Automating an ESG asset manager's annual portfolio investment review

Automating an ESG asset manager's annual portfolio investment review

An ESG-focused asset manager overseeing 10 billion dollars in assets has turned a two-week, error-prone annual ritual into a single day's work. QuantSpark replaced its manual portfolio investment review with an automated data pipeline and an interactive dashboard, cutting the review from 40 FTE days to one, a saving of 39 FTE days every year, while opening the door to monthly rather than annual reporting.

The problem

Before the engagement, the asset manager's investment review depended on analysts manually pulling data from disparate market-data and custodian systems, copying it into spreadsheet templates, running quality checks by hand, and assembling static slide decks. Each cycle took around two weeks and pulled in several team members, embedding key-person risk into a process that should have been routine.

Manual re-keying introduced errors, and the review's fixed, static format meant stakeholders received a snapshot they could not interrogate: no drilling into a number, no filtering by strategy, no re-cutting the attribution view without going back to the analysts and waiting for the next cycle. Because the review consumed so much specialist time, the firm could only justify running it once a year, even though the underlying portfolio data changed far more often than that.

What QuantSpark built

A two-person QuantSpark team spent four months building an automated data pipeline in Python and SQL that extracts and transforms data directly from the market-data and custodian systems, removing the manual copy-and-paste step entirely. That pipeline feeds an interactive business-intelligence dashboard of around 100 visuals, supporting drill-downs, slicers and filters, plus multi-level performance and attribution analysis.

Delivery followed an iterative pattern: build a piece of the pipeline and dashboard, test it, take it back to stakeholders for feedback, and refine, rather than attempting one long build-and-hand-over. That cycle both caught data-quality issues early and meant the eventual dashboard matched how the investment team actually wanted to interrogate performance, not how QuantSpark assumed they would.

How it worked end to end

Data starts in the asset manager's market-data and custodian systems. The Python/SQL pipeline pulls and reconciles it automatically, on a schedule, rather than waiting for an analyst to do it by hand. That clean, structured data lands in the BI dashboard, where stakeholders can drill into any number, slice by strategy or holding, and run multi-level attribution analysis themselves, on demand, rather than requesting a new slide from the team. The same pipeline and dashboard now also underpin a monthly reporting cadence, not just the annual review, because the marginal cost of producing an update is close to zero once the automation is in place.

Systems involved

The build combined the client's existing market-data and custodian systems as data sources, Python and SQL as the extraction and transformation layer, and a business-intelligence dashboard as the interactive front end. No specific vendor products are named in the public record; the description is deliberately kept at the category level to reflect what has been publicly disclosed.

The value delivered

The headline, measured result is the FTE-day saving: the annual review deck fell from 40 FTE days to one, a delivered reduction of 39 FTE days each year, not a modelled projection. That freed-up capacity, combined with the move from annual to monthly reporting, gave the investment team current portfolio insight far more often than before, using the same or less specialist time.

The interactive dashboard also improved data accuracy relative to the old manual re-keying process and reduced the firm's dependence on a small number of specialist analysts to produce the review at all. Converting the 39 FTE days into a pound figure would require the client's own analyst day rates, which are not part of the public record, so no monetary saving is claimed here: the FTE-day figure is the delivered, defensible measure.

What is not yet public

The public case study does not name the client, the specific BI platform used, the number of portfolios or systems integrated, or any usage statistics from the dashboard since go-live. Those gaps are flagged rather than filled, in line with keeping this story to what has been evidenced.