Growth and CRM often look at the same market from different angles.
CRM sees historical player value, bonus efficiency and signs of abuse. Growth sees future opportunity, including markets that may look weak simply because they have never been tested properly.
This tension becomes stronger in crypto-only iGaming. Smaller markets often have limited data, while poor-quality traffic and bonus abuse can appear quickly.
I used AI to combine the data and develop the decision logic for a framework that gave Growth room to test while making the risk rules clear and consistent.
Starting with data I could trust
The project began with 74 country records from a Rest of World CRM export, containing aggregated country-level user data such as GGR, deposit amounts, deposit frequency and abuse metrics.
A written analysis already existed, but it described 102 countries and different market rankings. I checked the source file directly and rebuilt the analysis from the underlying data.
AI was used throughout the project to bring the information and business rules together. It compared the country records, identified inconsistencies and combined CRM results with marketing priorities, compliance requirements and abuse signals.
I then used AI to test different segmentation approaches and refine the logic through several iterations. I guided the commercial principles, challenged classifications that did not make business sense and checked the results against the source data.
Every final recommendation remained connected to the original data and a visible business rule.
The CRM data provided useful historical signals about commercial value and user quality. However, some country samples were small or unstable. A weak result could indicate a poor market, limited activity or a market that had never received enough attention to demonstrate its potential.
Separating historical evidence from future decisions
I kept the original CRM analysis as a locked baseline and created a separate working view for Growth and Marketing.
The historical record stays intact, while the working view supports forward-looking decisions. Each country record shows the CRM signal, marketing priority, compliance status, risk-watch level, recommended segment and the reasoning behind it.
AI helped connect these different signals and apply the same decision logic across every country. It also produced the supporting documentation explaining how the model works and how each classification should be interpreted.
The final view was deliberately kept compact. Managers can understand the position of a market quickly and return to the full CRM data when more detail is needed.
Changing the market logic
The first version was too conservative.
Markets with limited CRM history or no current marketing focus could fall into the lower segment even when their wider market conditions were attractive. This gave historical performance too much control over future testing.
I worked through the logic with AI and revised the model so that allowed and structurally attractive markets could be classified as test-eligible, subject to risk watch and future review.
In the final framework, ROW-H means test-eligible and prioritised. Profitability is established later through live results.
A separate Risk Watch gives each market a low, medium or high monitoring level. It is intended to help Revenue and CRM move markets down when agreed traffic-quality or abuse signals deteriorate. Compliance restrictions remain a firm boundary regardless of commercial potential.
The operating principle is simple: give promising allowed markets a fair test, collect new evidence and respond quickly when that evidence changes.
The practical result
The finished model combines three elements: the locked CRM baseline, a compact market-decision view and a clear guide explaining how the framework should be used. It was delivered as both a working spreadsheet and a shareable Google Sheets version.
The final segmentation classified 52 markets as ROW-H, 20 as ROW-L and two as excluded, covering all 74 country records.
Its main value is the shared operating language it creates.
Growth can see where controlled testing is justified. CRM can see how historical quality affects the decision. Revenue has a defined direction for future market reviews, while compliance requirements remain visible throughout the process.
The underlying evidence remains intact, the reasoning behind each decision is visible and market classifications can evolve as better information becomes available.
How much more value can it create?
The next opportunity is to turn the Risk Watch into a live feedback loop.
Agreed thresholds for user quality and abuse would allow countries to be reviewed consistently as campaigns run. New commercial and traffic-quality data could feed back into the model, helping markets move from test-eligible to validated or into a lower-priority segment when conditions change.
Future automation could introduce timely alerts and regular market reviews. This would develop the spreadsheet into a continuously improving market-selection system.
The same logic could also support other commercial decisions where historical performance needs to be combined with controlled experimentation, including marketing channels, customer groups and products.
This project shows how AI can combine raw data, commercial judgement, compliance requirements and competing team priorities into a practical business process.
The framework already gives Growth, CRM and Revenue a common way to review market opportunity and risk. Every new test can make the next decision more precise, more responsive and more valuable.