Microbiome test, 6 criteria before you trust it
Before you trust a gut microbiome test, learn what each method measures and why the person reading it matters as much as the result.
Why can two microbiome tests give conflicting results?
A gut microbiome test can give a different result at two labs on the same sample, because the DNA extraction method, the lab protocol and the bioinformatics pipeline differ, and every choice leaves its fingerprint on the final number.
This is not a theoretical worry. In a 2026 study, NIST sent a standardized human stool material to seven direct-to-consumer testing services. The differences between companies were on the same scale as the biological differences between different donors, and the authors traced them to methods and weak quality control.1
The choice of DNA extraction alone has a large effect. Testing 21 extraction protocols on the same stool samples, an international team found that extraction had the largest effect of all the technical steps on what the analysis showed.6
So the same stool sample can end up in two different reports without either lab making a mistake. They measured in different ways. The six criteria below help you judge which report deserves your trust.
Is a microbiome test a diagnosis?
No. A gut microbiome test gives you a list of numbers. What they mean for you depends on your history, your symptoms and the experience of whoever reads them with you.
Two people with identical values on paper can need different approaches, depending on recent antibiotics, travel, a diet change or chronic stress. The number stays the same, and the context around it changes the meaning. The same logic, telling a theoretical mechanism from a measured finding, decides how much to trust any health claim.
Which testing method produced your number?
Behind every number sits one of five lab methods, and each looks at your sample from a different angle.
PCR and qPCR panels
Look only for the microorganisms the lab chose in advance, so they show exactly what they were told to find.
16S rRNA sequencing
Reads one shared piece of the bacterial genome and reports relative abundance, usually at genus level. It is the basis of most consumer tests.
Shotgun metagenomics
Reads all the DNA in the sample, reaches species or strain level and measures functional genes directly.
Culture
Detects only what grows on a plate, a small slice of the microbiome, but it is the only method that shows a living organism and its antibiotic sensitivity.
Mass spectrometry (MALDI-TOF)
Usually paired with culture. It identifies the species of whatever grew on the plate and does not scan the whole sample.
Why the sample material matters
Your report probably comes from one of these methods, most often 16S rRNA sequencing because of cost. A more basic question comes first: which biological material was tested.
My doctor said my urine test shows my gut microbiome is fine. Is that true?
No. Urine has its own urinary microbiome, with a different composition and far fewer microbes than the gut.7 A stool test measures the gut community, and a urine test looks at a separate, much sparser ecosystem. Gut microbiome methods are built around stool for exactly this reason.
What are the 6 quality criteria to check?
See all six together before the detail of each one.
| Criterion | What to check in your report |
|---|---|
| DNA extraction method | The report names the kit or protocol used. |
| Technical controls | A positive and a negative control ran with every batch. |
| Repeatability | The lab can show test-retest data on the same sample. |
| Detection limits | The report states the threshold below which a microbe counts as absent. |
| Relative or absolute abundance | It is clear whether percentages are a share of the sample or a real count per gram. |
| Measured or inferred function | It separates what was measured from what software predicted. |
1. Which DNA extraction method was used?
DNA extraction is the first step of every microbiome analysis, and it changes the final result most without the patient ever hearing about it. How well the lab breaks open the cell walls of each type of bacterium decides which microbes show up, and at what percentage.
- DNA extraction
- The lab step that breaks open the cell walls of microorganisms in a sample and isolates their genetic material before it is copied or sequenced.
Some bacteria, especially Gram-positive ones with a thick cell wall, need stronger mechanical or chemical lysis to release their DNA. A gentle protocol underestimates them even when they are abundant. That is one reason the international comparison ranked protocols partly by how they handled the ratio of Gram-positive to Gram-negative bacteria.6
2. Did the lab run technical controls?
A technical control is a sample of known composition that runs alongside yours through the whole process, so the lab can show the method worked on that day, with that batch of reagents. A negative control separates what came from your sample from what the reagents brought in.
Contaminating DNA was found in commonly used DNA extraction kits and other reagents, varied between kits and batches, and critically affected results from samples with low microbial biomass. The authors strongly advised sequencing negative controls alongside samples.2
A lab that does this as routine says so from the start.
3. Would the same sample give the same result?
Repeatability means the same sample, run twice at the same lab, gives close to the same result. It sounds obvious, yet it is one of the hardest things for a microbiome lab to deliver.
Blinded stool, chemostat and artificial community samples were sequenced by 15 laboratories and analyzed with nine bioinformatics protocols. Variability depended most on the specimen type, followed by DNA extraction, the handling environment and bioinformatics.3
In the NIST study, replicates from the same company were usually closer to each other than to other companies, but not always. One company gave one replicate a very different health picture from the other two.1 A lab that stands behind its quality publishes test-retest data or hands it over on request.
4. What are the detection limits?
Every method has a threshold below which a microorganism counts as absent, even if it is present at a very low level. How low that threshold sits depends on sequencing depth, meaning how many reads the machine produced for the sample, and on the cut-off the lab chooses.
In the NIST comparison, the cut-offs for reporting a microbe ranged from 0.1% down to 0.000008% of the reads.1 A shallow, cheap protocol can show no trace of an organism that a deeper protocol finds in the same sample.
5. Are the numbers relative or absolute abundance?
Relative abundance shows what percentage of the sequenced reads belongs to each microbe, a share within the sample. Absolute abundance shows a real count, usually per gram, and needs a separate measurement.
The difference changes the reading. If a beneficial bacterium such as Faecalibacterium prausnitzii shows a low percentage, it may have dropped, or something else may have multiplied next to it while its own count stayed the same.
Combining flow-cytometry cell counts with sequencing showed up to tenfold differences in microbial load between healthy people. The well-known trade-off between Bacteroides and Prevotella turned out to be an artefact of relative data.4
In practice, check whether your report gives percentages or counts. Most consumer tests give percentages, which is fine as long as you know it before deciding whether something increased or decreased. The same caution applies to a high result, such as Klebsiella high in stool or Pseudomonas high in stool.
6. Was the function measured or predicted?
Many reports include a functional profile, what your microbiome does, beyond who lives there. 16S rRNA sequencing reads one marker gene, not the genes that carry out those functions.
Software such as PICRUSt takes the 16S result and compares it with databases of fully sequenced genomes to predict which genes are likely present. It is a statistical estimate, not a measurement.5
- Reads all DNA in the sample
- Detects functional genes directly
- Reaches species or strain level
- Costs more
- Reads one marker gene
- Infers genes from reference genomes
- Usually stops at genus level
- Cheaper and more common
Shotgun data measure genes. 16S data predict them.
Profiling marker genes such as 16S rRNA does not give direct evidence of a community's functions. PICRUSt predicts the functional make-up from marker-gene data and a reference genome database.5
Infographic
DNA extraction method
The report should name the kit or protocol, because extraction changes the final result most without the patient hearing of it.
Technical controls
A positive and a negative control should run with every batch, to show the method worked on that day.
Repeatable results
The same sample run twice at the same lab should give close to the same result, and the lab can show test-retest data.
Detection limits
The report should state the threshold below which a microbe counts as absent.
Relative or absolute
Check whether percentages are a share of the sample or a real count per gram.
Measured or predicted function
16S sequencing predicts function with software. Only shotgun data measure genes directly.
Sources: Servetas et al. 2026, Costea et al. 2017, Salter et al. 2014, Vandeputte et al. 2017.
gutpathways.com
Save this infographicImage 4:5, with sources, from gutpathways.comWhy does a test result still need an experienced reader?
A report that passes all six criteria is useful, and it is still one piece of data inside a bigger picture. Even with flawless sequencing, some bacterial behavior is switched on and off in ways the DNA sequence alone does not show, and that is where clinical judgment comes in.
In the clinics where I work as a scientific collaborator, the most serious misreadings usually start with a technically correct lab whose result was read without a full history. A flawless test in inexperienced hands often leads to a worse decision than a mediocre test read with real experience.
Use the six criteria as a first filter for any microbiome test you consider. Then find someone who asks for your full history before telling you what the report means. If the test was ordered for gut symptoms, the steps that follow are set out in how to treat gut dysbiosis in the right order, and small intestinal problems need their own tests, described in how to test for SIBO.
Frequently asked questions
The sequencing itself is real, but accuracy across companies is uneven. In a NIST comparison of seven direct-to-consumer services using the same stool standard, the differences between companies matched the differences between two people.1 A test that documents its extraction method, controls, repeatability and detection limits deserves more trust.
It shows which microbes were detected in a stool sample and in what proportions, often with diversity scores and, in some reports, predicted functions. Shotgun tests can also measure functional genes directly. It does not diagnose a disease by itself.
Symptoms such as bloating, irregular bowel habits or discomfort after meals can suggest it, but they are not specific. A stool test adds data, and the meaning comes from reading it with your history, diet and medicines.
It is the percentage of reads from one microbe out of the whole sample. If another microbe multiplies, the first one's percentage can drop while its real population stays the same.4
It can be a useful starting point if the company states its extraction method, technical controls and detection limits. Without that information, and without someone to read it with your history, it is hard to know how much to trust it.
The report gives numbers. Someone with clinical experience and your full history can connect those numbers to your symptoms and suggest something specific to you.
References
7 studies
- Servetas SL, Gierz KS, Hoffmann D, et al. Evaluating the analytical performance of direct-to-consumer gut microbiome testing services. Commun Biol. 2026.
- Salter SJ, Cox MJ, Turek EM, et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol. 2014.
- Sinha R, Abu-Ali G, Vogtmann E, et al. Assessment of variation in microbial community amplicon sequencing by the Microbiome Quality Control (MBQC) project consortium. Nat Biotechnol. 2017.
- Vandeputte D, Kathagen G, D'hoe K, et al. Quantitative microbiome profiling links gut community variation to microbial load. Nature. 2017.
- Langille MGI, Zaneveld J, Caporaso JG, et al. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat Biotechnol. 2013.
- Costea PI, Zeller G, Sunagawa S, et al. Towards standards for human fecal sample processing in metagenomic studies. Nat Biotechnol. 2017.
- Whiteside SA, Razvi H, Dave S, Reid G, Burton JP. The microbiome of the urinary tract, a role beyond infection. Nat Rev Urol. 2015.
Found this useful? Share it with someone who needs it.
Cite this article
Prevedoros, T. (2026, August 31). Microbiome test, 6 criteria before you trust it. Gut Pathways. https://gutpathways.com/articles/microbiome-test-quality-checklist/
