How this works
Each experiment is a separate property with its own owner, deployed and operated independently. This platform observes them: it registers what is being tested, records what the market actually did, and decides whether to continue, hold, or stop.
Test in the market
A real website goes live against real users. No simulation, no projections standing in for demand.
Observe real signals
Search visibility, returning users, contact clicks, inquiries, quote requests, registrations, external references.
Decide, and record it
Continue, hold, or stop — recorded with its reasoning, so the judgement can be reviewed later.
Keep the learning
A stopped experiment is not a deleted one. Failure stays in the dataset because it is still evidence.
Experiment properties currently live
These properties are online and testing right now. They are separate projects; this platform observes them and does not modify them.
| Property | Domain | What it tests | Lang |
|---|---|---|---|
| Money Machine Intelligence | money.datumharbor.com | How real businesses create value and generate revenue | en |
| Commercial Failure Intelligence | failure.datumharbor.com | Commercial failures and the lessons they leave behind | en |
| Problem Discovery Engine | problem.datumharbor.com | Evidence-based discovery of commercially real problems | en |
| Qianmen — Defensive Scam Intelligence | scam.datumharbor.com | Scam patterns, warning signals and protective lessons | en |
| Enterprise AI Work Intelligence | aiwork.datumharbor.com | Real enterprise work, AI capability and deployment gaps | en |
| Global Innovation Radar | radar.datumharbor.com | Innovation signals and emerging capability shifts | en |
| 生意标本 (GSBI) | case.shengyibiaoben.cn | 记录真实商业样本，研究可复用的赚钱机制 | zh-CN |
No results, findings or playbooks are published yet. Nothing is shown here that cannot be traced to evidence, and no metric is displayed until it has been observed first-hand. Search, traffic, behaviour and commercial figures will appear only when there is real data behind them.
How a decision gets made
Every active experiment carries one lifecycle state and one decision. Lifecycle says where the test is; decision says what we do about it. They are recorded separately, and both are visible in the methodology.
What DatumHarbor is not
- It is not a directory of invented case studies. Every sample is a real property that really went live.
- It is not a prediction engine. An opinion — including an AI's — is not market validation.
- It does not take over the experiment properties. They are separate projects with their own owners.
- It does not report numbers it has not observed. Where something is unknown, it stays unknown.