The hardest question in any automation programme is not which tool to use — it is where to start. Most teams answer it by instinct, automating the workflow that annoys them most or the one a vendor demoed best. For a FINMA-regulated institution that instinct is expensive, because the first workflow you automate sets the template for everything that follows: if it is provable, measured, and auditable, it earns the mandate for the next one; if it stalls at the first compliance question, it poisons the whole programme. The discipline that avoids this is unglamorous — a simple 2x2 of impact against effort — but the axes have to mean the right things for a regulated institution.
Start with impact, and resist the temptation to define it as cost saved. In a fund management company or asset manager, the impact of automating a workflow has three parts: the senior time reclaimed from manual reconciliation and transcription, the operational risk removed when fewer manual touches sit between the data and the investor, and the audit defensibility of a process that runs on a schedule and reconstructs itself after the fact. A workflow that scores high on all three is worth far more than the labour line alone suggests. Effort is the mirror image: not engineering difficulty, but data readiness, how stable and well-understood the process already is, the weight of the review layer it needs, and how much compliance sign-off it will demand before it can go live.
The top-left quadrant — high impact, low effort — is where you begin, every time. These are the recurring operational workflows that already have a stable shape and a baseline you can measure today: periodic reporting, reconciliation, KYC and document extraction, the exception-handling around a calculation. The data already exists, the process is well understood, and because the baseline is measurable, the improvement is provable rather than asserted. This is the workflow that earns the programme its credibility, so it should be the one you ship first — not the most ambitious idea on the roadmap, but the one whose return you can put in front of finance and compliance without a caveat.
The top-right quadrant — high impact, high effort — is real, but it is a second or third move, not a first. Client- and investor-facing automation lives here: drafting investor communications, servicing enquiries, surfacing portfolio narratives. The impact is genuine, but the effort is heavy — the review layer is thicker because the reputational stakes are higher, adoption is gradual because trust is earned rather than switched on, and the return is harder to isolate. Sequence these deliberately, after a low-effort win has proven the approach and funded the patience they require. Attempting them first is how programmes acquire a stalled flagship and a nervous board.
The bottom half of the matrix is about discipline, not opportunity. Low-impact, low-effort workflows — internal scheduling, meeting summaries, knowledge search — are fine to pick up, but they must never be the headline; a programme that leads with them looks busy and proves nothing. The genuine trap is the bottom-right: high effort for low impact. A bespoke build for a workflow that runs a handful of times a year, or a novel capability that touches no measurable baseline, will consume attention and budget and return neither. In regulated finance the same square hides a second cost — every high-effort build is also a larger audit surface, so low-impact ones are doubly not worth it.
The worked examples follow the same logic across institution types. For a fund management company, the first-automation winner is almost always factsheet and performance reporting — high volume, stable template, measurable cycle time. When reporting across 39+ funds at a leading Zurich investment foundation was automated, that was precisely the entry point: the compressed cycle paid for itself in reclaimed senior time long before anything more ambitious was attempted. For an asset manager the winner is typically portfolio commentary and reconciliation; for an investment foundation it is the recurring regulatory and investor reporting that arrives on a fixed calendar. In each case the same quadrant wins — high impact, low effort, provable baseline — and the institution-specific detail only changes which workflow sits in it.
Which leaves the one rule that overrides the matrix: never automate a workflow whose baseline you cannot measure. Without a baseline you cannot prove the return, you cannot detect a regression, and you cannot answer the first question finance or an internal auditor will ask. A workflow that looks like a top-left winner but has no measurable starting point belongs back in discovery until it has one. The framework is only a way of enforcing a single habit — start where the impact is high and the baseline is provable, let that workflow earn the mandate, and let the proven case fund the next one. That is the same discipline that runs through everything we build.