Research Standards

What We Publish

Our reports are evidence reviews. Each one takes a question parents ask, such as whether baby trackers are safe for their data, and gathers what published research, regulators, courts and security researchers have found about it.

We don't run our own lab tests, surveys or clinical studies. When a report says something is true, it is because a source we link to says so, and you can open that source and read it yourself.

How a Report Is Researched

1. The question. We write down the question the report has to answer and split it into parts that can be researched one at a time. For the baby tracker report, those were app audits, connected devices such as baby monitors, the law, and what parents can do.
2. The search. For each part we look for the original sources first: the study, the regulator's decision, the court filing, the law itself, the company's own policy. We search widely and keep notes of every source with the passage it supports.
3. The draft. The report is written from those notes. Every factual claim carries a numbered citation to the source behind it, and the numbers lead to the full reference list at the end.
4. The check. Before anything is published, every claim goes through a separate check, described below.

We use AI research tools to search, take notes and draft. They make it possible to read far more sources than we could by hand, and they can also get things wrong. That is why every claim is checked against its source before a report goes up, and why every source is linked so you can check it too.

How We Choose Sources

We prefer the source closest to the facts. In order:

• Primary sources. Peer-reviewed studies, regulator decisions and enforcement actions, court records, laws and official guidance, and a company's own privacy policy or disclosures. A company's own statements are primary evidence of its stated policies, not independent verification of its practices.
• Original security research. Reports from the researchers who found a flaw, with their method described.
• Reporting from established news outlets. Used when a journalist did the original reporting, or when the primary record isn't public. A news report never stands in for a study we could read ourselves.

We don't cite forum posts, anonymous blogs, copies of pages hosted on other sites, or pages that only repeat a claim without saying where it came from. When a link breaks, we replace it with the original page or remove the claim.

Each reference says what kind of source it is ("Peer-reviewed article", "Regulator decision", "News report") so you can weigh it for yourself. Where a source is paywalled, the reference says so.

How Every Claim Is Checked

A separate check opens every cited source and compares each claim with what the source actually says. AI tools assist with this check. Cutiko reviews the report and is responsible for its accuracy before publication. Each claim gets one of three outcomes:

• Supported. The source says it. The claim stays as written.
• Partly supported. The source says something close but narrower. We rewrite the claim to match the source, for example "alleged" where a regulator alleged something and no court has ruled on it.
• Not supported. The source doesn't say it, or the page couldn't be opened to confirm it. The claim is removed or given a source that does support it.

The check covered 81 claims for our first report and 105 for our second. Each report's Methods section gives its own counts: how many claims were supported, how many were reworded, and how many were dropped and why. Each report shows a "Last reviewed" date: the last day its claims were checked against their sources.

Conflicts of Interest

Cutiko makes a baby tracker app, so we have a stake in the subjects we write about. This is how we handle it:

• We say so. Every report states this conflict at the end, under "Competing interests".
• The sources lead. We name other companies and products only where a cited source does, and we report what that source found, nothing more.
• We don't publish commercial rankings. We may report findings about specific products when cited research supports them, but we don't score competitors or present Cutiko as the preferred choice.
• We don't exempt ourselves. When a finding applies to apps like Cutiko, the report says so.
• No one pays for a report. Our reports are not sponsored, and no company named in a report pays us or reviews it before it is published.

Corrections

If you find a claim that doesn't match its source, or a source that has changed, please tell us on our contact page. We check it against the source and correct the report.

Material corrections and updates, such as a fixed error, a patched vulnerability or a new court ruling, are listed at the end of the report under "Corrections and updates", with the date and a short description of what changed. Spelling and formatting fixes are made without a notice. A report's "Last reviewed" date changes only when its claims are checked against their sources again.

If new evidence shows that a finding no longer holds, we don't delete it quietly. The finding stays in the report, marked as withdrawn, and the "Corrections and updates" list gives the date and the reason.

If new evidence changes a report's main conclusions, we write a new report rather than rewriting the old one in place. The earlier report stays online, marked as replaced, with a link to the new one, so anyone can see what we said before and why it changed.