The Validated Synthetic Data Principle

Every SaaS product starts with synthetic data. Demo dashboards, sample scores, placeholder charts. There is nothing wrong with that — you need to ship before you have real users.

But there is a massive difference between invented synthetic data and validated synthetic data. One builds trust. The other destroys it the moment a compliance officer asks: "Where does this number come from?"

The Problem with Fake Scores

In EU regulatory compliance, numbers carry legal weight. When your dashboard shows a company's CSRD readiness at 72%, that number implies a methodology. When an auditor asks for the source, Math.random() * 100 is not an acceptable answer.

// "Looks about right"
const csrdScore = 72;
const etsScore = 85;
const cbamScore = 45;
const overallScore = Math.round((csrdScore + etsScore + cbamScore) / 3);

This approach has three failure modes:

1. Trust collapse — The first enterprise buyer who asks "where does 72% come from?" gets no answer. Deal lost.
2. Migration debt — When you finally connect real data, every number changes. Users think the product is broken.
3. Legal risk — In regulated industries, showing compliance scores without methodology can be classified as misleading representation.

The Validated Synthetic Data Principle

Instead of inventing numbers, you build every synthetic value from the same official data that the real system will eventually use. The result: your demo data is synthetic in form but real in provenance — sector-bounded, regulation-grounded, and ready to upgrade to company-specific values.

// EU ETS benchmark — Commission Implementing Regulation 2021/447
{
  sector: 'steel',
  product: 'Hot metal',
  factor: 1.328,
  unit: 'tCO2e/t product',
  source: 'EU ETS Benchmark 2021-2025',
  methodology: 'ets-benchmark',
  year: 2024
}

How We Apply It at DWS IQ

Our compliance scoring engine uses this principle at every layer. Here are the six data categories and their validated sources:

The Upgrade Path Is Built In

The beauty of validated mock data is that upgrading to real data is not a migration — it is a configuration change. The scoring engine does not care whether the input came from a sector benchmark or a company's actual ESG report:

Same scoring engine. Same regulatory thresholds. Same deterministic logic. The only thing that changes is the input source — and each layer is strictly better than the last.

Five Rules for Validated Synthetic Data

Bottom Line

Fake compliance scores are worse than no scores. They create false confidence, fail under scrutiny, and require painful migration when real data arrives.

Validated synthetic data costs the same effort to build — you are just sourcing from EU regulatory databases instead of your imagination. And when a compliance officer asks "where does this 42% CSRD readiness come from?", you can answer: "EFRAG ESRS 2023 final set, 82 mandatory disclosures, cross-referenced with EEA sector averages for steel (NACE C24). Here is the breakdown."

That answer closes deals. The other one loses them.