What consumer DNA tests can and cannot tell you

For less than the price of a decent pair of trainers, you can spit in a tube and receive a report claiming to describe your ancestry, your disease risks and whether you’re genetically inclined to hate coriander. Some of that is solid science. Some is statistical entertainment wearing a lab coat. Knowing which is which matters, because the failure modes run in both directions: false alarm, and false reassurance.

Here’s an honest audit of what the kits can and cannot do: and what happened to the biggest company in the business, because it’s the best privacy lesson available.

What you’re actually buying

Almost all consumer kits don’t sequence your genome. They genotype it: they check several hundred thousand individual positions where DNA commonly varies between people, out of roughly three billion letters. Think of it as photographing a few hundred thousand road junctions and calling it a map of Britain.

That’s not a swindle, those positions are chosen because they’re informative, but it sets the ceiling. The kit can only report on the specific variants it looks for. Anything it doesn’t check simply doesn’t appear, which becomes important later.

What they can genuinely tell you

• Ancestry, roughly. Ancestry percentages are estimates made by comparing your DNA with the company’s reference populations. They’re decent at the continental scale, fuzzier at the regional one, and they shift when the company updates its reference data, which is why your results can quietly change years after you tested. Treat “12% Scandinavian” as an estimate with error bars, not a birth certificate.
• Relative matching. This is the genuinely strong feature. Shared DNA segments identify relatives with real reliability, which is why the tests are powerful for genealogy, adoption searches and connecting with unknown family.
• Carrier status. Some kits screen for variants relevant to conditions like cystic fibrosis: useful information when you’re planning children, though the NHS route matters if anything is flagged, and a clear result doesn’t rule out being a carrier of a variant the kit doesn’t cover.
• A handful of well-studied risk variants. A few findings are meaningful and well validated. But coverage is thin: the best-known example is that a leading kit’s BRCA report has historically covered only three specific variants out of the thousands known to raise breast and ovarian cancer risk. Meaningful when positive; nowhere near an all-clear when negative.

What they cannot tell you

• Your actual risk of most diseases. Heart disease, type 2 diabetes, most cancers and dementia are shaped by hundreds or thousands of genetic variants plus decades of environment and behaviour. Reports condense this into tidy “slightly elevated risk” statements that sound more predictive than they are. Your blood pressure, waist measurement and smoking status remain far better forecasters than any consumer genetic report: and unlike your genome, you can change them.
• A clean bill of health. This is the quiet danger. Because kits check only selected variants, “no variants detected” means “none of the ones we looked for”, not “low risk”. Researchers and clinicians have repeatedly warned about people taking false reassurance from a clear report, particularly with a strong family history. A BMJ analysis of direct-to-consumer testing made exactly this point: for individuals, results can mislead in both directions.
• Reliable lifestyle guidance. Reports on “genetic” diet response, muscle type, caffeine metabolism and sleep chronotype sit on thin evidence. None of it should change what you eat or how you train more than one honest week of paying attention would.
• Anything trustworthy from raw-data uploads. Many services let you download raw data and run it through third-party interpretation websites. A study in Genetics in Medicine found that around 40% of variants flagged in raw consumer data and sent for clinical confirmation turned out to be false positives. If a third-party report ever alarms you, treat it as unverified until a clinical lab says otherwise.

The surprises nobody orders

One category of result is more reliable than people expect: the relative finder. Every year, testers discover half-siblings, unexpected parentage, or a family story that doesn’t survive contact with the data. If several family members test, the truth tends to surface whether or not anyone wanted it. It’s worth deciding before you spit whether you’re genuinely open to that, and whether the relatives who share your DNA got a vote.

The privacy part: read this before the science sways you

Your genome is the one password you can’t reset, and recent history shows exactly how it can travel.

In 2023, credential-stuffing attacks against 23andMe exposed profile data linked to roughly seven million customers. In March 2025, the company filed for bankruptcy in the US, and its database of genetic information from more than 15 million people was treated as what it legally was: an asset to be sold. The US Federal Trade Commission publicly warned that any sale must honour the privacy promises made to customers, and in mid-2025 a court approved the sale of the company and its data to TTAM Research Institute, a non-profit founded by 23andMe’s own co-founder, for $305 million.

That outcome was relatively benign. The lesson isn’t. Privacy policies are corporate promises, and corporate promises are only as durable as the corporation. In an acquisition or bankruptcy, your data is on the balance sheet.

Three more things worth knowing:

• UK GDPR is on your side. Genetic data is special category data under UK law. You can request deletion of your data and destruction of your stored sample. Companies serving UK customers must provide a route to do both.
• Your choice implicates relatives. You share large fractions of your genome with your family. Police in several countries have identified suspects through relatives’ entries on genealogy databases; whatever you think of that, it illustrates that a DNA database entry is never only about you.
• Research sharing is usually opt-in. Check the boxes. Consent settings at sign-up decide whether your de-identified data is shared with research and commercial partners. Untick what you don’t want; you can usually withdraw later, but data already used stays used.

A five-point check before you buy

1. Can you use a pseudonym and a dedicated email address?
2. Does the company let you delete your data and destroy your sample, and is the process documented?
3. What exactly do the research-sharing tick-boxes opt you into?
4. Where is the company based, and does it commit to UK GDPR standards for UK customers?
5. What happens to your data if the company is sold? (If the policy doesn’t say, that’s your answer.)

If a service passes that check and your goal is ancestry or genealogy, a kit can be genuinely worth the money. We haven’t yet completed our review of specific providers, so no recommendation here for now.

When the NHS is the right route

If you have a strong family history (say, breast, ovarian, bowel or prostate cancer in close relatives, especially at young ages, or a known inherited condition in the family), don’t outsource that question to a £99 kit. Speak to your GP, who can refer you to an NHS clinical genetics service. Clinical testing looks properly at the genes in question, comes with genetic counselling before and after, and its results are actually designed to carry medical weight.

The verdict

Consumer DNA tests are a reasonable purchase for ancestry, genealogy and curiosity, provided you buy with your eyes open on privacy. As medical instruments, they’re weak: occasionally usefully right, structurally prone to false reassurance, and no substitute for the boring measurables (blood pressure, cholesterol, blood sugar, waist), that actually predict most men’s futures. Buy one for the family tree, not the crystal ball.

Common questions

How accurate are consumer DNA tests for ancestry?

Ancestry percentages are estimates, not facts. Kits compare your DNA with reference populations, so they are fairly reliable at the continental level and much fuzzier by region or country. Results often shift when a company updates its reference data, which is why your percentages can change years after you tested. Relative matching is the strong part: shared DNA segments identify close relatives with real reliability, which is why the tests work well for genealogy and finding unknown family.

Can a home DNA test tell me if I will get a disease?

No. Most common conditions, including heart disease, type 2 diabetes and dementia, depend on hundreds of variants plus decades of lifestyle and environment, which a consumer report cannot capture. A “no variants detected” result only means none of the variants the kit checks for, not low risk. Your blood pressure, waist measurement, cholesterol and smoking status predict your future far better. If serious illness runs in your close family, ask your GP about an NHS clinical genetics referral.

Should I trust a health result from a home DNA kit or a raw-data upload?

Treat it as unconfirmed until a clinical laboratory checks it. Consumer kits genotype a few hundred thousand positions, and third-party sites that interpret raw data have a poor record: one study found around 40% of flagged variants were false positives on clinical confirmation. Do not act on a worrying result, and do not panic over it. Book a GP appointment, take the report with you and ask about referral to NHS clinical genetics, where testing comes with counselling.

Can I get genetic testing on the NHS instead?

Yes, if there is a clinical reason. NHS genetic testing is free but needs a referral, usually because of a suspected inherited condition or a strong family history, for example breast, ovarian, bowel or prostate cancer in close relatives, especially at young ages. Start with your GP, who can refer you to a clinical genetics service. NHS testing looks properly at the relevant genes and includes genetic counselling before and after, so the results are designed to carry medical weight.

Can police or insurers get hold of my DNA test results?

Police in several countries have identified suspects through relatives’ entries on genealogy databases, and because you share large parts of your genome with family, an entry is never only about you. UK insurers are different. Under the ABI Code on Genetic Testing and Insurance (2018 edition) they do not ask for or use predictive test results, the only exception being Huntington’s disease for life cover over £500,000. GPs are told not to include such results in insurer reports.

Can I delete my DNA data after testing?

Yes. Genetic data is special category data under UK GDPR, so a company serving UK customers must give you a documented route to delete your account data and destroy your stored sample. Do it before any sale or bankruptcy if you can, because a database is a corporate asset that can change hands. Note that de-identified data already used in research stays used. If a company refuses or stalls, you can complain to the Information Commissioner’s Office, which is free.

Sources

• Genetic and genomic testing, NHS
• Horton et al. (2019), Direct-to-consumer genetic testing, BMJ
• Tandy-Connor et al. (2018), False-positive results in direct-to-consumer raw data, Genetics in Medicine
• Judge approves sale of 23andMe and its DNA data to TTAM Research Institute, NPR
• 23andMe bankruptcy and data sale coverage, CNBC
• Bankruptcy court approves sale of 23andMe, HIPAA Journal
• Special category data, Information Commissioner’s Office