Tag Archives: Clusters

How to Use Ancestry’s New Match Clusters and What They Mean

Posted on by
13

Ancestry recently introduced matches by cluster a new ProTools feature that clusters your matches together. And no, before you ask, this is not the long-awaited triangulation.

Ancestry’s new match clusters are a grid-based visualization of your shared matches.

Manual clustering was introduced by Dana Leeds in September of 2018, and, appropriately, named the Leeds Method.

You’ve probably seen similar automated clustering features at Genetic Affairs, MyHeritage, and GEDmatch. Now Ancestry has climbed on the bandwagon, too.

The purpose of clustering is to group your matches that also match each other together. Clustered matches don’t necessarily match on the same segments, which is what defines triangulation.

Cluster members who match each other share common ancestors. Grouping them together in clusters helps you figure out the ancestor for the entire group.

Note that clustering requires a ProTools subscription, in addition to either an Ancestry family history membership or AncestryDNA Plus. Ancestry’s subscription model can be viewed here.

Let’s take a look at Ancestry’s clusters, discuss how clusters work conceptually, and then how clusters can help us with our genealogy.

Viewing Your Clusters

Sign on, select DNA, Matches, and then “By cluster.”

Everyone in a cluster matches you, and many of your matches match each other too.

Ancestry, in their Clustering introduction and support article, explains that:

  • Clustering looks at matches on each side of your family between 65 cM and 1300 cM.
  • They exclude close family members, such as parents, children, siblings, half-siblings, aunts, uncles and their children. Many of the people designated as “Close Family” on my match list are included in clusters, so “Close Family” on your match list is not the same as “close family” who is excluded from clusters.
  • While cousins are very useful to separate matches, close relatives like parents and siblings would be in just about every cluster, so they aren’t useful.
  • Ancestry compares your qualifying matches to see who shares 20 cM or more of DNA with each other
  • People who all match you above 65 cM, AND each other at 20 cM or greater, after Timber, are candidates to form a cluster. Not everyone in a cluster matches everyone else, but everyone matches some other cluster members.

To be clear, Ancestry includes SOME qualifying people in clusters, but not all people who qualify. We don’t know how or why that decision is made, but not everyone who meets these qualifications is included in a cluster.

While that’s frustrating and confusing, the clusters we do have are valuable for determining where those people fit in the puzzle, plus, I’ve developed an easy workaround for those unclustered shared matches.

Let’s view your clusters.

Viewing Your Cluster Members

I have two separate Ancestry DNA tests because I tested on two different chip versions. I have intentionally used all of the tools on one kit, and have not on the other, so that I can see a test “in the raw” compared to one that has utilized all of Ancestry’s available tools.

I refer to the first test as my V1 “tricked out” version, and the second test, V2 is “less developed”.

My cluster results are somewhat different between the two.

Also, every time you sign in or click on “By cluster,” Ancestry recalculates your clusters, so they may be slightly different from one time to the next, or one day to the next. This could be a function of Clusters being in Beta, or maybe not. New matches may also force reclustering, of course, but I haven’t had new qualifying matches since clusters was released. Basically, Ancestry uses between first and third cousins to define clusters.

Roberta’s V2 “Less Developed” Test

I’m starting with the less developed V2 test, because I did NOT use SideView to designate which parent is which by assigning either identifiable unique ethnicity or known matches to a specific parent. Everyone who can should utilize SideView.

Ancestry does their best to assign clusters to one parent or the other, even if you don’t (or can’t) designate parental “sides,” meaning which parent is which.

At the top of the cluster page, you’ll see tabs for “All”, “Parent 1”, and “Parent 2.”

The default view is “All,” so clusters from both parental sides are included in this display, if you have clusters on both sides.

Even though I did NOT use sideview to designate which parent certain matches or ethnicity are from, Ancestry was able to identify some clusters from Parent 1 and some from Parent 2. In total, I have 9 clusters with a total of 92 different people in those clusters.

In this test, you can see the clusters at the top of the page, but my V1 “tricked out” test is different.

Roberta’s V1 “Tricked Out” Test

I have used SideView to indicate parental “sides” using my ethnicity and/or known close matches for this test.

If you have used Sideview to indicate which side is which, then your cluster selections will say “All,” “Maternal,” and “Paternal.”

Notice, though, that this test does NOT show any of my clusters at the top like the V2 test did, just the dropdown description boxes where you can view each individual cluster.

This is because I have more than 100 cluster members, but it’s anything but intuitive and is apparently what the message, “Chart view is available for clusters of 100 matches or fewer,” is trying to tell me. However, I had no idea what “chart view” was, or, without adding the totals from each cluster, that I have more than 100 cluster members. In other words, no one who sees this will know what is missing, or why. Now you know!

Fortunately, I have other tests available from other testers that I could check.

For example, I manage my Acadian cousin’s test. He is heavily endogamous and has more than 700 people in his clusters. His clusters don’t show at the top of his page either. The tests I manage with less than 100 cluster members all show their clusters when they first open their cluster page.

This restriction also pertains to the number of matches within any individual cluster. Essentially, a cluster or combined clusters of 100 people is just about all that can be displayed on a computer screen, left to right.

Regardless, either way, your entire group of clusters is shown together initially, either in the grid format, which they’ve named “chart format” if 100 or fewer, and in a list accessible via dropdowns for everyone.

My Clusters

I’m using my V1 “tricked out” test for the rest of this article because it’s the one where I’ve used all the available tools. Therefore, my best result should be obtained using this test.

Ancestry has created eight clusters for me with a total of 102 members, which is why I don’t see the nice little grid view at the top of my list, but my V2 test with only 92 cluster members displays the chart/grid view.

To view any individual cluster, click on the dropdown box. If the clusters are displayed at the top, scroll down to the dropdown boxes beneath the colorful cluster view.

After clicking the down arrow, here’s a view of my first cluster.

  • All of these people match me at a qualifying level meaning 65-1300 cM and are not a close relative.
  • Not everyone included in a cluster will match each other.
  • The colored cells indicate matches, meaning those people also match each other at 20 cM or greater.
  • The non-colored or “blank” cells indicate that those two intersecting people don’t match each other at 20 cM or more. It does NOT mean they don’t match each other at all, just not above 20 cM which is the lowest amount of shared DNA between your matches that you can see using ProTools.

The first person in this cluster, meaning “DP”, the person in the top row (also the first column), matches everyone else in the cluster. The second person, “ER”, matches everyone except five people in the cluster, and so forth.

Hovering over any colored cell tells you how closely these two individuals are related to each other.

Scrolling down below the cluster displays your match information to each cluster member, including whether they have a tree and their estimated relationship to you.

I label my matches by MRCA, or most recent common ancestor, in the notes field. If Ancestry can identify a common ancestor based on both of your trees, they will note that there is a “Common ancestor,” which is ThruLines. To view additional information, click on that link.

Each cluster can be traced back to an ancestral couple.

My first cluster has 27 members, and I had previously figured out how most of them are related to me, meaning our common ancestor. I had already labeled them accordingly in the Notes field, and also by creating “group labels” for each ancestral couple, which we’ll discuss in a minute.

Of these 27 cluster members:

  • 13 track back to Lazarus Estes and Elizabeth Vannoy
  • 3 track back to John Y. Estes and Martha Ruthy Dodson
  • 4 have private trees
  • 2 have no trees
  • 2 have very tiny trees
  • 2 people share multiple ancestors with me, so they may be in other clusters too
  • The common ancestor of 3 cluster members remains a mystery, but I know this is “how” they are related because they are a member of this cluster

Clusters may contain people with generational differences. For example, it’s very likely that this entire cluster descends from John Y. Estes and Martha Ruth Dodson, but 13 people can only be tracked to Lazarus Estes and Elizabeth Vannoy based on their trees alone. No one in this cluster can be traced to Elizabeth Vannoy’s parents. Since 3 cluster members can be traced to Lazarus Estes’s parents, the entire cluster probably originated with John Y. Estes and Martha Rutha Dodson.

However, for now, I’m assigning this cluster to Lazarus and Elizabeth.

Viewing Maternal and Paternal Groups of Clusters

By clicking on either Maternal or Paternal at the top, you see a view of multiple maternal or paternal clusters, so long as the total number of members isn’t over 100.

I have three clusters on my mother’s side: two fairly large ones, plus the small green third cluster.

I can tell by the matches, some of whom I know, that the purple cluster is my mother’s paternal side, and the blue cluster is my mother’s maternal side.

The green cluster also includes my mother’s paternal line. How do I know this? The grey cells indicate people who are members of both clusters. Grey cells are intersections between two different clusters.

Follow the first person, “DZ”, or any purple person, through the purple cells, across the blue cells to their first grey cell, then directly down to the green cell, and those two people are members of the green cluster too.

So, for any one person, to see how many clusters they are a member of, and who they match in every cluster, just follow their row straight across, left to right. Or straight down, if you prefer top to bottom.

If you’re wondering how someone could be in BOTH my mother’s maternal and paternal clusters, the answer is first cousins and their descendants who descend from both of my grandparents.

Cluster Members in Multiple Clusters

Please note that when viewing cluster members in the cluster dropdown boxes, that:

  • When someone is a member of BOTH the purple and green cluster, they are only listed as a member of ONE cluster, not both.
  • Therefore, any individual person is only listed once, not in each cluster of which they are a member.
  • This also means they are only counted once, not twice.

For example, the person in the first row, “DZ”, is a member of both the purple and the green cluster, but in the cluster dropdown, DZ is only listed as a member of the purple cluster, NOT the green cluster. It’s exactly the opposite for “MF” who is a member of both, but is listed only in the green cluster but not the purple cluster.

Looking back at the image, you can see that everyone in the green cluster is also a member of either the purple cluster, the blue cluster, or all three.

Someone that is a member of two clusters, but only listed in one cluster, was very confusing until I realized what was going on. This makes it unnecessarily difficult to identify clusters and associate them with ancestors.

However, I created an easy workaround.

While listing someone who is a member of multiple clusters in only one cluster makes it difficult to identify ancestors with whom clusters are associated, you can overcome this by creating a separate spreadsheet or chart and manually add the people associated with two or more clusters. Just follow each person’s row across left to right and use the grey squares in the cluster image. Of course, your analysis will reveal WHY they are members of multiple clusters.

This approach works as long as you don’t have more than 100 people on either your maternal or paternal side, respectively. If the page of clusters is larger than 100, you can’t see the multiple cluster image, so you’re out of luck tracking matches in multiple clusters because you need to see those grey cells.

One person who is a member of two clusters means that they are in a cluster for each of two different ancestral couples. For example, let’s say Cousin John is in a cluster for Joe Smith and Jane Johnson. He’s also in a second cluster for Jane Johnson’s parents. Cousin John could be in a third cluster too, for Joe Smith’s parents, or a different ancestral couple on his other parent’s side.

Every cluster has their own unique history and it’s your job to figure out which ancestral couple each cluster represents. .

For example, I’ve scrolled down on my Paternal Clusters to the bottom. I have five clusters, and you can see that many people are members of multiple clusters. Some people are in four clusters, counting the marks in the spaces for the various clusters for each match.

One person is a member of all five clusters, but I happen to know some of my matches descend separately from both sides of my father’s family – so we have pedigree collapse. These people could also be descendants of my aunts and uncles, for example, so we do share all of our ancestors on my father’s side.

It’s easiest to work with clusters if we create cluster groups.

Creating and Using Cluster Groups

Groups allow you to tag someone with various colored group labels that you define for your genealogy.

I created a new group for each of my 8 clusters. You can easily create the new group and tag everyone at the same time by clicking on “Add All” at the top, which opens your defined groups, at right. You can either select an existing group, or create a new one. You can assign this group identifier to everyone in this cluster, or just some people by checking their box (at far left), or not. Remember, your matches are only listed in ONE cluster, so you’ll need to add people into multiple cluster groups manually.

I’m using the grouping feature to track who is in which cluster or clusters, and who is not. Please note that I found assigning a group to everyone in the cluster using the “Add all” feature to be a bit buggy, so check closely to be sure the clusters are recorded correctly and everyone who should be labeled with a group cluster tag actually is. Also, be sure to click on “Save changes” at the bottom.

Returning to my primary DNA match list, now it’s easy to see who is and is not included in a cluster, or multiple clusters based on my group tags.

Of my first four matches, two are maternal and two are paternal, and they are assigned to a purple or a yellow cluster accordingly.

Who’s Missing?

To quote another genetic genealogist, many qualifying matches who clearly meet the cluster criteria “have been left on the cutting room floor.”

I noticed that several of my cousins are missing from my clusters. Known cousins are used to identify matches. While these people clearly don’t fall in the ecluded “Close Family” category, they are certainly close enough to be very useful, first to third cousins, and meet the cluster criteria.

Adding to the confusion, many who match me more distantly, AND match these people, ARE included in clusters.

So, if you think you’re imagining things, no, you’re probably not!

Let’s take a look.

The first person NOT included in a cluster is only my 10th match, “MB” a suggested second cousin with whom I share 238 cM.

You can see that the people both above and below her on the list are included in clusters. Even more confusing is that a ThruLine has been formed, which is what the “Common Ancestor” designation means.

That makes it even easier for me to identify the cluster, so one would think that matches with ThruLines would be a priority to include in clusters.

The second cluster criterion is that the match also matches other people in the cluster with 20 cM or more. Looking at our shared matches, that’s clearly the case. All of our closest common matches are also clustered, but “MB” is not.

I’m baffled.

We have 20 pages of common matches. Of the first 25 matches, 22 are clustered and 3 are not, which is also a bit baffling. All meet the criteria.

One reason that someone might not be clustered is that two matches are too closely related to each other, like parent and child, and the other person is already clustered. But that’s not the situation here. In fact, MBs adult child, my third cousin, is also on my match list and is also not clustered, although people on both sides of MB’s child are in clusters too.

As I work down my match list, by the 5th page or so, there’s little consistency between who is and is not a cluster member. Each match page displays 20 matches. On the 7th page, there are only 5 matches who are clustered, sprinkled between the rest who are not. All of those matches meet the criteria and so do our shared matches.

At the point on my match list where clustering ends, and no one else further down my match list is a cluster member, that person shares 67 cM with me, and they share 20 cM or more with all of our 31 shared matches. Of our shared matches, five share more than 65 cM with me, so no matter how you slice it, we all qualify to be in a cluster, several of us together. In fact, four of the other five are members of cluster 4, but the other two are not.

There are a total of 35 people who match me at 66 cM to 238 cM who clearly qualify to be in a cluster, but who are not. If the threshold is actually 65 cM, instead of “above 65 cM,” there are six more.

Easy Workaround

While having clusters formed with all of the qualifying members would be extremely useful, I’ve found a way to work around it, using my spreadsheet.We are going to use these clusters as seeds to grow into something better.

I’ve identified the ancestral couple associated with cluster members and labeled each cluster with their name. The omitted shared matches between me and cluster members should be in the same or a related cluster, barring issues like pedigree collapse and endogamy.

In this example, Omitted person #1 matches with both DZ and SL in cluster 1, so should be in Cluster 1. Omitted person #2 matches MF and LS in Cluster 2, so they should be a member of Cluster 2. Of course, I’ll be reviewing everyone’s trees and sometimes doing their genealogy for them to uncover our common ancestor.

So, while Ancestry’s clusters may not cluster everyone that they logically should, you can:

  1. Use the clusters that have formed
  2. Combined with shared matches to other cluster members
  3. To further identify, or at least find hints pointing to common ancestors

Now, let’s analyze the clusters.

My Cluster Results

So, what have I been able to do with Ancestry’s clusters?

Ancestors defined by clusters can be identified in multiple ways:

  • Because the tester is known
  • A match has a common ancestor in their tree
  • You extended their tree to find your common ancestor
  • A ThruLine has been formed

I’ve placed cluster numbers on ancestor couples identified as common ancestors with cluster members.

  • 1 cluster descends from my paternal great-grandparents (cluster 1)
  • 3 clusters descend from my paternal great-great-grandparents (clusters 2, 3 and 4)
  • 1 cluster descends from my maternal great-great-grandparents (cluster 6)
  • 1 cluster descends from my maternal great-great-great-grandparents (cluster 5)
  • 2 of the smallest clusters can be identified only to grandparents, meaning just the maternal or paternal side (clusters 7 and 8)

Conversely, that means clusters didn’t develop for:

  • My father’s grandfather, Joseph B. Bolton’s line
  • My mother’s grandfather’s line, Hiram Bauke Ferverda
  • My mother’s grandmother’s father’s line, John David Miller
  • My mother’s grandmother’s line, Ellenora Kirsch

How Can Clusters Kick-Start Your Genealogy?

The answer to how clusters can help you depends, in part, on your goals.

If you’re searching for unknown parents, grandparents, or even great-grandparents, and assuming enough other people have tested, clusters should work well for you.

  • Of my 8 clusters, all can be identified maternally or paternally, meaning those sets of grandparents.
  • Keep in mind, though, that I identified my maternal and paternal “side” through Sideview for my V1 “tricked out” test, so your mileage without having indicated parental “sides” may vary a bit. My V2 test where I did not select sides, still had about 90% of the clustered matches of my V1 test with Sideview. Clusters are essential for people seeking unknown, relatively closely related family members.
  • If you’re searching for unknown parents or grandparents, smaller clusters that include members from several larger clusters, especially all clusters on one of your parent’s sides, may be pointing to grandparents.
  • Please note that clusters always identify a couple, not an individual. As soon as you can identify which one of the couple by matching with someone who descends from one of that ancestors’ siblings, then you’re automatically bumped back another generation to their parents.
  • You may only be able to identify a cluster match to a generation closer in time.
  • Remember that Ancestry’s clustering is not triangulation, so your matches may not match on the same segment. You could match person A due to one set of ancestors, person B due to another set of ancestors, and A and B could match each other due to a third set of ancestors.
  • This, in part, is why clustering is useful, as it reduces, not eliminates, the possibility of that happening because you’re dealing with groups of people, not just 3, multiple match criteria, and larger size segments.
  • When pedigree collapse or endogamy is involved, the three (or multiple) people may match due to different ancestors that they can’t identify because the group of matches shares multiple or many ancestors. Think of either first cousins marrying each other a couple of generations ago, which is pedigree collapse, or endogamous groups like Acadians or Jewish people, isolated cultural groups who intermarried for generations.
  • Triangulation, which clusters are NOT, further reduces ambiguity because the same segment of DNA is being measured and compared. Ancestry does not offer triangulation, but both FamilyTreeDNA and MyHeritage do in various ways.

If you’re looking to collaborate with genealogists who descend from ancestors in the past three generations, contact matches in formed clusters.

If you’re looking to break through a recent brick wall, you may be able to do that. In part, it’s a roll of the dice depending on who has tested, the size of the testing pool where your ancestors are from, combined with the unknown internal Ancestry algorithm. For example, if you descend from ancestors in an under-tested part of the world, you may have fewer or even no clusters. To aid in breaking down brick walls, utilize clusters that do form as seeds to group additional people using your cheat sheet.

If you’re an experienced genealogist trying to break through a distant brick wall, Ancestry’s clusters, as they are today, probably aren’t going to help you much, but never say never. You don’t know where that desperately needed next hint might come from. If you’re hunting for the identify of a 4th great-grandparent, pay close attention to the common ancestors of the people in your closest cluster to that unknown ancestor in your tree.

Work on each cluster. If you find a cluster you can’t attribute to one of your ancestors, compare the ancestors in the trees of each cluster member, looking for commonality. Ancestors shared between them and not you may point to your brick wall..

Use the clusters as a starting point, and continue working down your match list. Use shared matches with cluster members to continue to associate your matches with clusters, even if Ancestry doesn’t assign them. Your cheat sheet spreadsheet is your friend, and so are notes and grouping tags.

Beta

If things aren’t working quite right, remember that Ancestry’s clusters are in Beta. Just try again later.

Ancestry has also noted that they are rolling this feature out in stages, and some members won’t be able to access Clusters until December 2025.

Ancestry has announced that soon you’ll be able to create custom clusters with specific matches and cM ranges.

Try Additional Cluster Resources

Each vendor has a different pool of people who have tested there.

Other vendors and third-party tools provide cluster resources and various types of automated tree-building. I have between 18 and 40 clusters using these various tools at different places.

Take a look and see how many clusters you have, and what you can do with them at:

_____________________________________________________________

Share the Love!

You’re always welcome to forward articles or links to friends and share on social media.

If you haven’t already subscribed (it’s free,) you can receive an e-mail whenever I publish by clicking the “follow” button on the main blog page, here.

You Can Help Keep This Blog Free

I receive a small contribution when you click on some of the links to vendors in my articles. This does NOT increase your price but helps me keep the lights on and this informational blog free for everyone. Please click on the links in the articles or to the vendors below if you are purchasing products or DNA testing.

Thank you so much.

DNA Purchases and Free Uploads

Genealogy Products and Services

My Books

Genealogy Books

Genealogy Research