The honest answer is that accuracy varies enormously depending on what is being tested and how the company built its product, and this is one of the most consistent concerns raised by professional genetics organizations. Most direct-to-consumer kits use a technology called genotyping, which checks a large number of specific, pre-selected positions in the DNA rather than reading every letter of a gene from end to end the way a full clinical sequencing test does. Genotyping is a reasonable tool for ancestry estimates and for looking at well-established, common variants, but it is far more prone to a specific kind of error, a false positive, when it is used to flag rarer, disease-causing changes in genes such as those linked to hereditary cancer.

This is not a theoretical concern. A widely cited study by a clinical genetics laboratory took forty-nine samples where consumers had used third-party services to reinterpret their raw direct-to-consumer data and had received a report describing a variant of concern in a disease-related gene. When those same samples were retested using accredited, clinical-grade sequencing, forty in every hundred of the flagged variants turned out to be false positives, meaning the person did not actually carry that genetic change at all. Some people had already begun planning preventive surgery or additional screening based on results that were simply wrong. The pattern was strongest for raw genotyping data run through unregulated third-party interpretation tools rather than a company’s own core report, but the underlying lesson applies broadly: a result generated outside a clinical laboratory, without the quality checks that accredited diagnostic laboratories are required to meet, should be treated as a starting point for a conversation with a qualified professional, never as a final answer.

There is a separate and equally important limitation even when a result is technically correct: what it means for a person’s actual health. A polygenic risk estimate for a common condition combines many genetic variants of small individual effect, and how well these scores predict risk depends heavily on which population was used to build the underlying research, meaning a score built mostly from data on one ancestry group can be considerably less accurate for a person from a different background, including most people from India. A negative result on a direct-to-consumer test, such as one that only checks three specific breast cancer gene changes out of many hundreds, tells a person almost nothing about whether they carry a different, equally important change in the same gene, and should never be read as clearing someone of hereditary cancer risk if their family history otherwise suggests concern.