AncestryDNA has just rolled out enhancements to their Clustering Program that let you “Create custom clusters”. At AncestryDNA > DNA > Matches > By Clusters/Pro > Create custom clusters. You must have the additional subscription for ProTools to access this program. I have not run it through its paces yet, but I wanted to review the Walk The Clusters Back (WTCB) concept, and ask for feedback on your experience with it.
The concept of WTCB is to adjust the cM range to focus on two generations at a time. The idea is to “solve” the Clusters for close relatives and then adjust the range down to include Matches in the next generation back, and then see where the Clusters separate into more distant Clusters. Start easy with a range of 90-400cM which is the recommendation for the LEEDS method to determine four groups. This would be roughly four Clusters with each one focused on a separate grandparent. Tag (by Dots or by Notes) every Match to the appropriate grandparent. Then drop the range to, say, 70-200cM to get mostly Clusters that include Matches who are 1C on a grandparent, and 2C on a Great grandparent. I don’t know of anyone who has found a “sweet-spot” range for each generation, and I suspect it might be different for each of us. The last time I did this WTCB I had to “fiddle” with the ranges – and never could find any range that gave me only Clusters with Matches from only two generations in each Cluster. So, get used to that.
The point is to notice when some Matches you’ve tagged to an Ancestor, then show up in different Clusters based on a new range – and then determine which sides are represented by the new clusters. Then tag all of those new Matches appropriately. Example: you have a Cluster with 20 Matches that is focused on a Great grandparent. Tag all the Matches with that grandparent (if not already tagged as a closer Ancestor). Adjust the range to add more Matches. Look in the new Clusters for the previously Tagged Matches – hopefully there are two new Clusters, but maybe three. From my experience there may be two Clusters with 15 to 25 Matches, each of which include some of the 20 Matches from the previous run. These new Clusters would represent the next generation back and the focus would only be one of the two parents of the previous Cluster.
Yes, it gets harder and harder with each new generation. The good news is that a Cluster with known Matches from one generation, can only morph into Clusters going back from that one Ancestor. This reduces the genealogy effort . If you’ve reviewed all of your ThurLines (and used ProTools to add even more Matches), you have likely tagged a lot of Matches out to 6C. So as the 4C and 5C and 6C Clusters start to form (as you reduce the cM range), you may already see the Ancestor for the new Clusters by looking at the Notes.
Use your judgment, and fiddle with the cM ranges. Please report back on your experience, and/or if you find a sweet spot for some range. Note that the sweet spot should include two generations – the one you’ve figured out and the next one you are working on.
[19P] Segment-ology: Walk The Clusters Back at AncestryDNA by Jim Bartlett 20251017
segment-ology 
Hi Jim — I, too, received Ancestry’s Custom Clusters capability this week, and have been experimenting with it. I picked a known 2nd cousin and asked it to give me a cluster that that cousin is in with a certain range of cMs, but instead it gave me three. I don’t understand why there are three. They are basically the same, but with different numbers of people in them. Without custom clustering, there are 4 people in the cluster; they are all either siblings or 1st cousins to each other and 2nd cousins to me. The three custom clusters add, respectively, 2, 3, and 6 additional matches to these basic four. The additional 2 are included in the 3, and the 3 are included in the 6. What is this trying to tell me?
Bill Josey
No matter who you are, the only value of your words and deeds is the love you leave behind.
LikeLike
Ho 17 corrispondenze dna con triangolazione che cambia da 7,1fino ha 9cM con persone che almeno una di loro e rom al 100% e a gruppi genetici dei rom.e altre 12 anno gruppi genetici dei rom.io nel mio albero fino ha 5 6 generazioni non ho cognomi che mi collegano ha loro.io sono Sud puglia e loro rom est europa almeno 13 di loro saranno tra rom e mezzi rom.in 3 mesi ho ricevuto altre due corrispondenze che si aggiungono a questo gruppo triangolato già esistente.con alcuni di loro 7 di loro ho solo un singolo segmento condiviso che va da 8fino ha9,4cM con un singolo segmento che triangola.Ma condividiamo piu di 10corrispondenze che vuol dire endogamia?.poi ho altre 9 corrispondenze che ho due segmenti condivisi con loro i più grandi triangola No con altri.poi ho una corrispondenza con tre segmenti il più grande triangola con altri.che significa?
Il ven 17 ott 2025, 20:52 Kevin Capodivento capodiventokevin6@gmail.com ha scritto:
LikeLike
Kevin, When you use such small segments – particularly under 10cM – there is a very high probability that the Common Ancestor is 10 to 20 (or more) generations back. For me, that would be well beyond my genealogy “horizon”! It would be well beyond most records. It would include over 1,000 potential Ancestor just 10 generations back; and over a million Ancestors 20 generations back. That far back, your Ancestors could come from almost anywhere… This does not necessarily mean endogamy, but it is highly likely that at even 10 generations back there was some intermarrying – what we call Pedigree Collapse. We all have some Pedigree Collapse in our Ancestry – there were not enough humans on the planet to have all of our Ancestors be different people. If I were you, I’d be moving to a differnt Triangulation to work on… Jim
LikeLike
Ma continuo a avere nuove corrispondenze che vanno come ingranaggio con il tg già esistente.Ma tu dici che è molto lontano.Ma tu dici che con queste persone che sono tra rom e mezze rom ce un antenato molto lontano?perché le corrispondenze di questa triangolazione hanno molti cognomi proprio di quelle popolazioni e loro hanno quelle origini ho piene ho mezze.
LikeLike
Kevin, As I understand the translation of your comment: you are still finding more new Matches that Triangulate on this segment. That is very common! For such a small segment, I would anticipate many Matches to be added. That is the way of DNA segments – you will have few Triangulations with very large segments, because we gernerally don’t have many close cousins. The smaller the shared segments, the more Matches we have. At Ancestry, I have over 100,000 Matches – over half of those are under 10cM. It doesn’t make any difference if they are Roma or not Roma – at 10 to 20 generations back, they could be from almost any population. A segment of your DNA is from one of your parents, one of your grandparents, one of your Great grandparents, etc (one Ancestor at each generation going back). Your Matches under 10cM could be from a Common Ancestor 15 generations back. Do you know your Ancestors 15 generations back? It would be an almost impossible task for me to figure out with my Ancestry… Jim
LikeLike
Ma tu dici che ci sono altre etnie?non solo quella popolazione dei rom?anche se tra di loro hanno si altre etnie ma hanno tutti asia meridionale e tutti gruppi genetici dei rom.mentre io con alcuni di loro ho medio Orientale e greco albanese che è la principale etnia proprio delle popolazioni rom dei gruppi genetici su mhyeritage.io su gedmach Oracle ho un Po di asia meridionale..
LikeLike
Jim.che vuol dire questa triangolazione con posizione genova iniziò e fine?puoi spiegarmi.
LikeLike
Kevin – all I can explain is that Triangulated segments represent your own segment of DNA from an Ancestor; and the Matches should have this same Ancestor, too. Jim
LikeLike
Si quando faccio confronto io e due corrispondenze alla volta il segmento ha volte e 9cM e ha volta diventa tra 8,5cM fino ga 7,1cM quando confronto io +2 corrispondenza alla volta.e certe volte hanno stesse posizioni iniziò e fine uguali.poi quando aggiungo fino ha 7corrispondenze diventa il segmento triangolazione su mhyeritage fino ha 7,1cM.e la posizione del genoma delle triangolazione cambia ma ha volte e tra 82678223m-84683203.
LikeLike
Kevin, Yes! You will almost never get the same shared DNA segment with different Matches – your got your segment down several generation of random recombination, starting with a large segment and reducing down to a small segment. The same thing happens for each Match – starting with the same Ancestor and randomly getting smaller with each generation. It’s actually quite amazing that we get an overlapping segment at all. And all the shared segments are different, but overlapping. It’s like paint a wall – each brush stroke covers part of the wall, taken all together, they paint the whole wall. The whole wall is your DNA from that one Ancester. The shared segments show just part of your DNA segment from the Ancestor. Taken together, they all show the full segment you got from the Ancestor. If you had access to one of your Matches (like at GEDmatch), you could look for *their* overlapping/Triangulated segments, and find the DNA segment they got from the Ancestor – it would be different from the full segement you got. So your Triangulated segment generally goes from smallest location to the largest location among all the segments that Triangulate. Jim
LikeLike
Grazie jim.sto imparando molte cose.queste corrispondenze che ho 17 in tutte tra di loro due coppie sono mezzi cugini primo grado e poi altri tra di loro sono cugini tra 2 3 4 5 grado secondo mhyeritage e con me tra 4 5 grado secondo te discendono da diversi figli antenati comuni.Come faccio ha vedere se discendono tra più figli da antenati comune?
LikeLike
Kevin, You have to rely on genealogy! The amount of shared DNA can give your some rough estimates as to how they are related to you and to each other, but beyond that you have to look at their Trees and do genealogy. Except for very close relationships (parent/child; sibling; aunt/uncle/niece/nephew), the DNA by itself is not precise enough… Jim
LikeLike
Kevin – review this post: https://segmentology.org/2016/02/05/anatomy-of-a-tg/
LikeLike
Kevin, Here is another refresher: https://segmentology.org/2025/09/15/boundaries-of-a-triangulated-segment/ and the blog post that comes after this one. Jim
LikeLike
Jim… I started playing around with the new custom-cluster tool on Ancestry… I am now able to obtain a larger cluster at the 2C and 3C levels (which also includes some double-3C) relationships since I am dealing with a double cousin scenario… One thought I initially had it would be interesting to extend the cluster box to a 3D type of graph- were height of square is the amount of DNA shared… All this may or may not be a good idea- does anyone out there have suggestions on how to do this… thanks George
LikeLike
George – I think the trick is to get both 2C and 3C in the cluster – adjust the upper threshold to take out any of your 1C; and adjust the lower threshold so it doesn’t include any 4C. You want to try to figure out the 3Cs based on a parent of the 2C. If you have a consensus, then Tag and/or Note and or jot down all the unknown Matches with this some 3C Ancestor – as Likely. Then adjust the upper threshold to eliminate the 2C; and adjust the lower threshold to add in some more Matches – hopefully in the 4C range. Repeat the analysis. I’ve got a feeling…. there won’t be a universal set of thresholds. I’m hoping for more feedback so we can establish a range – like the Shared cM Project does… Jim
LikeLike
I saw your post, and immediately jumped on to Ancestry, where I already have access to the custom clusters. Have done some playing, and am not yet super-excited.
I chose a match of my Mom’s whose CA is “Ben” (Mom’s 2G granddad) only the match “J” descends from Ben’s 2nd marriage. Ancestry created 6 clusters — that was unexpected. Two were just of matches descending from Ben’s 2nd marriage. One cluster included CA’s from Ben’s 1st and 2nd marriages, as well as some matches where the CA is Ben’s parents. Another cluster included descendants from both Ben’s marriages — and also his daughter (my mom’s 1G grandmother). Final cluster included matches descending from both Ben’s marriages and then his paternal grandparents.
I happen to match “J” as well so when I clustered on my own set of matches, although “J” was the match of interest, “J” only matched one person in the cluster, who matched 40 other folks in a tight cluster. Based on the research I’ve already done, the tight cluster folks likely relate to me/Mom via Ben’s mother. If the minimum match could be 15 cM rather than 20 cM that tight cluster might be more useful because it might break apart a bit.
Am looking forward to hearing your take when you’re ready to share, and that of your other readers, as I might be missing something.
LikeLiked by 1 person
Like you, I played a little bit, and it wasn’t a breakthrough. Perhaps I don’t have the hang of it yet. When I did WTCB a few years ago, I was able to impute to almost all my Matches down to 20cM. They are holding up well. I have been doing a lot with ProTools in the world of under 20cM… What I’m hoping is that I can again get down to most of my 20cM Matches and that that will help me with the rest of the Matches. A secondary goal is to try to link my Matches down to 20cM to extant TGs. With 20cM mostly focused on 4C1R/5C/6C, this stays in the ThruLines range, and I have thousands of Matches with MRCAs in that range at Ancestry. At the 6C range, that’s an average of 3 of my 372 TGs for each Ancestor; or 6 per couple. It’s going to be a struggle to try to like 6 TGs to each Cluster at that level, but I should be able to link some more than I have now. Onward… Jim
LikeLike
“Coming soon”
LikeLike
I am in the “coming soon” group anticipating this new feature. But for match groupings I always first suggest any “mature” testers (like ourselves) first eliminate all significantly younger matches since the 1R, 2R, 3R will completely disrupt your groups. My ranges and “sweet spots” are as follows:
2C 12-569 SS=65-165
3C 9-169 SS=30-130
4C 6-135 SS=24-100
LikeLiked by 2 people
pdtbill, thanks for posting the ranges you find useful. I am also mature enough that eliminating younger matches would be helpful! LOL I was puzzling over what you meant by 1R, 2R, 3R, but it just came to me that R means Removed. I’m just not used to seeing it by itself. Doh!
LikeLiked by 1 person
I am in the “coming soon” group anticipating this new feature. But for match groupings I always first suggest any “mature” testers (like ourselves) first eliminate all significantly younger matches since the 1R, 2R, 3R will completely disrupt your groups. My ranges and “sweet spots” are as follows:
2C 12-569 SS=65-165
3C 9-169 SS=30-130
4C 6-135 SS=24-100
LikeLike
Morning Jim – I did a lot of work on this back in 2020. I need to revisit it and will report back soon!
Preliminary play on the Ancestry new custom clusters was not what I had hoped for. Only goes down to 20cMs but haven’t seriously played with it yet.
LikeLike
While I have been able to run the clustering tool with my Pro Tools subscription for some time, I do not see any link to customizing it relative to a particular individual or for a selected cM range. I see the description on the Ancestry (or AncestryDNA) site, but no hotlink. Very frustrating.
LikeLike
I think it just rolled out yesterday, and required me to run the Clusters again from scratch and then let me change things…. Jim
LikeLiked by 1 person
There is now a new dimmed “Custom” button next to All, Paternal, and Maternal buttons. Hovering shows: “Coming soon! Create custom clusters with specific matches and cM ranges.” Looking forward to trying. Joe
LikeLiked by 1 person
Ho 17 corrispondenze dna su mhyeritage con triangolazione il segmento triangolato cambia tra 7,1-9cM e continuo a avere sempre corrispondenze che si aggiungono ha questo gruppo già triangolato tu che dici non.
LikeLike
Once you form a TG, with at least two cousins, that is solid. You can continue to add more Match shared segments, and long as they overlap and match you and at least one other Match in the TG. I have TGs with over 100 people in them. When you have a small, tight, TG under 10cM that keeps growing a lot, you should be careful. It might be a special arrangement of your maternal and paternal DNA in such a way that almost anyone could match by zig-zaging between you mother’s and father’s SNPs. Be sure each of the Matches match *two* other Matches on the same segment. If it still holds, it may well be a pile-up area – it’s really part of your DNA, but it comes from a very distant Ancestor. Like the Nearderthal tiny segments I have – not much good for genealogy research. Jim
LikeLike
Si questo tg in 4 mesi ho avuto altre due corrispondenze che si sono aggiunti al tv esistente.io ho il test di mia madre ma a lei non corrispondono,e dopo che ho testato mia madre continuo a avere nuove corrispondenze che si aggiungono a questo tg già esistente,insomma corrispondono questi tg solo al mio test.
LikeLiked by 1 person