Using DNA Matches to Break Through Brick Walls
A practical workflow for turning a list of DNA matches into a documented ancestor: clustering, shared cM analysis and targeted testing.
A DNA test alone rarely names an ancestor. What it gives you is a list of living cousins who inherited segments from the same ancestors you did. The skill is in organizing those matches and combining them with documentary research. Here is a workflow that works.
Choose the right test
- Autosomal DNA (AncestryDNA, 23andMe, MyHeritage, FamilyTreeDNA Family Finder) is inherited from all your ancestral lines and is reliable for roughly the last five to seven generations. This is the test for most questions.
- Y-DNA follows the direct paternal line, father to son. It is excellent for surname questions and unknown fathers.
- Mitochondrial DNA follows the direct maternal line. It changes slowly, so it's mainly useful for testing a specific hypothesis.
Test where the database is largest (usually Ancestry), and upload your raw data to GEDmatch, which compares testers from every company. Most companies no longer accept uploads from their competitors, so to reach another company's matches you'll usually need to test there too. See Which DNA Test Should You Take?
Step 1: Define the question
"Who were Sarah Miller's parents?" is answerable. "Tell me about my ancestry" isn't. Identify the target couple or person, and figure out which of your living relatives descend from that line. Testing the oldest generation available, such as a grandparent or great-aunt, doubles the useful DNA compared to testing yourself.
Step 2: Sort matches into clusters
Group your matches by which side of your family they share. The Leeds Method is simple: list your matches in the 90–400 cM range (roughly second to third cousins), and color-code them by the shared matches they have in common. Most people end up with four clusters, one for each grandparent. Several sites now offer automatic clustering tools.
Step 3: Estimate relationships with shared cM
The amount of DNA you share, measured in centimorgans (cM), suggests a range of possible relationships. Use the Shared cM Project tool at DNA Painter to see which relationships fit a given number. Treat the answer as a probability range: 200 cM could be a first cousin once removed, a half first cousin, or a second cousin.
Step 4: Build your matches' trees
Most breakthroughs come from building your matches' family trees backward until they intersect. Do it in a separate, unattached "research tree" so you don't clutter your main tree. Use ordinary records (census, obituaries, marriages) to extend matches' trees, then look for a shared couple.
Step 5: Test hypotheses
Once you have a candidate ancestral couple, use DNA Painter's What Are the Odds? tool to compare where you would fit under each hypothesis. Then confirm with documents. DNA evidence must be correlated with records like any other evidence under the Genealogical Proof Standard.