Send the same stool sample to four companies and you get four different microbiomes
Direct-to-consumer gut testing has a reproducibility problem large enough to invalidate the product. Variation between providers analysing identical samples is as large as the variation between different people.
There is a straightforward way to test a test. Split one sample, post the halves to different laboratories, and see whether the answers agree. Somebody did this to the direct-to-consumer gut microbiome industry, and the results should end the category in its current form.
A 2026 analytical evaluation published in Communications Biology assessed the performance of consumer gut microbiome testing services. The headline finding is the one that matters: variability between providers analysing the same material was on the same scale as the biological variability between different donors.1
Read that again slowly, because it is easy to underweight. The difference between two companies looking at your gut is about the same as the difference between your gut and a stranger’s. Which company you posted to carries roughly as much information as who you are.
The quality control problem underneath it
The evaluation attributed the spread to methodological variability and insufficient quality control rather than to anything inherent in sequencing.1 Some companies produced consistent results across replicates. Others clearly did not.
One detail deserves isolating. A provider returned a replicate in which more than 50% of sequencing reads were unidentified, and issued a consumer report from it anyway.1 More than half the data was noise and the customer received a confident document describing their gut flora.
That is not a measurement problem. It is a disclosure problem, and it tells you what the report is for.
The test does not know what it does not know, and the report is not designed to tell you.
What the clinical literature says
The reproducibility finding lands on top of an evidence base that was already thin.
There are no regulatory-approved clinical microbiome diagnostic tests.1 The international consensus statement on microbiome testing in clinical practice concluded that evidence supporting clinical usefulness is scarce, and that commercial providers sell direct-to-consumer tests without proven value in practice.3 Reviews of microbiome variability reach the same place from the methodological side: the difficulty of disentangling correlation from causation, limited clinician familiarity, and no established framework for clinical translation.4
The harm is downstream, as usual
The blood draw equivalent here is a stool sample, so the direct physical harm is nil. The documented harm is what people do with the report.
Clinicians report patients seeking medical help on the basis of at-home microbiome results,1 which is a particular problem when the underlying science is unsettled: a consultation that begins with an unvalidated result has to be spent dismantling it. Regulatory analyses of the category describe medical, economic and dignitary harms as the relevant categories, which is unusually blunt language for a policy paper.5
There is also an opportunity cost that nobody itemises. Somebody spending several hundred euros on a microbiome panel has that money unavailable for measurements with outcome evidence, and there is a long list of those.
Why grade D
Grade D means marketed ahead of its evidence: measurement unreliable, or the claimed inference unsupported. Consumer microbiome testing satisfies both conditions, and the first one comprehensively.
Scope matters, because the underlying science does not deserve this grade. Gut microbial ecology is a serious research field, the associations with metabolic and immune conditions are real, and therapeutic applications in defined conditions such as recurrent C. difficile infection are established medicine. The grade attaches to the consumer product: a posted sample, a composition readout of contested reproducibility, and a personalised dietary plan derived from it.
What would change the grade
Reproducibility first, and everything else second. If providers published replicate agreement statistics, the proportion of unidentified reads per sample, and their sequencing and bioinformatics pipeline, a buyer could distinguish the competent from the rest. Nothing about that requires new science, and no provider we are aware of publishes it.
Beyond that, the field needs validated thresholds tied to defined clinical actions, and prospective evidence that acting on a microbiome result produces a better outcome than the generic advice it usually reduces to. Until then, the honest summary of what these tests recommend is: eat more fibre and a wider range of plants. That is good advice. It costs nothing, and it does not depend on what your report said.
Questions this article answers
- Are consumer gut microbiome tests accurate?
- They are not reproducible enough to be called accurate. A 2026 analytical evaluation found that variability between direct-to-consumer providers analysing the same sample was on the same scale as the biological variability between different donors, attributed to methodological differences and insufficient quality control. One provider issued a consumer report from a replicate in which over half the sequencing reads were unidentified.
- Is a microbiome test worth the money?
- On current evidence, no. There are no regulatory-approved clinical microbiome diagnostic tests, the international consensus statement on microbiome testing in clinical practice describes the evidence for clinical usefulness as scarce, and the reproducibility data means your result depends substantially on which company you posted the sample to. The dietary advice these tests generate is generally advice that applies to almost everyone regardless of their microbiome.
- Can a microbiome test tell me what foods to eat?
- It can generate a food list, but the list is not reliably derived from your microbiome. Two problems compound: the composition measurement itself varies between providers as much as it varies between people, and the step from a composition reading to a personalised dietary recommendation rests on associations rather than demonstrated causation. Eat more fibre and more plant diversity is sound advice and does not require a test.
- Is the gut microbiome important?
- Almost certainly yes, and that is a separate question from whether a consumer test measures it usefully. The research field is serious and advancing. The gap is translational: disentangling correlation from causation, the absence of validated clinical thresholds, and no approved diagnostic. A field can be genuinely important while its consumer products are not yet worth buying.
- Evaluating the analytical performance of direct-to-consumer gut microbiome testing services. Communications Biology. 2026.
- Direct-to-consumer microbiome tests prove wide variability. AGA / Gastroenterology news, April 2026.
- International consensus statement on microbiome testing in clinical practice.
- Navigating microbiome variability: implications for research, diagnostics, and direct-to-consumer testing.
- Is the current regulatory framework for direct-to-consumer microbiome-based tests sufficient to protect consumers from medical, economic, and dignitary harms?
More from Baseline
Claim check: “our patients reversed their biological age by six years”
A recurring line in longevity clinic marketing, and a clean worked example of how a defensible measurement becomes an indefensible statement.
Neko Health alternatives: what to do about a 100,000-person waitlist
Neko built the best-designed cheap health check in Europe and cannot serve the demand. Wait times run six to twelve months. Here is what the alternatives actually replace, and what none of them do.
The cancer blood test finally has a big trial. Read the sensitivity number twice.
PATHFINDER 2 put Galleri through 35,878 people and produced the strongest evidence any multi-cancer blood test has. It also showed the test misses about six in ten cancers, and both facts are compatible with it being useful.