If you have spent any real time on FamilySearch, you already know the frustration. You search a name, you get nothing, and you assume the record simply does not exist.
Often it does exist. It is just sitting inside a deed book, a probate file, or a courthouse ledger that nobody ever indexed. Your ancestor might be right there on the page, named as a witness to someone else's land sale or listed as a neighbor in a boundary description, and there was never a way to search for that until recently.
FamilySearch Full-Text Search changes that. It reads the actual text on the page using AI, rather than relying only on the small set of names and dates a volunteer typed into an index decades ago. This guide will walk you through how to use it, from your first search to the wildcard tricks that experienced researchers rely on.
What FamilySearch Full-Text Search Actually Does
FamilySearch has billions of digitized historical images on its site, but only a fraction of them have ever been indexed by name. Indexing takes an enormous amount of volunteer time, and plenty of record types, especially land records, probate files, and court records, were simply too dense and too varied to index the traditional way.
Full-Text Search gets around that problem using artificial intelligence and handwriting recognition technology. Instead of waiting for a human to type in names one at a time, the AI reads through the whole page and creates a transcript, which then becomes searchable.
That means you are no longer limited to searching the name of the person the document was officially about. You can search for any word that appears anywhere in the text, which opens up witnesses, neighbors, heirs, tenants, and anyone else who happened to get mentioned along the way.
FamilySearch currently has close to two billion records available through Full-Text Search, drawn heavily from United States land and probate collections along with a growing number of court and notary records from other countries. New collections get added on a regular basis, so coverage keeps expanding, but it is not complete everywhere yet. Some counties and time periods will have thin or missing coverage simply because those specific images have not been processed by the AI yet.
How to Find the Tool and Start a Search
Where It Lives on the Site
Full-Text Search is built into the regular FamilySearch search experience, though it is easy to miss if you do not know where to look.
Sign in with your free FamilySearch account, then run a normal search from the "Search for an Ancestor" box on your home page. Once your results load, look at the filter options in the left sidebar and click "Full Text Search." This opens a separate results view built specifically around AI transcribed text, and it will not show up automatically in your regular all collections search results.
You can also go directly to the tool through FamilySearch's search menu, or by searching for "Full Text Search" from the main site navigation. FamilySearch has been steadily moving this feature out of its original home in FamilySearch Labs and into the main search experience, so if a link you find online sends you to Labs instead, that is still a valid way to reach it.
Entering Your First Search
Once you are in the Full-Text Search tool, you will see separate fields for Keywords, Name, Place, Year Range, and Image Group Number.
For a first search, keep it simple. Enter a surname in the Keywords or Name field, then add a place and a rough year range if you know them. Click search, and you will land on a results page that you can then narrow down using the filters described below.
It helps to start broader than feels natural. A tight search with too many filters can accidentally exclude the exact record you are looking for, especially since AI transcription is not always perfect and place names or dates can occasionally be misread.
Filtering Your Results by Locality and Collection
A search for a common surname can return thousands of results, which is not useful on its own. This is where filtering becomes important.
On the results page, the left sidebar lets you filter by Collection, Year, Place, and Record Type. Clicking into any of these narrows your results immediately, and you can stack multiple filters at once.
Filtering by Place is usually the single most effective narrowing tool, especially once you get down to a specific county. If you know your ancestor lived in a particular county during a particular decade, filtering to that combination alone can take a search from thousands of results down to a manageable handful.
Record Type filtering is worth exploring too, since it lets you focus specifically on land records, probate records, or court records depending on what kind of document you are hoping to find. If you already suspect your ancestor bought or sold property, filtering to land records first saves you from digging through probate files that will not have what you need.
Using Wildcard Search Operators
Old records are full of spelling variations, and the person writing a deed or will in 1840 often spelled a surname differently than the way your family spells it today. Wildcards help you search all of those variations at once instead of guessing one at a time.
FamilySearch supports two wildcard characters. The asterisk stands in for multiple characters, and the question mark stands in for exactly one character. You need at least three actual letters in your search term for a wildcard to work, and you can use up to four asterisks in a single search.
If you are not sure how a name ends, put the asterisk at the end, so searching `Thib*` will catch Thibault, Thibodeau, and other variations that share the same start. If you are not sure how a name begins, put the asterisk at the start instead, so `*bou` would catch names ending that way.
The question mark works well for single letter substitutions. Searching `Sm?th` will catch both Smith and Smyth in one search, which is exactly the kind of small spelling difference that trips up a standard name search.
Wildcards are especially useful for surnames with European, Slavic, or Southern United States origins, where spelling was rarely standardized and a single family might appear under three or four different spellings across their own lifetime.
Search Strategies for Finding People Missing From Standard Indexes
Once you understand the mechanics, the real skill in Full-Text Search is learning what to search for beyond just a name.
Pairing a surname with a specific year often works better than a name search alone, since AI transcribed text can be inconsistent about which date on a page it associates with a person. Searching a surname alongside a year like 1834 can surface a deed, a will, or a court entry tied to that exact time period even when a plain name search misses it.
Occupations are another strong angle. If you know your ancestor worked as a blacksmith, a miller, or a tanner, searching that occupation together with the surname can narrow a broad search down to records that specifically describe their trade.
Wills and estate inventories tend to list specific personal items by name, so searching an object like a pocket watch or a particular type of livestock alongside a surname can turn up records that a name search alone would never catch.
Boundary descriptions offer a similar path into land records that never mention a name search would catch on its own. Old deeds frequently describe property using landmarks and directions, so searching a phrase like a specific creek name or a distinctive boundary marker can lead you to a deed for property your ancestor owned, even if the transcription of their name itself came out slightly wrong.
Quotation marks are worth using whenever you want an exact phrase rather than scattered individual words. Searching "last will and testament" in quotes, for example, keeps those words together instead of returning every record that happens to contain any of them separately.
It also helps to try your ancestor's name in more than one order. Some historical indexes and record formats list the surname first, so searching both "William Crisp" and "Crisp William" as separate phrase searches can catch entries you would otherwise miss.
Reading the AI Transcripts With a Careful Eye
The transcripts you see next to each Full-Text Search result are genuinely useful, but they are not perfect, and treating them as a finished, verified source is a mistake worth avoiding.
Handwriting recognition technology still struggles with certain things. Faded ink, old cursive styles, and Latin text in particular tend to produce transcription errors, so a name or date that looks slightly off in the transcript is worth double checking against the original image before you trust it.
This tool has also become a genuinely important way to locate enslaved ancestors and their descendants, since deeds, probate files, and plantation records frequently mentioned enslaved individuals by first name only, without ever including them in a standard index. Searching a first name alongside terms that would have appeared in that historical context can surface records that were effectively invisible before Full-Text Search existed. It is difficult and sometimes painful material to work through, but for many families it is the only path back to information that would otherwise be lost entirely.
Whenever a result looks promising, always click through to view the actual scanned image rather than relying on the transcript text alone. The transcript is a doorway into the record, not a replacement for reading it yourself.
Citing What You Find
When you find a record through Full-Text Search, cite the original document image the same way you would cite any other FamilySearch record, rather than citing the AI generated transcript itself.
If you want to reference the transcription specifically, because you are quoting language from it directly, make a note that the text came from an AI generated transcript and that you verified it against the original image. Future you, or a family member using your research later, will appreciate knowing exactly how confident to be in what you found.
Why This Kind of Discovery Matters
There is something different about finding your ancestor's name in the actual body of a record instead of as an entry on an index card.
An index tells you a name existed. A full text result might tell you your ancestor witnessed a neighbor's land sale, or that they were listed among the personal property in someone's estate, or that they owned fifty acres along a creek that still has the same name today. Those details do not just confirm a person existed. They start to sketch out an actual life.
That is really the whole point of tools like this one. The record was always there. Now it is finally possible to find it.
If a search like this leads you to a story worth remembering, whether it is a discovery about where your family came from or a detail that finally answers a question you have had for years, it is worth writing down somewhere it will not get lost again. That is exactly what Memoracy was built for, giving you a simple place to record the stories behind the names you find, so the next generation does not have to go searching for them the way you just did.
Sign up and start your first story on Memoracy today.
