Perspectives · Data Governance
Why Generic Data Tools Fail Investment Teams, and the Fix
Generic cloud data tools don't understand investment data out of the box. Why asset managers hit that gap, and what a purpose-built data model changes.
· 13 min read
The Promise and the Letdown
You bought the modern cloud data warehouse, connected your BI tool, and your team can now run SQL against billions of rows in seconds. The demo was impressive, and the migration felt like progress. Then someone asks a simple question: "What's our exposure to this issuer across all funds and accounts?"
The query comes back wrong. Not because the SQL is broken, but because the platform has no concept of what an issuer is, how positions aggregate across tax lots and custodians, or how a corporate action last week changed the instrument you thought you were holding. Your analyst spends the next two days reconciling spreadsheets to get the right answer.
This is the gap that mid-market asset managers keep running into. The tools are powerful, but they are generic, and investment data is not.
For a long time data infrastructure was about storage and speed. Get everything into one warehouse, make queries fast, and call it done. In 2026, it’s become clear how important meaning is. Can your platform actually understand what a position is, how a fund hierarchy rolls up, what "as of today" means when a corporate action is pending? If it can't, you haven't transformed your data into action. You've just built a faster way to get the wrong answer.
For mid-market institutional asset managers, this gap is especially painful. You don't have a team of data engineers to build a custom domain model from scratch or a Chief Data Officer establishing enterprise-wide governance. You have a lean operations team, a few analysts, and growing volumes of data from custodians, administrators, and vendors that need to make sense before they can drive decisions.
The core problem is generic tools don't understand investment data. The gap between "powerful platform" and "investment-aware platform" is where mid-market firms lose time, accuracy, and alpha.
What "Understanding Investment Data" Actually Means
To see why generic tools fall short, it helps to be specific about what investment data actually requires.
A generic cloud data platform is, at its core, an engine for storing and querying data. It handles tables, columns, joins, and aggregations extremely well. What it doesn't come with is a business data model built in. That still has to be built by the customer, along with the semantics that make the data meaningful1.
Investment data has a conceptual structure that is remarkably stable. Portfolios, positions, trades, securities, and a handful of other core entities have formed the backbone of the industry for decades. In fact, there are only around ten fundamental entities that matter, and that number has barely changed in half a century1. These entities have relationships, rules, and behaviors that any practitioner takes for granted:
Instruments have identifiers that change over time (CUSIPs, ISINs, Bloomberg tickers), and a single issuer can have dozens of them.
Positions aggregate across tax lots, accounts, and custodians, and they carry bitemporal meaning: what you held on a given date versus what you know today about what you held on that date.
Corporate actions (splits, mergers, dividends, name changes) retroactively change what a position means.
Fund hierarchies determine how exposure rolls up from holdings to sub-funds to master funds.
Valuations depend on pricing sources, settlement conventions, and accounting rules that vary by asset class.
A generic platform treats all of this as rows in tables. It doesn't know that a corporate action should retroactively adjust a position's cost basis. It doesn't know that "as of today" has a different meaning when a dividend is pending. It doesn't know that two records from different custodians refer to the same holding.
This is why the standard AI playbook also fails when pointed at investment data. Gartner found that 63% of organizations either do not have or are unsure if they have the right data management practices for AI, and predicts that through 2026, organizations will abandon 60% of AI projects unsupported by AI-ready data2. The issue isn't the model. It's that the data underneath lacks the semantic structure to make AI answers trustworthy.
Gartner makes the same point at the architectural level. Their research argues that agentic AI efforts will fail without semantic and structural data modeling. It is essential to implement both data modeling, to describe the kinds of information that are important, and semantic modeling, to capture and communicate business meaning. Combining the two into a semantically enriched data layer provides the foundation for resilient AI and agentic readiness3.
In other words, the problem isn't the compute engine; the problem is the missing layer between the engine and the business meaning of investment data.
The Real-World Cost for Mid-Market Asset Managers
Research shows that the consequences of this gap show up in the daily work of mid-market asset managers. Alpha FMC's research on asset management data strategies found that most firms describe themselves as "Getting Organised" on the data maturity curve, and none described themselves as a "Data Innovator." The consultancy notes that mid-sized firms are possibly the most complex in relation to data responsibility: none have filled a Chief Data Officer position, and almost half are still siloed by business function4. For a mid-market firm, this means data ownership is fragmented across teams with no single accountable party, and each team encodes its own definitions into its own tools.
Alpha FMC's broader research on the impact of AI on asset manager IT architecture reinforces the point. Their industry operational research indicates that data fragmentation is the leading challenge limiting front-office effectiveness, with firms' data teams spending the majority of their time on operational tasks rather than analysis5. Industry research suggests that 70 to 80% of enterprise AI initiatives fail to deliver meaningful business value, and in financial services these rates are often worse due to data fragmentation, regulatory requirements, and legacy architecture5.
Citisoft's 2025 Transformation Agenda survey, which gathered insights from nearly 70 asset managers, asset owners, and insurers, found that smaller firms under $100 billion AUM are pursuing transformation across data, risk, and performance but face structural blockers: fragmented systems, spreadsheet reliance, and limited internal expertise are slowing execution. Up to 40% of these firms augment vendor solutions with manual tools6. Their 2026 survey adds that few firms operate through a single data delivery model, and nearly half identify data quality and accuracy as their biggest data and technology challenge. The challenge is not only collecting or storing data but creating a connected, governed, and scalable data environment7.
For a mid-market asset manager, this plays out in concrete ways. Your operations team spends days pulling data from multiple systems and reconciling across asset classes. Analysts export portfolio data, reconcile figures offline, and manually stitch together factsheets, regulatory filings, and investor letters. When definitions drift between teams, the same metric can show different values in different reports, each internally correct but collectively inconsistent. Accuracy, auditability, and alpha all suffer.
Alpha FMC's operations survey confirms that data quality, shortages of AI-skilled talent, and integration challenges are the primary barriers to deeper AI adoption across firms of all sizes, with smaller firms remaining more constrained8. The irony is that the tools meant to solve this problem often make it worse. A generic data warehouse centralizes the data but doesn't centralize the meaning. You've moved from five disconnected spreadsheets to one connected warehouse that still doesn't know what a position is.
A generic data warehouse centralizes the data but doesn't centralize the meaning.
Makeshift Fixes Teams Try Today
If you're running a mid-market asset management firm and dealing with this right now, there are practical steps you can take without buying a new platform. These are genuine improvements. They reduce friction, cut down on definition drift, and make your existing tools more useful. But it's important to be honest about what they can and can't do.
1. Build a Lightweight Internal Data Dictionary
Start by documenting what your key metrics mean: AUM, exposure, NAV, committed capital, called capital, unrealized G/L. Get the definitions in writing, agree on them across teams, and make the document accessible. This sounds basic, but the research consistently shows that definition drift is one of the most common sources of reporting errors. When two teams use different definitions of the same metric, each report is internally correct but collectively inconsistent.
Citisoft's research underscores why this matters: nearly 90% of firms plan to overhaul their data operations and governance by 2027, recognizing that clean, governed, and integrated data is essential for agility and intelligence6. A data dictionary is a small but meaningful first step toward that governance.
A data dictionary won't make your platform compute these metrics correctly. But it will reduce the arguments about what the numbers should be, and it gives new team members a reference point instead of tribal knowledge.
2. Create a Simple Semantic Layer in Your BI Tool
Most modern BI tools support some form of semantic modeling: defined metrics, calculated fields, and relationships between tables. If you're using a tool like Sigma, Looker, or Power BI, you can define your key investment metrics once in a semantic layer so that every dashboard and report uses the same calculation.
This is a real improvement over ad hoc SQL. It means when someone asks for "net exposure," the tool applies the same formula every time instead of letting each analyst write their own. But it's a thin layer. It defines metrics; it doesn't model the underlying investment domain. It doesn't know that a corporate action should retroactively adjust a cost basis, or that positions need bitemporal handling. You're encoding the tip of the iceberg, not the foundation.
Alpha FMC describes the semantic layer as an abstraction that unifies diverse data sources by embedding meaning through ontologies and knowledge graphs, so that every downstream consumer works from the same governed definitions9. A BI tool's semantic layer is a small piece of that vision, but it doesn't deliver the full domain model.
3. Assign a Single Data Owner Per Domain
Even without a formal Chief Data Officer, you can designate one person as the owner of each data domain: reference data, positions, valuations, performance. Their job isn't to build a platform. It's to be the final word on definitions, track data quality issues, and ensure consistency within their domain. Alpha FMC's research shows that having a single party responsible for data helps avoid silos and can drive enterprise-wide value from data initiatives4.
This is a governance fix, not a technology fix. It prevents some errors but doesn't build the underlying model.
4. Document Your Bitemporal Requirements
Investment data is bitemporal: the date a transaction occurred and the date you learned about it are often different. A trade executed on Tuesday might not appear in your custodian file until Thursday. A corporate action with an ex-date last month might arrive in today's feed. If your team doesn't explicitly track both timelines, reporting will be wrong in ways that are hard to detect.
Documenting these requirements won't make your platform handle them automatically. But it will help your team build workarounds (like effective-dated tables or "as of" snapshots) and catch errors before they reach investor reports.
The Honest Assessment
These four fixes are worth doing. They buy you time, reduce friction, and improve consistency, but they are stopgaps, not solutions. The root cause remains that your platform has no investment-specific data model. To fully solve this on a generic cloud platform like Snowflake, you need to build that domain model yourself. Alpha FMC notes that leading firms are solving this through the creation or re-architecture of enterprise investment data platforms that centralize data ingestion, validation, and enrichment across the investment lifecycle, a significant undertaking that requires establishing data governance frameworks, building a common investment taxonomy, and layering analytics tools10. A data dictionary, a BI semantic layer, and a data owner don't change that. They make the gap more manageable, but the gap is still there.
For a mid-market firm with a lean team, that build is usually not feasible. You'd be spending engineering resources on constructing something that already exists in purpose-built platforms.
The Missing Layer: Investment-Specific Data Models
What's needed is a domain-specific data model. Not a generic schema adapted for finance, but one purpose-built for investment data from the ground up.
Alpha FMC makes the case clearly: for most asset managers and owners, fragmented systems and inconsistent data standards present a major hurdle. Data centralization alone is not the goal; achieving consistent, actionable insights to support efficient decision-making is10. The data model isn't an afterthought; it's the foundation. Without a coherent data model, fund managers simply create more silos1. With one, every downstream system consumes the same definitions, the same relationships, and the same business rules.
Alpha FMC describes the semantic layer as the missing link in asset and wealth management: many firms have modernized data management, integration, quality, and governance, yet the missing piece is data connectedness via a semantic layer that infuses raw data with meaning, relationships, and interpretability9. This is the layer that generic platforms don't provide. It's also the layer that mid-market firms can't realistically build themselves.
How Aquata Solves This
This is where a purpose-built investment data platform changes the equation. Aquata is designed specifically for the investment industry. It's not a generic data platform with financial features bolted on. It's a domain-specific data platform built around investment data from the start.
A central piece of this is Aquata's OpenSpec layer. OpenSpec provides investment-specific, vendor-agnostic data models organized into business domains like Reference Data and Treasury. These models are purpose-built for investment data rather than adapted from a generic schema. That means the platform already understands what an instrument is, how positions aggregate, how fund hierarchies roll up, and how corporate actions affect valuations. Your team doesn't have to build that understanding from scratch or maintain it as the market evolves.
For a mid-market asset manager, this is the difference between spending engineering resources on building a domain model and spending that same time on actually using your data to drive decisions. Aquata consolidates all investment data into a unified data layer, connecting data from every custodian, administrator, and vendor into a single source of truth. Your team stops chasing, collating, and cleaning data, and is equipped to unlock its value.
With Aquata, you can make all your investment data accessible, actionable, and AI-ready. For a firm that has been making do with generic tools and makeshift fixes, this is the natural conclusion to a problem you've been working around for too long. Aquata transforms fragmented data into unified, governed data that you can act on, without the overhead of building a domain model yourself.
Sources
- 1.There Is A Blind Spot In The Modern Data Stack That Will Kill The Buy-Side's AI Ambitions, A-Team Insight a-teaminsight.com.
- 2.Lack of AI-Ready Data Puts AI Projects at Risk (press release, February 2025), Gartner gartner.com.
- 3.Agentic AI Efforts Will Fail Without Semantic and Structural Data Modeling (March 2026), Gartner gartner.com.
- 4.From Manufacturing to Mining: Asset Management Data Strategies Reach for 'Knowledge Alpha', Alpha FMC alphafmc.com.
- 5.The Impact of AI on Asset Manager & Asset Owner IT Architecture, Alpha FMC alphafmc.com.
- 6.2025 Asset Management Transformation Agenda, Citisoft citisoft.com.
- 7.2026 Asset Management Transformation Agenda, Citisoft citisoft.com.
- 8.Key Trends Shaping Asset Management Operations, Alpha FMC alphafmc.com.
- 9.Unlocking Investment Insights: Build First, Consume Better with Semantic Layers, Alpha FMC alphafmc.com.
- 10.Unlocking the Power of Total Portfolio: Unified Data for Asset Managers, Alpha FMC alphafmc.com.
