Ashkenazi ancestry is one of the most reliably detected things in consumer genetics. Sephardic ancestry is one of the least. Most people don't realise these are completely different problems — and that the percentage on their screen was produced by very different evidence in each case.
Why Ashkenazi ancestry is easy to see
Ashkenazi Jews passed through a severe population bottleneck: the community descends from a small founding group, on the order of a few hundred effective founders, roughly 600 to 800 years ago. Then it grew very fast, mostly marrying within itself.
That history leaves an unmistakable signature. Two Ashkenazi strangers typically share long stretches of DNA inherited from those founders, so they look to the software like distant cousins. Even a small amount of Ashkenazi ancestry stands out, because it arrives with those shared segments attached.
Why Sephardic ancestry is hard to see
Sephardic, Mizrahi and North African Jewish populations did not go through the same kind of bottleneck. They are larger, older and genetically closer to their neighbours — to Iberians, Italians, Greeks and people of the Levant.
So when a consumer test reports "Sephardic," it is usually picking up generic Southern European or Near Eastern ancestry that looks similar. The algorithm cannot tell your Spanish great-grandmother from a Sephardic one, because in most of the genome they look alike.
The claim you'll see everywhere
You will find, all over the internet, the statement that "about 20% of Spaniards have Sephardic ancestry." It comes from a 2008 study by Adams and colleagues that looked at Y-chromosome markers.
It has been heavily criticised, for a clear reason. The markers used cannot distinguish Jewish ancestry from the Near Eastern and North African ancestry that reached Iberia by many other routes — Phoenician traders, Greeks, Romans, Byzantines and, above all, eight centuries of Islamic rule. Some of that "20%" is Jewish; much of it is not; the method cannot tell which. Later work puts the figures far lower. Quote the 2008 number, if you quote it at all, with that criticism attached.
The more useful question is never "what is my percentage?" It is "how much do I share with Jewish references compared with someone of my background who has no Jewish ancestry?" That contrast is what we try to measure — explained in how we compare you to medieval Jews.
Sources
- Adams et al., American Journal of Human Genetics 83 (2008) — cite with its critique
- Studies of the Ashkenazi founder bottleneck (e.g. Carmi et al., Nature Communications 2014)
- Chacón-Duque et al., Nature Communications 9:5388 (2018) — haplotype-based converso estimates against Iberian controls