Most people arrive at ancient-DNA analysis holding a 23andMe or AncestryDNA result and one question: why doesn't this match? "62% Italian" from one company, "48% Southern European" from another, and then an ancient-DNA model that talks about Anatolian farmers and steppe herders and never mentions Italy at all.
None of those results is wrong. They are answers to different questions, computed against different reference sets, and the disagreement is built in. This guide sets the two kinds of test side by side so you can decide which question you are actually asking — and shows how the second kind runs on the same raw file you already downloaded.
What a consumer ethnicity estimate measures#
23andMe, AncestryDNA, MyHeritage and their peers compare your genotypes with a reference panel of living people whose four grandparents came from the same place. Your genome is split into windows, each window is assigned to the reference group it most resembles, and the shares are summed into the familiar pie chart.
Three properties follow:
- The categories are modern. "Italian", "British & Irish", "Balkan" are labels for present-day populations. They describe where people who look like you genetically live today.
- The categories are the company's. Each vendor draws its own regions, so the same genome gets different pie charts from different companies — a fact every vendor's own help pages acknowledge.
- The categories are recent. Modern populations are themselves mixtures of older ones. A "100% Sardinian" estimate is a single modern label sitting on top of a deep history of Neolithic farmers, hunter-gatherers and later arrivals that the label does not decompose.
That is a useful product. It answers "which present-day populations am I most similar to?" well, and it is very good at finding living relatives. It does not answer where that similarity came from.
What an ancient-DNA ancestry test measures#
An ancient-DNA test compares your genome with excavated individuals — genomes sequenced from skeletal remains and published in the scientific literature — and asks how your ancestry can be described as a mixture of populations that lived thousands of years ago. The reference panel is public and versioned: ours is the Allen Ancient DNA Resource (AADR) v66, roughly 23,265 samples across roughly 6,015 population labels, explained in AADR explained.
The categories are therefore ancient populations, not modern countries: Western Hunter-Gatherers, Anatolian Neolithic farmers, Western Steppe Herders, Iron Age and Roman-period groups. The result is not "62% Italian" but something like "54% Anatolian farmer-related, 31% steppe-related, 15% hunter-gatherer-related" — the components that, layered over millennia, produced the modern population the consumer test named. Each of those source populations has its own page in the ancestry population directory, and three of them are explained in depth in Steppe ancestry, Hunter-gatherer ancestry and Neolithic farmer ancestry.
Why the two disagree, and why that is correct#
Put the two side by side for one imaginary genome:
| Consumer estimate | Ancient-DNA model | |
|---|---|---|
| Reference | living people | published ancient genomes |
| Categories | modern regions | ancient populations |
| Time depth | recent centuries | thousands of years |
| "62% Italian" becomes | the answer | a mixture of farmer, steppe and hunter-gatherer components |
| Can it reject a model? | no | yes (qpAdm) |
| Same result from every vendor? | no | reproducible from the same panel and model |
A person of Italian ancestry should score high on a modern Italian reference and should decompose into Anatolian-farmer, steppe and hunter-gatherer components in an ancient model — because that is what modern Italians are made of. The two results describe the same genome at two different depths.
The three forms an ancient-DNA test takes#
"Ancient DNA test" is a category, not one method, and the honest version of the category offers three different instruments — described together on the ancient DNA test page:
Formal modelling (qpAdm). Your raw file is merged with the ancient panel and modelled as a mixture of chosen ancient sources using allele-frequency statistics. It returns a weight, standard error, z-score and confidence interval per source, a p-value that can reject the model, and the full model record behind it, downloadable as plain text. This is the method most published ancient-DNA papers use. What a qpAdm test is covers it for buyers; ours is the qpAdm analysis from €29.99.
Coordinate analysis (Global25). Your genome is represented as a 25-number coordinate row and compared with ancient and modern reference rows by distance, admixture fitting and PCA. Fast, visual, always returns an answer, cannot reject one. What are Global25 coordinates? explains the input; the Global25 analysis is €29.99.
Segment matching (Ancient Matches). Your file is scanned against every individual in the panel for stretches of shared genome — which particular buried people you overlap with, and where on your chromosomes. This is identity by state, not identity by descent, and the distinction is the whole point of Ancient DNA matches explained. Ours is Ancient Matches, €29.99.
The same raw file works for both#
You do not need a new test. Every consumer company lets you download the raw genotype file behind your ethnicity estimate, and that file is the input to qpAdm and Ancient Matches. The number of markers it contains — roughly 600,000 to 700,000 on current 23andMe and AncestryDNA chips — decides how much of the ancient panel it overlaps, which decides how tight the resulting statistics can be. Before buying anything, run it through the free Raw DNA File Check: it reports what it detected and a plain-language verdict per analysis.
Global25 is the one exception: no testing company produces G25 coordinates, and neither do we. They come from the independent Eurogenes service, either by your own request or through our €15 Coordinate Concierge — see How to get your Global25 coordinates.
What an ancient-DNA test will never tell you#
A consumer test's weaknesses are well known; an ancient-DNA test's are different and worth stating before you pay:
- It never names an ancestor. A source population is a statistical proxy — a group of published genomes whose ancestry profile your genome is compatible with drawing on. It is not a statement that those buried people were related to you.
- It does not do recent genealogy. It cannot find cousins, cannot tell you which grandparent carried which component, and says nothing about the last few centuries in the way a consumer test's "communities" do.
- It is only as good as the file and the model. Coverage bounds the standard errors; the choice of sources and outgroups decides what the numbers mean. That is why the numbers are published in full, and why a passing model is "not refuted" rather than "confirmed" — see How to read qpAdm results.
Which one should you run?#
Run the consumer test to learn which living populations you resemble and to find relatives. Run an ancient-DNA test to learn why — which prehistoric and historical populations, in what shares, produced that resemblance. They are complements, not competitors, and the raw file the first one gave you is the ticket to the second. The terms used above are defined in the glossary.
References#
- Lazaridis, I. et al. (2014). Ancient human genomes suggest three ancestral populations for present-day Europeans. Nature, 513, 409–413.
- Haak, W. et al. (2015). Massive migration from the steppe was a source for Indo-European languages in Europe. Nature, 522, 207–211.
- Mallick, S. et al. (2024). The Allen Ancient DNA Resource (AADR): a curated compendium of ancient human genomes. Scientific Data, 11, 182.
- Novembre, J. et al. (2008). Genes mirror geography within Europe. Nature, 456, 98–101.
