Preqin~3 min read

The form was never the reason fund managers stopped sharing.

Three weeks, one PM, one engineer, the head of engineering, and Legal and Sales on opposite sides.

  • Role: Product strategy and design
  • Timeline: 3 weeks
  • Team: 1 PM, 1 engineer, head of engineering, me
  • Impact: 5% → 15% retention
  • Platform: Web
0%

Account retention

up from 5%

0mo

Due diligence

down from 10 months

Both held through 2021, when I left Preqin. Retention comes from a two-week test with 20% of traffic diverted to the experiment and the rest left on the existing form, run alongside usability sessions on the same build. Due diligence comes from shadowing partners through a cycle, with deal closure for that same 20% read off an internal platform. Retention has a control group. Diligence is a before-and-after with none. Still missing: what retention counts (accounts, funds, or seats), and how many deals closed in that 20%.

Overview

Fund managers stopped contributing performance data, and the three easiest fixes aimed at a problem nobody had.

Preqin sells access to data that fund managers supply for free. When they stopped contributing, every subscriber got less for the same money. Six options were on the table. The three cheapest to build all made submission easier.

Worked with: 1 PM, 1 engineer, the head of engineering, Legal, and Sales; 6 partners interviewed.

What I owned

I owned

  • Reframing the problem from collection to exchange, and the argument that took the three cheapest options off the table
  • Research with partners at large and mid-market private equity funds
  • Both experiments: the leaderboard that backfired and the quartile ranking that replaced it
  • Keeping the form to what someone could complete in one sitting, despite a push for 32 fields

Decided with others

  • Getting Legal and Sales to the same answer: I brokered it; they made the call
  • The cohort floor beneath anonymization, with the head of engineering

Research

Partners were withholding numbers they were embarrassed by.

I interviewed 6 partners at large and mid-market private equity funds. The second question below changed the brief: what does a fund manager get for sharing a number that might embarrass them?

Decision: exchange, not collection

Winning that argument cost the cheapest roadmap.

If the cost of contributing is effort, tooling helps. If it is exposure, tooling speeds up submission for people who were never going to submit. Every research signal pointed at exposure.

The cheaper builds needed nobody to agree on what a contributor gets back. Moving to exchange meant asking for engineering time against a benefit I could only argue for.

Decision: the leaderboard

My first answer was a named leaderboard, and it backfired.

I considered a public leaderboard, a private benchmark, and no comparison. I was confident enough to ship a named ranking and did not plan mitigation, though the product was about to publish what partners guarded most closely. A partner said: “This doesn’t reflect who we are.” I removed it.

The research had already predicted this. I shipped it anyway.

The leaderboard as built. A bar chart titled Profile views by LPs in the past year ranks fifteen named firms from Norfund down to Xen Technologies, above a peer group panel listing named funds by how much performance data each has contributed, and a Submit your data button beneath.
What partners actually saw: their own position, named, beside everyone else's. The prompt to contribute sits at the bottom of the page it undermined.

Redesign

The version that worked kept the comparison and removed the identity.

Legal would not accept anything identifiable. Sales would not accept something partners could not compare with. I proposed an anonymized quartile position in an inline field, and both said yes. The timeline went from five sprints to two.

Anonymity holds only if the peer set is large enough to hide in. So the product requires four funds to share a benchmark and eight to share a median.

The Evolution of Quartile Ranking chart: ten stacked columns, one per reporting quarter, each split into first through fourth quartile bands with a marker showing where this fund's net multiple falls against its peers. No peer is named anywhere on the chart.
The screen from the cover, flat and readable. There is no name anywhere on it. That is the entire difference between this and the leaderboard, and the only thing that made the comparison something Legal would sign.
The custom benchmark builder. A panel headed Welcome to your custom benchmark creation states three rules: a minimum of four funds is needed to share the benchmark, with eight funds the median can be shared, and the submitter's own fund data is required. Below it, an empty Your fund row above a Peers and competitors table.
The cohort floor, written into the product rather than left to policy: four funds to share anything, eight before a median. It is the control that makes anonymity hold.

Scope

A short inline form shipped, and the 32 requested fields did not.

Every field had an internal advocate. I refused them because of the cumulative cost, and limited the form to what someone could finish in one sitting. The data model stayed thinner than the business wanted. We never tested the 32-field version, so this rests on reasoning.

The benchmark table with a single editable cell. The top row reads Your firm and Your fund with one focused, empty Called percentage input; ten named peer funds sit underneath it with their figures already filled in.
What survived the thirty-two: one field, in the row it will be compared against.

Results

Retention rose from 5% to 15% and diligence cycles shortened from ten months to four.

I do not know what the leaderboard cost

The retention figure cannot separate recovery from damage. Did the later result overcome the harm, or cover it up? Getting the diagnosis right did not stop me getting the first remedy wrong.

Sam Cusano