Every Christmas, thousands of DNA kits are unwrapped; every January, thousands of results arrive; and by February I'm having the same conversation over tea: "It says I'm 12% Scandinavian. We don't have any Scandinavians. Is the test wrong?"

The test isn't wrong. The marketing just points you at the least reliable page of your results. I have no affiliation with any testing company, so here is the independent version.

Ethnicity estimates: interesting, not evidence

That coloured pie chart is an estimate, produced by comparing fragments of your DNA against the company's reference panels of modern populations. Three things follow:

  • It changes. Companies update reference panels; your "12% Scandinavian" can quietly become "4% Norwegian" next year without your DNA changing at all.
  • Borders blur. "English" vs "Scottish" vs "North-Western European" distinctions sit at the edge of what the science can do. Broad strokes (British Isles vs Southern Europe) are solid; fine percentages are weather forecasting.
  • It looks back ~500–1,000 years, mostly. Your Viking percentage isn't a secret grandfather; it's the long sediment of medieval migration.

Enjoy the pie chart. Don't build on it.

Cousin matches: the part that finds people

The page most users skim past — DNA matches — is the genuinely powerful one. The test compares your DNA against every other customer and lists people who share segments with you, measured in centimorgans (cM). The more you share, the closer the cousin:

  • ~3,400 cM — parent or child
  • ~2,500 cM — full sibling
  • ~850 cM — first cousin (or half-sibling territory — ranges overlap)
  • ~230 cM — second cousin
  • under ~90 cM — third cousin or further, where the haystack starts

A match doesn't tell you how you connect — but shared matches cluster into family lines, and those clusters can be mapped against a documented tree. This is how unknown parents are identified, brick walls fall, and — fair warning — how family surprises surface.

When DNA says something unexpected

A parent who isn't a match. A first-cousin-level stranger. An ethnicity that genuinely cannot fit. These discoveries are more common than anyone admits at the dinner table, and they deserve two things: verification (lab mix-ups are rare; misread results are not) and care (living people, old decisions, and other families are involved). Go slowly, confirm with records, and decide who needs to know before you tell anyone.

The method that actually works

DNA alone builds nothing. Records alone can't see past an adoption or an unrecorded father. The method is the loop:

  1. DNA suggests a connection (a cluster of matches around one line)
  2. Records test it (censuses, certificates, parish registers)
  3. The confirmed segment anchors the next suggestion

Run that loop patiently and trees grow through walls that records alone could never breach.

Staring at a results page right now? My DNA Discoveries Session (£95) is an hour of independent walkthrough — your ethnicity estimate translated, your top matches grouped, and a written plan for which threads to pull. And if a surprise has just landed in your inbox: you're not the first, it will be handled with discretion, and a free 15-minute call costs nothing.