Data sourced from Dakota Private Markets, the private fund performance platform powered by Dakota. Learn More | Request Access
Most of the conversation about private market benchmarks is about getting them. The data is scattered across dozens of sources, and when you do get it, two managers can report the same metric and both be defensible.
Both problems are real. But there is a third one that survives even after you solve them: the benchmark you finally get your hands on is often answering a different question than the one you asked.
In this article, we'll cover four reasons a published benchmark can be accurate and still be the wrong tool, and what a benchmark has to look like before it is worth acting on.
4 Reasons Benchmarks Fall Short
1. The Peer Group Is Almost Always Too Broad
"Private equity, 2019 vintage" is not a peer group. It is a category.
Inside it sit a $200M lower-middle-market buyout fund doing founder-led deals in the Midwest and a $10B megafund running take-privates. Same asset class, same vintage, almost nothing else in common. Different entry multiples, different leverage, different exit routes, different competitive sets.
A median drawn across both tells the smaller manager very little, and it tells the allocator evaluating them even less. The narrower the cohort, the more the number means, and most published benchmarks stop narrowing several steps before the point where it starts to matter.
2. Survivorship Quietly Flatters the Median
Benchmarks are built from funds that report. Funds that stop reporting leave the sample.
The ones that stop are not random. A manager winding down a disappointing fund has little reason to keep submitting figures, and no obligation to. The result is a benchmark assembled disproportionately from survivors, which pulls the median up and makes the bar look lower than it was.
This is not a reason to ignore benchmarks. It is a reason to treat a median as a soft floor rather than a midpoint, and to ask how a dataset was assembled before reading a quartile breakpoint off it.
Dakota Private Markets lets you build the peer group yourself, filtering 18,000+ funds by vintage year, asset class, sub-strategy, geography, fund size, and portfolio company sector, so the comparison matches the fund in front of you rather than its category. Request access.
3. A Quartile Breakpoint Is a Moving Target
Top quartile sounds like a fixed standing. It is a snapshot of where a line sat on the day the data was pulled.
As more funds in a vintage report, as marks move, and as realizations land, the breakpoints shift. A fund that was top quartile in one quarter can sit in the second the next without anything changing inside its portfolio. Managers quote the best version of this they ever held, and that version is usually the oldest one.
The practical move is to ask when a quartile claim was calculated, and against how many funds. Both answers tell you more than the label does.
4. The Median Hides the Number That Actually Matters
A single midpoint is the least interesting output a benchmark produces. The spread is where the information is.
Across funds tracked in Dakota Private Markets, median TVPI sits within a fairly narrow band by strategy, roughly 1.28x for private credit up to 1.53x for private equity. The dispersion around those midpoints is not narrow at all. Venture capital spans about 0.73x between top and bottom quartile. Private credit spans about 0.35x.
That gap is the real finding. It says manager selection carries roughly twice the consequence in venture that it does in credit, and no median anywhere in that dataset tells you so. A benchmark that reports only the middle has thrown away the part that should shape where a team spends its diligence hours.
What a Usable Benchmark Looks Like
Four tests, and a benchmark has to pass all of them before it is worth acting on.
- It is narrow enough to be a real peer set. Vintage and asset class are the starting filters, not the finishing ones. Sub-strategy, geography, fund size, and ideally portfolio company sector are what turn a category into a cohort.
- You know how the sample was built. Which funds are in it, how many, and whether records are verified or self-submitted.
- It is current, and you know as of when. A benchmark without an as-of date is a number without a shelf life.
- It shows the distribution, not just the midpoint. Quartiles at minimum. The spread is the output that changes decisions.
Most published benchmarks pass one or two of these. The ones worth building a process around pass all four.
Build the Benchmark, Don't Borrow One
The fix for a benchmark that falls short is not a better published number. It is the ability to construct the cohort yourself.
Dakota Private Markets holds 18,000+ private funds and 159,000+ performance records across seven asset classes, with Net IRR, TVPI, DPI, and RVPI on every record and every record reviewed by Dakota's research team before publication. Filter to the peer set that actually applies, see the full distribution rather than a midpoint, and export it to Excel or CSV for the diligence file.
Request access to Dakota Private Markets.
