Executive Summary
Actuaries often ignore the new BI platform, instead rebuilding risk models in Excel before a board meeting. This creates a bottleneck and relies on one or two key people.
This isn't just about inefficiency. The Excel models are usually undocumented and hard to audit, which creates a genuine regulatory and valuation risk the board may not be aware of.
The answer isn't another dashboard or more training. It's about changing the data's structure to build a governed, auditable Semantic Layer. This is where the actuarial logic can be properly codified, which is what's needed to earn their trust.
Why actuaries still use Excel for board reporting
It’s the day before the risk committee meeting. The leadership team is looking at a smart, expensive dashboard showing key reserving and loss ratio metrics. But the Chief Actuary knows what’s really happening. Their team is frantically exporting raw data to CSVs, running it through a complex set of macros in a master spreadsheet that only one or two people truly understand.
They aren't doing this because they are resistant to change. They are doing it because their professional credibility is on the line. For an actuary, a number isn't just a number: it's a liability. If they cannot trace its lineage from the source policy and claims systems all the way to the final cell in their model, they will not, and should not, trust it.
This reliance on Excel is the symptom of a broken data architecture, not a failure of your people. You’ve likely invested six or seven figures in a modern data stack, but it can feel like you've only created a faster way to access numbers nobody can defend under pressure.
Why the 'single source of truth' isn't trusted
Your data team will tell you they have built a "Single Source of Truth." They are probably right, from their perspective. The data warehouse is centralised, and the pipelines are running. The problem is that the most critical component is missing: a shared, transparent layer of business logic.
The definition of 'Incurred Claims' or 'Loss Adjustment Expense' is buried in a complex dbt model or a tangle of SQL scripts. The actuaries can't see it, they can't audit it, and they certainly can't approve it. To them, the dashboard is a black box, and in the world of risk management, black boxes are not acceptable.
In my experience, this points to a problem with Data Trust. I see this in many high-volume, high-regulation businesses I work with. The technical teams build a perfect engine, but they forget the people who have to fly the plane need to be able to inspect it themselves. Without a robust Data Governance framework that prioritises transparency over just speed, your BI tool is little more than a very expensive calculator that everyone ignores.
How to build trust with a semantic layer
The only way I've seen this get solved is to stop patching the reports and instead rebuild the foundations of trust. The fix isn't about a specific technology, it's about the structure of your data. It involves creating a deliberate, governed space between your data warehouse and your BI tool.
This is the role of a Semantic Layer. It's where we take the business logic out of the complex SQL scripts and the hidden Excel formulas and place it into a single, version-controlled, human-readable repository.
This process transforms your data setup from a collection of separate reports into a genuine Single Source of Truth that can withstand the scrutiny of not just your board, but your regulators.
This is more about people than technology
Let's be honest. Implementing this is not a simple task. It requires taking away the spreadsheets that your most critical team relies on. You should expect some resistance.
The initial phase involves slowing down to codify what might be decades of institutional knowledge. It's a process of translation and negotiation. You are mapping the complex reality of your business onto a structured data model. This is difficult, meticulous work that can't be rushed.
You will probably move slower for a quarter to move faster, and with more confidence, for the next five years. The goal is to make the central, governed system so transparent, reliable, and powerful that going back to an offline spreadsheet becomes the far riskier option.
By building this foundation, you aren't just fixing a reporting problem. You are creating a system where your best people can focus on analysing risk, not hunting for the right data. That is a reliable way to scale an InsurTech business with confidence.