The Analyst Productivity Gap: Why Your Most Expensive People Are Doing the Wrong Work

Forrester put a number on it: 70–80% of analyst time at financial services firms goes to data preparation: pulling it, transforming it, reconciling it, reformatting it, before any analysis can begin. That statistic has been cited in boardrooms and strategy decks for years.

What happens after someone cites it is usually: not much.

The number is well known. The root cause is less discussed. And the firms that have actually closed the gap didn't do it by working faster or hiring more analysts. They did it by automating the workflows that were consuming the time in the first place.

What 70–80% Actually Looks Like in Practice

Abstract percentages are easy to absorb and forget. Here is what 70–80% of analyst time on data prep looks like on the ground.

A research analyst at an investment management firm arrives Monday morning with three reports to complete before the portfolio manager meeting at 10am. The first requires pulling performance data from SEI, blending it with holdings from Bloomberg, and reconciling against the ABOR. The second requires FactSet attribution data formatted against a custom internal benchmark, not Russell 2000, the firm's own composite. The third requires comparing current holdings against a quantitative screening output that lives in a separate system, accessed differently depending on which analyst built the original screen.

None of those steps involve analysis. All of them are necessary before analysis can happen. And all of them are being done manually, by a person who was hired to think, not to move data.

The operations lead at one investment firm described it as being "tethered to the computer" during reporting periods. The technology team described spending all their bandwidth maintaining legacy processes, with zero time for value-add work. Both descriptions capture the same reality from different vantage points.

Where the Time Actually Goes

The data prep problem in financial services has three specific sources, and they're worth naming precisely, because automation addresses them differently.

Fragmented data sources. Most investment management firms access market data, performance data, and reference data from three or more vendors: FactSet, Bloomberg, SEI, MSCI, and others, each delivered in its own format, on its own schedule, with its own quirks. Before data from these sources can be used together, someone reconciles them. That work happens manually, usually in Excel, usually every reporting cycle.

Undocumented transformation logic. Over years, analysts build macros, scripts, and workarounds that translate raw vendor data into the formats the investment process requires. A custom benchmark adjustment. A holdings calculation that accounts for a specific share class. A reconciliation step that exists because two systems define "trade date" differently. This logic is rarely documented. When the person who built it leaves, the firm either loses the logic or spends weeks rebuilding it.

No governed access layer. When analysts can't get data easily from a trusted central source, they go get it themselves: directly from vendor systems, from shared drives, from each other. Each team develops its own approach. The result is what one technology director described as "islands of data": Excel spreadsheets, Access databases, and direct OLTP connections, none of which can prove who accessed what data or produce a consistent answer to the same question.

What Automation Actually Does

Automation doesn't eliminate analysis. It eliminates the steps between having data and being able to use it.

When the data preparation workflow is automated, the 70–80% problem inverts. Reconciled, governed, ready-to-query data is available when the analyst needs it. The reporting that took three hours of manual assembly before the 10am meeting runs automatically and is ready before the analyst logs in. The attribution analysis uses the correct custom benchmark because the logic is encoded in the pipeline, not in someone's head.

A $90B AUM investment firm rebuilt their financial reporting workflows in Snowflake after years of running them through Excel-based processes. Report processing became 100x faster. Real-time analytics on $10M+ in monthly transactions became possible. The gain in analyst time was not incremental: the category of work that consumed most of the day was automated.

A $146B AUM asset manager reduced the time required to fulfill new data item requests from six months to one week, after rebuilding their FactSet data feeds natively in Snowflake. The requests that previously generated months of IT backlog now resolve in days because the underlying data infrastructure handles them.

The Prerequisite That Doesn't Come With the Tool

Analyst time savings from automation are real and measurable. They also require something that most firms don't have before they start: a governed, centralized data layer that automation workflows can depend on.

Alteryx workflows are powerful for research automation. Snowflake provides the performance and access control layer. Fivetran and dbt handle the pipeline and transformation. But none of them solve the fragmented-data problem on their own. If the input data is still living in three vendor systems with no unified access layer, an Alteryx workflow over the top of that produces faster bad data.

The firms that achieve the productivity gains build the foundation first: a single source of truth for market data, performance data, and reference data, with clear ownership and documented transformation logic. Automation then sits on top of something stable.

That foundation is not a multi-year enterprise program. It is a set of specific decisions made once per data domain: what is the authoritative source, how is it accessed, who owns it, and then automated so those decisions don't have to be made again.

The Shift That Doesn't Reverse

When an investment team has experienced self-service access to clean, governed, ready-to-query data, they do not go back to the manual workflow. The shift is immediate and permanent.

Analysts who spent most of their week preparing data start spending most of their week analyzing it. Portfolio managers stop waiting for the report to be built and start asking questions the data can answer today. Technology teams stop maintaining legacy macros and start building capabilities the investment process has wanted for years.

The 70–80% number is not a fact of life in financial services. It is a description of what happens when data preparation is not automated. The firms closing that gap are doing it now, while the competitive window is still open.