What Are Shared DNA Matches?
A shared match — sometimes called an "in common with" match — is someone who appears on both your match list and one of your matches' lists. On its own that sounds like a footnote. In practice it's the single most powerful tool on any DNA testing site, because it turns an undifferentiated list of two thousand strangers into recognizable family groups. If you and Match A both match Match B, C and D, those people are probably connected to you through the same branch — and that's how unknown matches get identified. This page covers how to read shared matches, how to group them, and the traps that catch people out. To convert the numbers into likely relationships, the DNA match calculator is the companion tool.
What Exactly Is a Shared Match?
If you match Person A, your testing site can show you which of your other matches also match Person A. Those are your shared matches with A.
The logic behind why this is useful: DNA is inherited in branches. You get roughly half your DNA from each parent, a quarter from each grandparent, and so on. Anyone who matches you on your father's mother's line will tend to match other people on that same line — because they all descend from the same set of ancestors.
So a group of people who all match each other and you almost certainly connect through one branch of your tree. You don't need to know how; you just need to see the grouping.
Where to find it: every major testing service offers this, under names like "shared matches", "in common with", "relatives in common" or "DNA relatives in common". The mechanics differ slightly but the principle is identical.
The thresholds differ too, and this matters. Some sites only show shared matches above a certain size — commonly around 20 cM — so a small match may show no shared matches at all even though connections exist. That's a limitation of the display, not evidence of anything.
If the terminology here is new, DNA matches explained and centimorgans explained cover the groundwork.
How Do You Use Shared Matches to Identify Someone?
The standard method, and it's more mechanical than people expect.
Step 1: Start with a known relative. Pick a match you've already identified — a confirmed second cousin, an aunt, anyone whose relationship you're certain of. This is your anchor.
Step 2: Look at their shared matches with you. Everyone in that list connects to you through the same part of the tree as your anchor does. If your anchor is a maternal cousin, this whole group is maternal.
Step 3: Repeat with anchors from other branches. Ideally you want one confirmed match from each grandparent line, which lets you divide most of your list four ways.
Step 4: Place the unknown match. Take the person you're trying to identify and look at their shared matches with you. Whichever cluster they overlap with tells you which branch they're on.
Step 5: Narrow within the branch. Compare the amount of shared DNA against the likely relationships — the DNA match calculator converts centimorgans into probable relationships, and how much DNA do cousins share gives the expected ranges.
Step 6: Build their tree, not yours. Work forward from the common ancestors your cluster points to, and look for where the unknown match fits. This is often the real work, and shared matches are what tell you where to do it.
The single biggest accelerator is testing close relatives. A parent, aunt, uncle or older cousin immediately splits your match list in a way nothing else can — anyone matching your mother is maternal, full stop.
What Is Clustering, and How Do You Do It?
Clustering is shared matching applied systematically, and it produces the clearest picture most people ever get of their match list.
The idea: group your matches so that everyone in a group shares matches with each other. Each group corresponds to one ancestral line.
Doing it by hand, which works fine for the top hundred matches:
- List your strongest matches in a spreadsheet
- For each one, note who their shared matches with you are
- Group people who overlap heavily
- Label each cluster once you can identify anyone in it
Doing it with tools. Automatic clustering — the Leeds Method and the various colored-grid tools built into or alongside the testing sites — does this in minutes rather than evenings. DNA Painter guide covers the visualisation side.
What you should see: most people get four main clusters, one for each grandparent line, plus smaller satellite groups. Seeing four clean clusters is itself confirmation that your understanding of your ancestry is broadly right.
What the clusters mean: each represents descendants of one ancestral couple. You may not know who that couple is yet — but you know the group belongs together, and identifying any one member labels the whole cluster.
Common cluster patterns:
| What you see | What it usually means |
|---|---|
| Four clean clusters | Normal — one per grandparent line |
| Only two or three | Some lines under-represented in the database |
| Many overlapping clusters | Endogamy, or a recent shared ancestral community |
| A large unexplained cluster | An unknown branch — often the interesting one |
| One person in several clusters | Related through more than one line, or a data artefact |
Where Does This Go Wrong?
The traps, in rough order of how often they catch people.
Shared matches are not triangulated segments. This is the big one. If you and A both match B, that does not prove you all share the same segment of DNA. You might match B on chromosome 4 while A matches B on chromosome 17, through a completely different ancestor. Shared matching groups people; it doesn't prove a common segment. That's what triangulated groups are for.
Small matches mislead. Below roughly 20 cM, a shared match may be coincidental rather than genealogical. Below 10 cM, a substantial proportion of matches are false positives. Build your clusters on strong matches and treat small ones as supporting evidence at best.
Endogamy breaks clustering. In populations that intermarried over generations — Ashkenazi Jewish, Acadian, Amish, Mennonite, island and isolated rural communities — everyone matches everyone. Clusters merge, shared DNA amounts overstate closeness, and standard methods need heavy adjustment. This isn't a fault in your analysis; it's a property of the population.
Double relationships. If your grandparents were cousins, or two siblings from one family married two siblings from another, matches land in more than one cluster legitimately.
Absence proves little. No shared matches with someone may mean the site's threshold hides them, that the connection is distant, or that few relatives from that line have tested. It doesn't demonstrate the match is spurious — and a genuinely missing expected match has its own explanations, covered in why don't I match a known relative.
Different sites, different answers. Each company tests a somewhat different set of markers and applies different thresholds, so shared match lists don't align perfectly across services. Uploading your raw data to more than one site where permitted broadens the picture considerably.
How Do the Testing Sites Differ?
Worth knowing, because the feature behaves differently depending on where you tested, and people often assume a limitation is a finding.
AncestryDNA shows shared matches down to about 20 cM, and its clustering-adjacent tools group matches by likely common ancestor. Its database is much the largest, which usually means more matches and better cluster coverage. It doesn't offer a chromosome browser, so you can't confirm segments there.
23andMe shows relatives in common and does provide a chromosome browser, so segment comparison is possible. Its database is large but oriented more toward health customers, and many users don't build trees.
MyHeritage shows shared matches with a lower threshold, plus a chromosome browser and built-in triangulation. Strong European coverage, which matters if your lines are outside the US.
FamilyTreeDNA offers shared matching, a chromosome browser and segment tools, plus Y-DNA and mitochondrial testing that the others don't. Smaller autosomal database.
GEDmatch isn't a testing company but a comparison site — you upload raw data from elsewhere and compare against people who tested with different companies. That cross-company comparison is its whole value, and its tools for shared matching and triangulation are more granular than most.
The practical implication: if you're stuck, uploading your raw data to additional sites that accept transfers is usually more productive than analyzing the same match list harder. A cousin who tested at a different company is invisible to you until you're both on the same platform.
What Should You Actually Do With Your Match List?
A practical workflow, in order:
1. Test or recruit close relatives. A parent is worth more than any analysis technique. Failing that, an aunt, uncle, or a first cousin on the line you're researching.
2. Identify three or four anchors — matches whose relationship you're confident about, ideally one per grandparent line.
3. Cluster your top 100 to 200 matches, by hand or with a tool.
4. Label every cluster you can, using the anchors.
5. Attack the unlabelled cluster. That's usually where the discovery is. Build the trees of its strongest members forward and look for the point they converge.
6. Note the shared amount for each significant match and check it against expected ranges before committing to a relationship. Numbers narrow possibilities; they rarely prove one.
7. Write down what you conclude and why. DNA reasoning is easy to reconstruct wrongly six months later, and the chain of inference matters as much as the conclusion.
8. Contact people. A short, specific message — "we share 187 cM and both match the Hutchings cluster; do the names Hutchings or Barlow mean anything?" — gets far more replies than a generic hello.
Frequently Asked Questions
What does "shared matches" mean on a DNA test?
Shared matches are people who appear on both your match list and one of your matches' lists. Because DNA is inherited in branches, people who all match each other and you usually connect through the same ancestral line — which is what makes the feature so useful for grouping.
Do shared matches mean we all descend from the same ancestor?
Usually, but not necessarily. Shared matching shows that each pair matches; it doesn't prove you all share the same DNA segment. Proving a common ancestral segment requires triangulation, which compares the actual chromosome positions.
Why do I have no shared matches with someone?
Most often because the match is small and falls below the site's threshold for displaying shared matches — commonly around 20 cM. It can also mean few relatives from that line have tested, or that the connection is distant.
What is DNA match clustering?
Grouping your matches so that everyone within a group shares matches with each other. Each cluster generally corresponds to one ancestral line, and most people see four main clusters — one per grandparent line — plus smaller satellite groups.
Why do all my matches seem related to each other?
Usually endogamy — descent from a population that intermarried over many generations, such as Ashkenazi Jewish, Acadian or isolated rural communities. Everyone matches everyone, clusters merge, and shared DNA amounts overstate how closely related people actually are.
How do I identify an unknown DNA match?
Look at which of your existing clusters they share matches with, which tells you the branch. Then compare the shared centimorgan amount against expected ranges for candidate relationships, and build their tree forward from the ancestral couple the cluster points to.
A note on what to expect. Clustering a match list is genuinely satisfying, and it is not the same as solving anything. It tells you where in your tree to look, and the identification still comes from ordinary genealogy — building trees, reading records, contacting people. DNA narrows the search; documents close it.
Shared matches are what turn a list of strangers into a map. Find three or four anchors you're sure of, cluster the rest around them, and work on the group you can't explain — that's where the answer usually is. Just remember that grouping isn't the same as proving a shared segment. Convert the numbers into likely relationships with the DNA match calculator.
