← Back to blog
April 16, 2026· 6 min read· Georges

The Fragmentation Tax

Why Growing Ecommerce Brands Can't Make Decisions Fast Enough

The Fragmentation Tax

A CFO asks what customer acquisition cost looks like by channel, after accounting for returns and fulfilment costs.

The data exists. Revenue in Shopify, ad spend across Meta and Google, email attribution in Klaviyo, shipping costs in the fulfilment provider's dashboard, and cost of processing returns in financial systems. Getting an answer means manually exporting from five systems, reconciling customer IDs that don't match across platforms, adjusting for different time zones, accounting for edge cases like partial returns and store credit, and dealing with the fact that Meta uses 7-day attribution windows while Google uses 30-day.

Four hours of work. The answer is still incomplete.

This isn't an edge case. It's the default state for most small and mid-market ecommerce companies. The fragmentation tax isn't the cost of the tools themselves, but the cost of decisions made too slowly or with incomplete data because compiling that data takes too long.

Why Unifying Data Alone Doesn't Solve It

The standard advice: build a data warehouse, hire analysts, unify everything.

Enterprise retailers who've done exactly this still struggle. Industry research shows BI adoption rates remain stuck at 20-30% of employees despite years of investment (BARC/Eckerson Group 2022, Gartner 2025). These aren't broken tools or failed deployments. They're tools that require technical expertise most companies don't have.

Unified data is necessary but not sufficient.

You still need someone who knows SQL to query the warehouse. Someone who understands ecommerce operations, e.g. how store credit flows through systems, why inventory velocity varies by category, how promotional mechanics affect margin calculations. Someone who can connect operational changes to business outcomes across multiple data sources.

Most companies don't have this combination. McKinsey found 77% of companies lack the necessary data talent and skill sets to perform required tasks in mission-critical areas, and the shortage is getting worse as AI adoption drives demand for technical talent. Mid-market companies can't justify £80K-120K per analyst. Enterprise teams that can afford analysts face a different constraint: data scientists spend 60-80% of their time on data preparation—cleaning, organizing, and fixing data issues—rather than analysis (CrowdFlower 2016, ResearchGate).

The average company now runs 101 SaaS applications (Okta 2025). Each generates data. Each uses different definitions. Marketing counts sessions and clicks. Finance counts paying customers. Operations counts SKUs shipped. Context-switching between systems happens dozens of times per day.

Ecommerce operators understand the business. How discounts cascade through margin calculations. Which product categories drive repeat purchases. When seasonal inventory needs to shift. They just can't write SQL or build data pipelines.

The Infrastructure Problem Beneath BI Tools

BI platforms like Tableau and Looker are powerful when you have the right foundation. Most companies don't.

These tools visualise data. They don't automatically integrate it from multiple sources. They don't encode your business logic. They don't make analysis accessible to non-technical operators.

This is why implementations take 6-12 months and require significant ongoing investment, on top of analyst headcount. Someone needs to build the pipelines, maintain the data models, and create new and update dashboards. The technical capacity you were trying to build now goes into simply keeping the system running.

And this is where teams usually get stuck. A dashboard showing conversion rates dropped 8% tells you something changed. It doesn't tell you why, or what to do about it. That requires someone to dig into the data, join tables across platforms, apply business context, isolate variables. The kind of work that creates three-week backlogs when questions route through an analyst / data science team.

When BI investments go unused, it's not because companies chose the wrong tool. It's because the foundation was built for reporting, not decision-making.

What This Costs

Questions that should take minutes take days. Or, decisions get made with incomplete data because complete data takes too long to compile. Teams operate from different numbers pulled at different times using different definitions.

Marketing wants to shift budget but can't get a clean, comparable view across channels. Operations wants to optimise inventory but can't easily connect demand signals to product movement. Finance wants to model scenarios but can't quickly assemble the inputs to test assumptions.

The problem isn't the technology. It's that using it requires technical expertise, manual work, or both. And by the time you get the answer, the decision window has often closed.

Why the Obvious Fixes Fall Short

Most companies turn to one of three solutions: hire data analysts, buy more sophisticated BI platforms, or use LLMs like ChatGPT or Claude. All three make sense at enterprise scale with dedicated teams and substantial budgets, up to a point. But for small to mid-market companies, they create new bottlenecks, new dependencies, and new costs that quickly exceed what the business can support.

In my next post, I'll show why each of these approaches are at best sub-optimal and at worst fail, and why they all miss the underlying infrastructure problem.

Georges

Key Takeaways

  • The fragmentation tax isn't the cost of SaaS tools—it's the cost of decisions made too slowly or with incomplete data

  • Unified data is necessary but insufficient; you still need technical expertise to use it

  • BI adoption rates remain stuck at 20-30% of employees (BARC/Eckerson Group 2022) because the infrastructure beneath them doesn't exist

  • Three gaps create the problem: tool fragmentation (101 SaaS apps), talent shortage (77% of companies lack resources), and the technical barrier between data and operators

  • Standard fixes (hiring analysts, buying BI platforms, using LLMs) assume ongoing investment in technical expertise that small to mid-market companies can't sustain

Frequently Asked Questions

Q: Why don't data warehouses and BI platforms solve this?

A: They're necessary but insufficient. BI tools visualise data, but they assume you've solved integration, can maintain pipelines, and have technical expertise to operate, and get the insights you need from them. Industry research consistently shows BI adoption remains stuck at 20-30% of employees despite years of investment (BARC/Eckerson Group 2022).

Q: Can't we just hire data analysts?

A: 77% of companies lack necessary data talent (McKinsey 2025). Small and mid-market ecommerce companies can't afford teams at £80K-120K+ each. Data scientists spend 60-80% of their time on data preparation rather than analysis (CrowdFlower 2016). And data scientists typically take months to get up to speed on domain knowledge.

Q: What's the difference between BI tools and data infrastructure?

A: BI tools visualise data. Infrastructure handles what comes before: automated data integration, encoded business logic, and interfaces that make analysis accessible to non-technical users. Most companies invest in BI without building what needs to come first.

Q: How do large ecommerce companies handle this?

A: They build custom infrastructure over years with dedicated data teams, automated pipelines, encoded business logic. It's expensive and requires sustained investment, but it works at scale if done right. Small to mid-market companies face the same analytical challenges but can't justify the cost or headcount needed to build and maintain that infrastructure.

Q: Is this really about technology or is it about process?

A: It's about infrastructure. The technology exists, but using it requires either technical expertise (SQL, data modelling, pipeline management) or manual work (exports, reconciliation, spreadsheet analysis). Infrastructure should automate what currently requires expertise or manual effort, making analytics accessible across the organisation rather than bottlenecked through technical specialists.

Sources & Data

Know what to do next

Agenie connects your stack and turns your metrics into decisions. Live in minutes, no SQL required.

No credit card required  ·  Or book a demo