Documents and AI
Uploading many documents at once
What to decide before dragging in two hundred files, because undoing it afterwards costs more.
Sooner or later a big block has to go in: a folder from a finished site, the archive of an incoming client, a history being migrated. It can be done, and the result depends almost entirely on three decisions taken before you start, not during.
The three decisions
- 1
Which file each item goes to
Dropping two hundred files into a generic place is fast and creates a problem you already know: a drawer.
- 2
What gets read and what does not
Not everything deserves extraction. What will be consulted by content, yes; the rest can stay as archive.
- 3
And what to do with duplicates
In a large block there always are: the same document under two names, or two versions of one.
Important
The second changes the cost most and is the one nobody decides: **every document that is read is an action**. Uploading a two-thousand-file archive and asking for all of it to be read consumes two thousand times, and of those two thousand perhaps fifty will ever be consulted. Reading what will be used and leaving the rest as archive is the difference between bringing your history over and paying to bring it twice.
What to check before dragging
| Check | Why beforehand |
|---|---|
| That names say something | Afterwards they are already in with IMG_4821 as the title |
| That there are no protected files | They upload, but cannot be read or searched |
| That the block belongs to one matter | Mixing two sites in one batch is what creates the drawer |
| And that someone will review what comes in | A bulk upload with no review is archiving, not management |
Watch out
The practical nuance that saves an afternoon: **upload a small sample first**. Twenty files of the same type tell you whether the scan quality works, whether the names make sense and whether extraction picks up what you care about. If something is off, fixing it with twenty takes minutes; discovering it with two thousand already in does not.
After uploading
The last looks obvious and is the most repeated error in long migrations: the same folder gets brought over twice because nobody noted that it already was.
Worth knowing
If the block is very large, split it into batches by matter rather than by size. Ten batches of twenty documents from one site each are manageable; one batch of two hundred from ten different sites is a problem to unpick by hand.
›Is there a per-file limit?
Yes, and it is usually generous: if a file will not fit, it is almost always scanned at far higher resolution than needed.
›Can I upload whole folders?
It depends how you do it; what matters is which file they land in, not how many there are.
›What if I put the whole block in the wrong file?
Move it, do not delete: moving keeps the trail and does not consume again.
A real case
The situation
A company migrates a finished site's archive and asks for everything to be read.
What you do
- Uploads a sample of twenty before the two thousand
- Decides which types get read and which stay as archive
What you get
The history comes in tidy and consumption stays on what will actually be consulted.
The situation
Three hundred documents are uploaded and nobody knows which failed.
What you do
- Checks the load's result
What you get
Only what failed gets retried.
The situation
The load cuts out halfway.
What you do
- Checks what went in before repeating
What you get
What is already uploaded is not duplicated.
The situation
They all land in the same file undifferentiated.
What you do
- Prepares the mapping before loading
What you get
Each document reaches its place.
The situation
They are loaded with names that say nothing.
What you do
- Names or classifies before uploading
What you get
They are locatable afterwards.
The situation
A big load runs without a prior test.
What you do
- Tries a small batch
What you get
The problem is found with ten rather than three hundred.
This article answers
- upload many files at once
- import a folder of documents
- bulk document upload
- how many documents can i upload at once