I can ask a strap on my wrist how my sleep debt this week compares to the same week last year. It answers in under a second, and the underlying data is one tap away if I want to argue with it.
Meanwhile, an institutional investor whose commitment runs into the hundreds of millions receives its quarterly reporting as a PDF, forty-five days after the quarter closed. If it wants that same figure broken down by vehicle, it sends an email and waits.
We should be more embarrassed by that gap than we are. We are not, because we have a comfortable explanation ready.
The objection, and why it does not hold
Whenever I make this comparison in a room of administrators, the answer comes back within seconds. Whoop is not audited. Whoop has no co-investors sitting behind confidentiality walls. Nobody signs off on a recovery score. All true, all fair.
But follow the argument to the end. If regulation were the binding constraint, the least regulated corner of a structure would be its most transparent one. It is not. The unregulated SPV three levels down is usually the darkest box in the whole chain: a spreadsheet, an inbox, and someone's memory of what was agreed in 2019.
The real difference is duller, and more damaging to us. Whoop defined its data model once, before it shipped a single device. Every metric it will ever display was specified up front: what it means, how it is derived, at what granularity, with what timestamp attached. We define ours per client, per deal, per vintage, and usually after the fact, in the week before a reporting deadline, by whoever happens to be closest to the file.
That is not a compliance constraint. It is a sequencing choice, and we keep making it in the wrong order.
We customise at the wrong layer
The industry has settled into a debate that I think is framed incorrectly.
GPs and their investors want reporting shaped to their needs: by strategy, by vehicle, with look-through to the asset. At the same time, everybody on both sides wants a leaner and cheaper operating model, and administrators are being pushed hard to automate and standardise. The conclusion usually drawn is that one of the two has to give. Either bespoke reporting survives, or the margin does.
It only looks that way because of where we have put the customisation. Today it sits at the output layer: a hand-built template per client, per investor, per vehicle, maintained manually, re-keyed every quarter, and inherited by whoever survives the last reorganisation. Every new request adds a permanent line of manual cost. Of course that does not scale. It was never going to.
Move the standardisation down to the data layer instead. One definition of commitment, of drawn capital, of recallable distribution, of an entity's position within a structure, applied consistently from the master fund down to the smallest holding company. Do that, and customisation at the output layer becomes close to free. Ten reporting formats over one well-defined dataset is a configuration exercise. One reporting format over ten inconsistent datasets is a permanent reconciliation team.
Standardisation is not the enemy of tailored reporting. It is the precondition for tailored reporting anybody can afford.
Why we have not done it
Three reasons. None of them are technical.
The bespoke work is billable.
An administrator that automates a reporting package it currently charges for on an ad hoc basis is voting to reduce its own revenue this year in exchange for a better margin in three. Very few management teams are measured on that horizon. This is uncomfortable to write, and I would prefer it were not true.
The platforms are heavy.
Migrating a book of business off an established accounting platform is genuinely brutal, and for any individual manager the rational move is to defer it by one more year. Repeat that decision for a decade and you arrive here.
The structure is fragmented, and this one is chronically underestimated.
The fund sits on the fund platform. The SPVs, the holding companies and the corporate entities frequently sit somewhere else entirely: a domiciliation system, a separate corporate secretarial tool, a share register in Excel. "From the top fund down to the smallest vehicle, including corporate" sounds like a modest ambition. It is in fact the hardest sentence in this article, because it crosses three systems that were never designed to speak to one another, and because the timeline of a structure (the incorporations, the transfers, the liquidations) lives almost entirely outside the accounting platform.
ILPA has made this urgent rather than merely interesting
Until recently this was a conversation you could postpone indefinitely. The ILPA templates have removed that option. The updated Reporting Template, covering capital account statements and capital call and distribution notices with integrated fee and expense detail, is to be implemented from the first quarter of 2026. The new Performance Template follows, with the first reports landing in 2027 once funds have completed four full quarters.
What is striking, watching houses prepare, is that building the template is not the hard part. The hard part is the field mapping: reconciling definitions across Lux GAAP, IFRS and US GAAP, and discovering along the way how many places inside a single firm hold a slightly different version of what should be the same number.
The template is doing us a favour. It is an audit of our data definitions, disguised as a reporting requirement.
The advantage of having nothing to migrate
Here is what I expect to become the uncomfortable competitive story of the next few years.
Firms with no legacy hold a structural advantage that is very difficult to buy back later. Not because their technology is more sophisticated, but because they took the steps in the right order: definitions before dashboards, data model before the first client is onboarded. It means they can say yes to a demanding client without that yes costing them anything.
That is the part that gets missed when this is discussed as a technology race. The prize is not a nicer portal. The prize is that the client who wants real oversight (genuine look-through, the full timeline of every entity, figures traceable back to source) becomes a profitable client to serve, rather than an exception the operations team quietly dreads.
The incumbent carrying fifteen years of accumulated bespoke templates has to choose between margin and service. The firm that defined its data properly at the outset never faces the choice.
What direct access actually requires
If a GP is going to have direct access to everything, the honest objection is not confidentiality. It is that direct access means seeing numbers before they have been reviewed. Draft NAV, in the client's hands, on a Tuesday afternoon. Administrators are right to be nervous about that, and "we will give you access once it is final" is precisely how the PDF has survived this long.
The answer is not to withhold the data. It is to ship the status alongside it. Every figure should carry its own state (draft, reviewed, audited) with an as-of timestamp and a traceable source. That is unglamorous metadata work. It is also the entire difference between a portal people trust and a portal that gets ignored in favour of an email to the relationship manager.
And then, AI
Only once that groundwork exists does the interesting part become possible: asking a structure a question in plain language and getting an answer back, with the workings attached. That is the Whoop experience, and it is genuinely within reach for us.
But it is the cherry, not the cake. A language model placed on top of undefined data does not produce uncertainty. It produces fluency. It will answer confidently, in complete sentences, and it will be wrong in ways that are far harder to catch than a wrong PDF. In our industry, a confident wrong NAV is considerably worse than a slow correct one.
The cherry cannot hold the cake up.
So: agree the data dictionary at onboarding rather than in year three. Standardise underneath, customise on top. Ship the status with the number. Do that, and it turns out the two worlds, the standardised one and the tailored one, were never in conflict at all.
We were just building them in the wrong order.