My main tree holds a little over seven thousand people.
It doesn’t feel nearly as satisfying as it did when I was clicking those little green leaves…because I can no longer account for most of it. Somewhere in there is my direct line, maybe up to second great grandparents, now finally being carefully worked and properly sourced. Around it is a great drift of people I added in my first flush of enthusiasm, when Ancestry waved a leaf at me and I clicked it, and then clicked the next one, and the one after that.
That is how you end up with the third cousin once removed of the second husband of your fourth cousin twice removed’s wife in your tree…
Nearly everyone who has built a tree online has done it. The software is designed to make adding people frictionless and removing them slightly awkward, and enthusiasm does the rest. But there comes a point where the tree stops being an asset and starts being a liability — where you cannot answer the only question that matters about any person in it:
Why is this person here?
If I put someone in my tree for a reason — a DNA match I was chasing, a witness who kept reappearing, a household I needed to reconstruct — then they have earned their place, and I can tell you what the reason was. I have many many such floating people – my research branches. But – if I put them there because they were a random hint on a distant person attached to my tree…I probably couldn’t tell you why they were there.
There is something to be said for starting over entirely from scratch. But, I decided on a different approach : this year I started taking people out and going through a process of validating everyone who is under 2nd GGP level first…then working from there.
Duplicates first, then the strays
So the first pass is looking for duplicates. Same person, entered twice, usually because I built them once from a census and once from a marriage record eighteen months apart and never noticed. Merge, check nothing was orphaned, move on.
Then the strays. These are the people with no reason — the in-laws of in-laws, the collateral drift, the whole suburbs of a tree that grew because clicking was easy. They are slowly going.
I will be honest about how that feels. It feels like undoing work. It is not. Every one of those people is still sitting in the records they came from, and if a reason to include them ever appears they can come back in ten minutes with a citation attached, which is more than most of them had the first time.
The speculative branches are detached to form research branches — but they are not deleted
The harder job is different, and it isn’t a deletion at all.
There are long branches in my main tree that are genuinely speculative. Families I think connect to mine.
They will live in the valuable ‘research zone’ rather than on my main tree. Households in the right parish with the right surnames and the right given-name patterns, that I have been working on for years without ever proving the join. They are real research. They are just not proven, and they should not be sitting in my main tree as though they were.
So those come out of the main tree and go into separate research trees, where the whole point of the tree is that nothing in it is claimed.
The obvious hazard is duplication — the same people existing in two trees, drifting apart, each collecting half the sources. I would rather solve that up front than discover it in two years, so the rule I am working to is – to the best of my ability – check a research person doesn’t already exist floating around somewhere.
I would be glad to hear how other people handle this one. It is the part of the plan I am least confident about.
Back to the beginning, one profile at a time
With the tree pruned, the actual work starts — and it starts at the top of my own line.
I am working from my second great-grandparents downward, on the maternal side first. Direct line first. Then their children, then theirs.
Every profile gets the full treatment, and “full” means something specific:
- Every census the person’s dates and residence make mandatory — not the ones I happened to find, the ones they should appear in. If a census is missing, that is a documented gap with a reason, not an absence I have quietly stopped noticing. When I go to a new ‘area’ I check to see what should be available for everyone, and use it as a guide (AI helped create it!)
- Birth, marriage, death and burial, each carrying a real citation. A date sitting on a fact with no source behind it is not evidence. It is something I typed once.
- Newspaper clippings – I have the subscriptions, I should use them.
- Land and deed recordings – as much as I can find – and transcriptions as most are handwritten
- Prison records (yep) or similar.
- The record image saved locally, named so I can tell what it is without opening it, in that person’s own folder.
- A written note on anything unresolved — including the things I would rather not write down, like the two Patricks I have not yet definitively told apart on my father’s side
It is slow. A single profile done properly can take an evening, and there are hundreds of them. But it is the only version of this work that compounds, because everything I do afterwards stands on it.
And I have a running Notion database filled with TODOs as I go. I’ll get to a point where I can filter all the records I need to get from X place…and get them all at once.
What “backed up” actually means here
I have watched enough people lose thirty years of research to be twitchy about this.
The live tree is on Ancestry. The records are on my own machine, one folder per person, organised by side, then birth surname, then parent couple, then person. The local copy and the online tree are meant to say the same thing, and part of every profile session is making sure they do — because a tree that only exists on a subscription website is a tree you are renting.
That sync is the least glamorous part of the whole project and probably the most valuable. When I get hit by a bus, a completely reproducible tree is still available, even if Ancestry no longer exists.
And now, finally, a research log
The genuine change this year isn’t the pruning. It is that I have started logging searches.
For years I recorded what I found. I did not record what I looked for and did not find, which meant that every eighteen months I searched the same collection for the same person and got the same nothing, having entirely forgotten the first two times.
Now every session gets written up: what I searched, where, with which spellings and which date range, and what came back — including, especially, when nothing did. A negative search is a result. It tells the next person, who may well be me in two years, that I’d done this already. I wish I had started doing this years ago. I have Claude scan my log whenever I search something now, to be sure I’m not repeating myself.
There is a version of this hobby where you accumulate names, and a version where you build something another researcher could pick up and continue. I spent a long time doing the first – but I really want to be doing the second. It is one of the reasons this website exists!
Seven thousand people, and I could not tell you why half of them were there. I would rather have a smaller tree I can defend.
If you are working through the same clean-up — particularly the business of splitting speculative branches into research trees without ending up with two versions of everyone — I would like to hear how you are managing it.