it makes the gap visible to the reviewer rather than invisible.
It does not get you a decision. The scorecard tells you the same questions were asked. Whether the answers add up to a deal is judgment, it stays with the investment committee, and any vendor suggesting otherwise is selling something you should not buy.
The turnover argument
There is a second reason to write the standard down, and it is the one that tends to land with founders rather than with deal teams.
The associate who leaves takes the version of the standard she was carrying. The partner who leaves takes more. Bain's work on PE knowledge transfer has found that a large majority of firms have no systematic process for moving deal knowledge between investment professionals, which matches what most partners will tell you privately: the firm's actual diligence method has never been written down anywhere, and it is reconstructed from scratch roughly every four years as the bench turns over.
A firm that has written its standard down and has it applied on every deal is not just running more consistent diligence this quarter. It is the only version of the firm that gets better at diligence over a decade, because it is the only version where a lesson learned on deal nine is still being applied on deal forty.
This is the problem HuxleyIQ was built around: taking a firm's own criteria and applying them, in the firm's own order, to every deal that comes through.