Why 30-40% of public family trees are wrong (and how to fix yours)

Studies of public family trees on Ancestry and FamilySearch consistently find that roughly a third of records contain at least one significant error. Most of those errors are inherited, not made — copied from someone else's tree decades ago and never re-examined. Here are the five most common patterns, why they spread, and how to audit your own.

Published May 11, 2026 · FamilyTreesAI

The size of the problem

Genealogists have been writing about tree accuracy for years, and the conclusions are remarkably consistent. A 2014 review of public trees by professional genealogist Robert Charles Anderson sampled colonial American lineages and found error rates above 30% — frequently because a single 19th-century mistake had been copied forward into thousands of modern trees. More recent academic work on collaborative genealogy platforms shows similar patterns: errors don't just appear; they propagate.

The bigger your family on these platforms, the more likely you've inherited a mistake from someone who built their part of the tree before you were born.

The five most common error types

1. Impossible biological dates

A mother is recorded as giving birth at age 9, or at 63. A father dies eight years before his recorded child is born. These mistakes happen when a date is mistyped, when two people of the same name are merged, or when a generation has been quietly skipped. They're the easiest to catch and the most reliable indicator that something nearby is also wrong.

2. Duplicate ancestors

Two records exist for the same person — usually because their name was spelled slightly differently in two different sources, or because two researchers built the same ancestor from different starting points and never noticed the overlap. The tree branches and you end up with two great-great-grandfathers when you only had one.

3. Missing or wrong parents

A child is attached to the wrong couple — typically because two families in the same town had children with the same name, and the wrong one got linked. This is the error type most likely to be inherited from another researcher's tree without anyone re-verifying the source documents.

4. Conflicting marriages

Two people are recorded as married, but one of them is also recorded as married to someone else on the same date. Or a marriage is dated before either party was born. These appear when census, vital records, and family lore disagree and nobody resolves the conflict.

5. Geographic impossibilities

A child is born in Oregon in 1862, but the family is documented in Pennsylvania through 1865. A marriage is recorded in a county that didn't exist yet at the time. These almost always indicate a transcription error or a date off by 10-20 years.

Why errors spread

Collaborative genealogy is built on copying. When you find someone who's already researched your great-grandmother, the easiest thing to do is import their work. Most people do. Most of the time, that's fine. But when a researcher in 1998 attached the wrong parents to a name, every tree that imported that record now carries the same mistake. The further down the family line you go, the more mistakes are baked in — and the less likely anyone's going to question them.

The errors aren't malicious. They're just old. Most of the wrong records in your tree were probably added by someone who's been dead for a decade, working from a source nobody else has looked at since.

How to audit your own tree (the manual approach)

If you want to do this by hand, here's the order of operations that catches the most errors fastest:

  1. Sort every person in your tree by year of birth. Scan the list for anyone born more than 45 years after their mother or 70 years after their father. These are your highest-confidence errors.
  2. List every couple where the mother's recorded birth year is later than the child's. Same logic for fathers, allowing for posthumous births within 9 months.
  3. Look for duplicate names within three generations. Especially common names — John Smith, Mary Johnson. If two records have the same name and overlapping life spans, one of them might be a duplicate.
  4. Check every marriage date against both spouses' birth years. Anyone married before age 14 or after age 80 is worth a second look.
  5. Verify your source documents. For every person, ask: what's the actual primary source for this birth, marriage, and death? If the answer is "another tree on Ancestry," that's a red flag.

This process takes hours per branch. Most people give up after one or two passes — which is exactly why the errors persist.

How AI auditing works

The audit logic above is mechanical. It's exactly the kind of task an AI can do thoroughly in a few minutes across thousands of records — flagging every impossible date, every duplicate, every conflict, every geographic inconsistency, with confidence scores and the specific reason for each flag.

That's the core of the Tree Auditor feature in FamilyTreesAI. You upload your GEDCOM from any service (Ancestry, MyHeritage, FamilySearch, RootsMagic), and the AI runs the same audit a professional genealogist would — but on every person, every record, every relationship, in one pass. You get a sorted list of errors, ranked by confidence. You decide what to fix.

The AI doesn't change anything on its own. It surfaces the problems. You're still the genealogist. We just give you a researcher who reads 100,000 records before lunch.

Audit your tree free.

Join the FamilyTreesAI interest list, and the first 5 ancestors on your tree get audited free when beta opens — no payment required. Drop your GEDCOM, see what's broken, decide if it's worth fixing the rest.

Get Notified When Beta Opens →

What to do right now

If you've never audited your tree, do the date-sort pass today. It takes 15 minutes and almost always surfaces at least one obvious mistake — usually a typo from years ago. Fix that one. Note where you stopped. Come back when you have time, or wait for the auditor.

The point isn't a perfect tree. The point is one that's verifiable — where every person, every date, every relationship traces back to a real source you've actually looked at. Most public trees fail this test. Yours doesn't have to.

This article is part of the FamilyTreesAI Family Legacy Library. We curate from vetted partners and write our own — no paywalls, no sponsored content. Read the rest at familytreesai.com/resources.