Every iGaming operator talks about becoming data-driven, yet inside many organisations there is a second analytical ecosystem operating beneath the official one. It rarely appears in strategy presentations or vendor discussions, but it influences daily decisions across CRM, finance, trading, acquisition and VIP management far more than most executives realise. It lives in exported CSV files, manually adjusted Excel sheets, private Sheets links shared over Slack and “temporary” KPI trackers created years ago that somehow became permanent operational dependencies.
Almost every mature operator has it.
Not because the people inside are incompetent or resistant to technology. The spreadsheet underground usually emerges inside companies that are heavily investing in analytics. The irony is that the more dashboards, reports and KPIs an organisation accumulates without proper governance underneath, the more likely people are to create unofficial versions of reality around themselves.
It almost always starts with good intentions. A CRM manager exports campaign data because the dashboard is missing one segmentation option. Finance builds a reconciliation file to validate payment provider settlements. A trader creates a personal margin tracker because the official reporting refreshes too slowly during peak events. An affiliate manager keeps a separate CPA validation sheet to double-check the numbers before approving invoices. An analyst manually adjusts a metric before a board meeting because “the logic is slightly different this month.”
Each workaround feels small and harmless. Most are created to solve legitimate operational problems. Over time, these isolated workarounds evolve into parallel reporting systems with their own logic, assumptions and corrections. Months later, the organisation finds itself in a situation where finance reports one version of NGR, CRM works with another, acquisition measures FTDs differently again, and product quietly stopped trusting all of them long ago.
At that point, the business technically still has a BI platform. Operationally, it has drifted into shadow analytics.
What makes this especially interesting is that spreadsheet culture is rarely caused by a lack of dashboards. Most operators already have more reporting than their analysts can realistically consume. The real problem is usually fragmentation. Departments build metrics independently, definitions evolve without ownership, dashboards multiply faster than governance structures, and eventually people lose confidence that the “official” numbers truly reflect operational reality. Once that trust begins to disappear, departments naturally start building private safety mechanisms around themselves.
This is the moment when the language inside organisations magically changes. People stop asking “what is the correct number?” and begin asking “which version are you using?” That shift may sound minor, but operationally it changes everything. Analytics stops being a shared decision-making foundation and becomes a negotiation process between competing interpretations of the business.
The cost of this problem is much larger than most operators realise because spreadsheets themselves appear inexpensive. There is no major system outage, no dramatic failure and no obvious crisis. Instead, the cost appears everywhere in small, invisible ways. Analysts spend increasing amounts of time validating numbers instead of interpreting them. Meetings become longer because attendees argue about definitions before discussing actions. Executives request additional exports because they no longer fully trust existing dashboards. Knowledge becomes fragmented across Slack messages, screenshots and undocumented manual adjustments hidden inside personal files.
The result rarely registers as chaos in the dramatic sense. What spreads inside the organisation is something quieter and more dangerous: hesitation.
This is one of the reasons why semantic governance is becoming increasingly important in modern analytics architectures. Metrics and Attributes are no longer just technical BI concepts. They are operational infrastructure that allows organisations to establish one shared analytical language across departments. Without that consistency, even sophisticated dashboards and advanced visualisations eventually become difficult to trust because different parts of the organisation continue operating on slightly different assumptions.
The rise of AI is likely to expose this problem even faster. Many companies currently believe AI will eliminate reporting chaos by making analytics more accessible and conversational. In reality, AI systems inherit the quality of the analytical environment they operate within. If an organisation already has conflicting KPI definitions, duplicated business logic and fragmented ownership of metrics, AI does not solve those inconsistencies. It accelerates them.
This is why so many AI projects disappoint after the PoC stage despite impressive demos. The model itself is rarely the problem. The problem sits underneath, in an organisation that never established a clean semantic foundation in the first place. AI Agents can only reason consistently when the Metrics, Attributes and governance structures beneath them are consistent as well. Otherwise, faster answers simply produce faster confusion.
Perhaps the most dangerous aspect of the spreadsheet underground is that the most problematic spreadsheets are usually the successful ones. The abandoned file nobody opens poses very little risk. The dangerous one is the spreadsheet that quietly became critical to operational workflows without ever formally becoming part of the company’s infrastructure. Once key business logic lives outside governed systems, organisations begin accumulating invisible dependencies around individual employees, undocumented processes and manual interpretations of data. Historical context disappears, onboarding becomes harder and analyst turnover suddenly becomes a strategic risk because nobody fully understands how certain “important numbers” are actually calculated anymore.
Many operators still underestimate how much of their business logic currently exists outside their official platforms. In some organisations, entire operational decisions rely on files stored locally on someone’s laptop or maintained by a single person who has gradually become the unofficial translator between competing KPI realities.
That model becomes increasingly fragile as the industry moves deeper into AI-native operations.
The operators likely to gain the biggest advantage from AI over the next few years will not necessarily be those with the largest data departments or the most sophisticated machine learning models. More likely, the winners will be the organisations that managed to eliminate analytical ambiguity before introducing large-scale automation and AI-driven decision layers. Companies where everyone trusts the same definitions, where governance exists beneath self-service analytics and where operational knowledge lives inside systems rather than scattered across invisible spreadsheet ecosystems.
Because in the end, the future competitive advantage in iGaming may have less to do with who has more data, and far more to do with who finally escaped the spreadsheet underground.
