Are gut microbiome tests worth it?
The experiment that settles the reliability question
Most criticism of consumer microbiome testing is theoretical. This is not.
Researchers led from the US National Institute of Standards and Technology, publishing in Communications Biology, took a single human faecal reference material, homogenised it so that every portion was effectively identical, and sent aliquots to seven commercial microbiome services, in triplicate, for 21 tests in total.
The design is what makes it powerful. Because the biological material was standardised, every difference in the results had to come from the companies' methods rather than from any real difference between samples. Any variation observed is measurement error by construction.
What came back:
- Reported abundances varied widely across five clinically relevant genera, including Bacteroides, Bifidobacterium, Clostridium, Roseburia and Faecalibacterium.
- The comparator ranges differed between companies, so even identical measurements would have been graded differently against each provider's idea of normal.
- Variability between providers was on the same scale as the biological variability between different people. The choice of company mattered as much as whose gut it was.
- Replicate samples from the same stool were classified as both healthy and unhealthy, with divergent dietary and functional recommendations attached.
A test that grades one sample two different ways cannot be used to grade you. Everything downstream of the result, the score, the dysbiosis rating, the personalised food list, inherits that instability.
Why the results differ so much: what is actually being measured
The variation is not carelessness. It comes from real, defensible methodological choices that different companies make differently, and understanding them explains why the problem is hard.
Sequencing method
Most services use one of two approaches. 16S rRNA amplicon sequencing reads one marker gene: cheap, well established, and typically resolving only to genus level. Shotgun metagenomics sequences everything present, resolving to species and revealing functional genes, at higher cost. The gap is large. In comparison work, shotgun metagenomics classified around 74% of microbial reads to species level against roughly 10% for 16S.
Each has its own biases. 16S is distorted by unequal amplification between species. Shotgun may not sequence deeply enough to catch rare organisms. Two companies using different methods are answering measurably different questions.
Bioinformatics and reference databases
After sequencing, raw reads are matched against a reference database using a pipeline of algorithmic choices. Different databases and different pipelines yield different taxonomic calls from identical data. None of this is visible to the customer.
The comparator problem
This is the deepest issue and it is not solvable by better sequencing. There is no established reference range for a healthy gut microbiome. No agreed composition, no validated diversity threshold, no clinical cut-off. Each company therefore constructs its own comparison population and its own scoring, which is why the same numbers earn different verdicts. The 2024 international consensus statement in Lancet Gastroenterology & Hepatology considers microbiome testing investigational and does not recommend it for routine clinical use.
The sampling problem, before any sequencing happens
- One stool is one moment. Gut microbial composition shifts with diet, illness, travel, sleep and medication, sometimes substantially within days. A single sample is a snapshot.
- Stool is not the gut. What you sample is largely luminal content from the distal colon. The mucosa-associated community, which is arguably the more relevant one for interacting with your immune system, is not well represented, and the small intestine is not represented at all. That matters, since conditions like SIBO live in the small intestine.
- A single stool is not homogeneous. Different portions of the same specimen can give different profiles, which is precisely why the NIST team had to homogenise their reference material to run a fair test.
- Recent antibiotics distort everything for months.
Where the science is real, stated properly
This is the part that a purely sceptical page gets wrong, and it matters more than the criticism. The gut microbiome is not a wellness invention. It is one of the more important areas in modern medicine, and its causal significance has been demonstrated, not merely suggested.
The clearest proof is faecal microbiota transplantation for recurrent Clostridioides difficile infection. Replacing a person's gut microbial community with a donor's cures a serious, sometimes fatal disease, and it does so at rates well above standard antibiotics in randomised trials, with meta-analyses reporting cure in roughly three quarters of patients against around half on antibiotics. It is recommended in guidelines for recurrent disease.
That is about as strong as causal evidence gets: change the microbiome, cure the illness. Anyone who tells you the microbiome does not matter is as wrong as the company selling you a score for it.
The gap is between the microbiome matters and we can read yours and tell you what to eat. The first is established. The second is not, and conflating them is the entire business model.
The named products, fairly
Assessed on what each has actually published rather than on how they market.
ZOE
ZOE deserves more care than the rest, because it is backed by genuine peer-reviewed science and treating it as equivalent to an unpublished product would be inaccurate. The PREDICT programme, run with researchers including teams at King's College London and Massachusetts General Hospital, studied over a thousand participants including twin pairs, and produced real findings about how differently people respond to identical meals.
Two things are worth being precise about.
- PREDICT measured postprandial metabolic responses, principally blood glucose, triglycerides and insulin after standardised meals. It was not a study of gut symptoms, bloating or IBS.
- The microbiome's contribution was modest. In PREDICT it accounted for approximately 6% of the variation in glucose response and 8% of the variation in triglyceride response. That is a genuine and interesting signal. It is not a basis for a personalised food score, and it is far smaller than the marketing impression most people take away.
The outstanding gap is outcomes. No randomised trial has yet shown that following ZOE's personalised recommendations produces better health results than standard evidence-based advice, which is to eat more fibre, more plants, and less ultra-processed food. That is the trial that would settle it, and it has not reported.
Viome
Uses RNA sequencing, which in principle reports on microbial activity rather than only presence, a reasonable idea. The translation from that data into specific food and supplement recommendations, including its own supplement products, has not been independently validated. Recommending supplements you also sell is a structural conflict regardless of the underlying science.
Ombre, Thorne, Tiny Health and the wider field
Broadly similar: 16S or shotgun sequencing, a proprietary scoring layer, and food or supplement recommendations. The scoring and the recommendations are the unvalidated parts, and they are the parts you are paying for.
Stool testing that is validated
The contrast is instructive, because it shows what a validated stool test looks like: one clinical question, one threshold, one decision.
| Test | Question it answers |
|---|---|
| Faecal calprotectin | Is there intestinal inflammation? Separates IBD from functional disorders and avoids unnecessary colonoscopy. |
| Faecal immunochemical test (FIT) | Is there occult blood? Used in bowel cancer screening programmes. |
| C. difficile toxin / PCR | Is this diarrhoea caused by C. difficile? |
| Stool culture and pathogen PCR | Is there an infectious cause of this diarrhoea? |
| Faecal elastase | Is the pancreas producing enough enzyme? Below roughly 100 µg/g suggests insufficiency. |
Where a consumer microbiome test genuinely has value
Fairly stated, because they are not worthless:
- Curiosity is a legitimate reason to buy something. If you want to see what is in your gut and can afford it, that is your call. The mistake is treating the output as a diagnosis.
- Research contribution. Several companies feed anonymised data into genuine research, and the field needs data at scale.
- Marginal nudges. A strikingly low diversity result prompting a conversation about dietary variety is not nothing. It is a nudge, not a finding.
- The field is moving. A verdict in 2026 is not permanent. Standardised reference materials, of the kind used in the NIST study, are exactly how a field grows out of this problem, and this page will be updated when it does.
Two results these panels commonly return get sold on as diagnoses and deserve their own reading: a candida overgrowth flag, and a biofilm interpretation.
A note on how these are reviewed elsewhere
Consumer microbiome tests carry some of the most generous affiliate commissions in this category and are among the most heavily promoted products in consumer health. A "best gut health tests compared" article that enthusiastically recommends three of them, all with tracking links, is not necessarily wrong, but it is not disinterested either.
This page has no affiliate links, and no page on this site does. That is a deliberate decision and it is the reason the ratings here are worth reading.
What to do instead
If the question is "what is causing my symptoms", the sequence is unglamorous and it works: exclude the treatable conditions with validated tests, then identify your own triggers by observing what actually happens, repeatedly, against your own baseline rate.
That second half is the part a stool test cannot do for you, because the answer is not written in your microbiome. It is written in what happens after you eat. Minthe establishes your personal base rate first, corrects for the fact that testing many foods at once produces false positives, and asks you to confirm each suspect with a guided test rather than stopping at a correlation. The method is published in full, including what it cannot do.
Minthe is a wellness and self-tracking tool, not a medical device. It does not diagnose or treat any condition and isn't a substitute for professional medical advice. If you have red-flag symptoms such as blood in your stool, unintentional weight loss, difficulty swallowing, persistent vomiting or a fever alongside gut symptoms, see a clinician rather than tracking them.
Sources
- Servetas SL, et al. Evaluating the analytical performance of direct-to-consumer gut microbiome testing services. Communications Biology. National Institute of Standards and Technology.
- International consensus statement on microbiome testing in clinical practice. Lancet Gastroenterology & Hepatology, 2024.
- Berry SE, et al. Human postprandial responses to food and potential for precision nutrition (PREDICT 1). Nature Medicine, 2020.
- Updated systematic review and meta-analysis of faecal microbiota transplantation for recurrent Clostridioides difficile infection. eClinicalMedicine, 2020.
- Comparative analyses of 16S rRNA amplicon sequencing and shotgun metagenomics for taxonomic resolution in stool samples.
Frequently asked
Are consumer microbiome tests accurate?
Not reliably, and this has been measured directly rather than merely argued. Researchers at the US National Institute of Standards and Technology sent identical aliquots of a single homogenised human faecal reference material to seven commercial services in triplicate, 21 tests in total. Because the material was standardised, any difference in results came from the testing methods rather than from biology. Reported abundances for clinically relevant genera varied widely, comparator ranges differed between companies, and in one case replicate samples from the same stool were classified as both healthy and unhealthy, with different dietary recommendations attached.
Does that mean the microbiome does not matter?
No, and this is the distinction that matters most. The gut microbiome is unambiguously real, important, and causally manipulable. The clearest proof is faecal microbiota transplantation for recurrent Clostridioides difficile infection, where replacing a person’s gut community cures a serious disease at rates well above antibiotics in randomised trials. The microbiome is real science. Consumer scoring of it is the part that has not been validated.
Is ZOE different because it has published research?
ZOE’s PREDICT studies are genuine, peer-reviewed nutrition science and it would be unfair to lump them in with unpublished products. But it is worth being precise about what they established. PREDICT measured how people vary in their postprandial blood glucose, triglyceride and insulin responses to standardised meals. Within that, the gut microbiome accounted for a modest share of the variation, around 6% for glucose and 8% for triglycerides. The studies did not test whether the commercial programme improves health outcomes compared with standard dietary advice, and no randomised trial has shown that it does.
Can a microbiome test tell me which foods trigger my bloating?
No. There is no validated pathway from a stool sequencing result to an individual food trigger, and the personalised food lists these services produce have not been shown in trials to outperform generic advice or to reduce symptoms. Working out your own triggers requires observing what actually happens when you eat something, repeatedly, compared against how often you get symptoms anyway.
What stool testing is actually validated?
Several things, and they are worth having when indicated. Faecal calprotectin distinguishes inflammatory bowel disease from functional gut disorders. Faecal immunochemical testing is used in bowel cancer screening. Toxin testing and PCR identify Clostridioides difficile. Stool culture and PCR identify infectious causes of diarrhoea. Faecal elastase screens for pancreatic insufficiency. These are validated for specific clinical questions, which is exactly what a microbiome wellness score is not.