The starting position
The asset manager had a recognisable problem. AUM had grown 60% over five years. Operational headcount had grown almost in lockstep. Fee compression was running steeper than cost reduction. The COO had spent two years trying to break the linkage between AUM growth and operational headcount through automation and vendor platform upgrades, with modest results. The board was patient but the patience was running out.
The specific friction was concentrated in the middle and back office. NAV production ran on a nightly batch cycle that involved three vendor platforms (one for fund accounting, one for portfolio management, one for performance attribution), each with their own data models and reconciliation logic. Failures were common enough that the operations team had a permanent "exception list" that grew through the day and was reconciled overnight. End-of-month close took 11 days. ESG reporting under SFDR had become a quarterly fire drill that consumed disproportionate management attention.
The COO sponsored the engagement. The Chief Operating Officer's goal was to break the AUM-cost linkage structurally rather than chip away at it incrementally.
What we found in the diagnostic
Six weeks of portfolio audit and current-state mapping confirmed the structural picture. The asset manager had invested significantly in front-office quant capability — sophisticated models, strong data scientists, market-leading research stack. None of that sophistication had reached the middle or back office. The operational backbone was running on workflows designed for an era of batch processing and human exception handling.
The key insight was that the constraint was not a vendor problem. The vendors were fine — Aladdin, SimCorp, and the custodian platforms were all delivering what they were designed to deliver. The constraint was that the operating model around the vendors was still treating their outputs as inputs to a human-driven workflow, instead of treating them as feeds into an action data layer that could power continuous decision-making.
We also found significant existing infrastructure to build on. The data team had already built parts of a normalised reporting layer, primarily for management reporting. It wasn't designed for action — definitions weren't enforced at capture, freshness was unpredictable, lineage existed only on paper — but it was a meaningful starting point that meant the data layer rebuild could be incremental rather than greenfield.
What we redesigned
The blueprint phase produced a target-state operating model for NAV production built around the five enablement pillars:
Production function: NAV workflow rebuilt in BPMN 2.0 around the principle of continuous, exception-driven operation. Prices, positions, corporate actions, and external feeds flow into the system as they arrive. Anomaly detection runs continuously rather than at end-of-day. Exceptions are surfaced with full context to operations, who handle them in flow rather than at a batch reconciliation. The end-of-day "NAV strike" became a confirmation rather than a calculation event.
Data layer: A normalised action layer built around the position-level wide-row, joining holdings, valuations, corporate action status, fund admin reconciliation, and external feeds in real time. Captured at the point where the source systems updated rather than reconstructed from batch extracts. Definitions enforced at capture time. Lineage observable in production for any value flowing into the NAV calculation. Designed specifically to be reusable by adjacent workflows — performance attribution, regulatory reporting, ESG operations.
Decision rights: A formal decision rights matrix mapping which exceptions the system resolves automatically (with full audit), which escalate to operations, which escalate to the fund accountant or external administrator, and which trigger fund manager or compliance review. Material thresholds preserved (UCITS and AIFMD requirements don't move) but the routine exceptions handled automatically at high velocity.
Embedded governance: A second-line risk specialist embedded with the team alongside the existing depository review function. Decision logs, lineage, and model risk file produced as a by-product of build. The depository conversation was constructive throughout — they had been asking for exactly this kind of observability for years.
Feedback flywheel: Every exception resolution captured as structured feedback. Anomaly detection model retrains weekly on operational data. Operations exception rate is monitored as a leading indicator of model drift or upstream system changes.
How the delivery ran
Phase 2 delivered the blueprint in 10 weeks. Phase 3 (activation) ran for 8 months.
The first three months were data layer construction and integration with the three vendor platforms. The work was unglamorous but the operational pattern depended on it. We treated vendor integration as part of the data architecture rather than a separate workstream and the team accepted that this was the binding constraint.
By month 4 the new workflow was processing live data alongside the legacy batch process. By month 6 the batch process had been wound down for routine cases. By month 8 the new workflow was the primary operating pattern and the legacy infrastructure had been decommissioned.
The operations team's role changed visibly. The team is the same size as it was 12 months ago — but managing roughly 40% more fund volume. The role profile is different (more comfortable with system telemetry, more analytical, less procedural) and the hiring profile for new operations staff has been updated accordingly.
What the outcomes look like
Eight months after activation:
- 94% of NAV exceptions resolved before market open through the continuous workflow, compared with approximately 60% under the legacy batch model.
- Month-end close compressed from 11 days to 5. The team no longer treats month-end as a sprint; the work is distributed across the cycle.
- The data layer built for NAV production is now feeding 3 adjacent workflows — performance attribution, AIFMD/UCITS regulatory reporting, and the early stages of an ESG operations rebuild. Each new workflow inherits the foundation rather than rebuilding it. This is the compounding effect we promised at the start of the engagement and it is what makes the second project cheaper than the first.
- Operating cost-per-fund dropped 40% on a run-rate basis. Combined with the 40% volume growth handled by the same team, the structural cost-to-serve change is substantial.
- Depository and external administrator dialogue is now data-driven rather than report-driven. Reconciliation breaks are surfaced and resolved in real time rather than at the end of cycle.
Why this was harder than it sounds
Three challenges defined this engagement:
The vendor density was the structural problem. Three vendor platforms with three data models meant the data layer rebuild had to navigate the vendor landscape rather than pretend it didn't exist. We treated vendor integration as part of the architecture work rather than a separate problem and that decision was the difference between a programme that landed and one that would have stalled at month 6.
The front office's AI maturity created a misleading sense of overall maturity. The quant team was sophisticated and the executive team's intuition about "what AI can do" was shaped by the front office. The operational backbone was structurally less mature, and getting the executive team to invest in operational AI work — which is less glamorous than front-office quant — required honest framing about where the actual cost base sat.
Fee compression was the binding commercial constraint. It was also the strongest case for the work. The asset manager's margins were under structural pressure that was not going to reverse. The cost savings from this engagement bought genuine breathing room and made subsequent transformation work easier to fund. We named this in the diagnostic and it shaped how the executive team prioritised the work.
What changed for the asset manager
The cost outcomes are real and they matter. The deeper change is the platform: the asset manager now has a reusable action data layer that is feeding adjacent workflows and will continue to feed new ones for years. The next engagement (regulatory reporting redesign) will start halfway built. The one after that (ESG operations) will start further along still.
This is the compounding case in action. The first project is the proof. Each subsequent project costs less than the one before and lands faster. Two years from now, this asset manager will be operating at structural cost-to-AUM levels that competitors who have not done the work will struggle to match.
That is the case for starting now.