Best practice – How to use Devonthink

Hello everybody,

I have been trying to find a practical and clear solution for my document management for several years now. I also bought Devonthink (Pro) a few years ago, but I’ve never really warmed to it - probably because it’s very complex and I didn’t really understand the possibilities.

I’ve tested many different document management solutions, but none of them really appealed to me. So I’ve been looking at Devonthink more and more intensively lately and I like it more and more.

I would therefore like to know how others work with Devonthink and whether there are “best practices” for the different application possibilities - apart from the answer “That depends mainly on you”.

My goal: To archive private and business documents, almost exclusively PDFs and (very few) Office documents from Pages/Numbers and their counterparts from Microsoft, Open Office, Google Docs… Other files (project data etc.) I still manage in my folder structure and archive them there.

My first approach was to create several databases for “Business”, “Private”, “Accounting”, and “Rentals”, but I’m not sure if this makes sense or if I should prefer to create a “big” database where I create the above mentioned areas as subfolders.

I would like to collect all business documents (incoming invoices and outgoing invoices) and make them available to my tax consultant on a regular basis. These would therefore have to have a chronological order. The other documents should be sorted alphabetically “by sender/recipient”.

I have read on several pages that Devonthink can automate some things, but I have not yet found out on what basis this happens and what exactly can be achieved with it. In the documentation I did not find much about this. Can I use it to sort files into folders and assign tags or does it go further and convert information from text recognition into meta data?

Recognition / OCR / Meta-Information
Is it possible for Devonthink to automatically detect the date of a scanned file (not the file date but the date contained in the file’s content) and set it as the creation date? Can the sender of a document be recognised and set as author or creator or can a tag be applied?

Does it make sense to use tags to specify the sender or should it rather be displayed in the folder structure? Is it possible to arrange the tags in a meaningful hierarchical order, like this:

Client A
Client B
Service provider A
Service provider B

Sender/Recipient Category
Print shop
web host
Tax consultant

document type
Incoming invoice
Outgoing invoice

Many questions… I would be happy to hear how others handle this.

Many thanks and best regards

1 Like

Very extensive and organised spec. I take note it seems to be focused (fullyl?) on getting stuff into storage, but you mention nothing about how, when, why, etc. you use USE the information after storage.

yes, DEVONthink can be a resource for you. But so could simpler use of OSX finder and folders.

Random thoughts:

Automation. Read in manual about “See also and classify” and “Rules”.

Date Recognition. I never explored this with DEVONthink. But I use Hazel to look at incoming documents where I’ve instructed Hazel how to find relevant information from the document and then change the file name, e.g. with date of bank statement. Need to design the rules specific to each type of document. A good file name makes everything simpler, including DEVONthink.

Tags. I tag incoming (manually), but I find I never ever use the tags later. Muscle memory keeps me tagging, but i don’ know why I keep doing it. Sigh.

Categories. Yes, can put documents (and replicants) into the relevant groups in DEVONthink. I used to do that sort of thing, but now when I want to find an invoice (or invoices) sent to Client A (not a frequent thing, by the way) I just use search. I do “group” documents by client, but stuff just is thrown into there. I search (with tool or by inspection) when I want to find something.

Just feels to me what you are settting up is premature, e.g. before even trying to learn a bit about how you will put stuff into and get stuff out of DEVONthink. You can always drag and drop into a different structure with real documents once you learn more how to use make the effort valuable to you.

Remember though, I’m the sort of person who does not like to spend any time on creating and maintaining the “perfect” filing system. What I want is a system which allows me to productively/effectively FIND stuff, and a perfect filing system is only one way to scratch that itch. Reminds me too much of my first corporate job on a big project. The file room was huge with many staff managing it. Sometimes used by the project staff (or their delegate) to get information out, but frankly not as much as i expected. Was a good learning for me. Yes, we needed a filing system for such an endeavour, but scores of filing cabinets was over the top.

1 Like

In addition to the remarks of @rmschne:
I’m using DT also mostly to store personal and business finance documents. For me (!), it has proven useful to organize the documents in different databases according to tax requirements: business/rental/private. As @rmschne, I do use tags, but I’ve yet to find a use for most of them. They might come in handy if you have to group invoices like teleom, insurance … not sure about that, though.
Instead of using tags for incoming/outgoing, I use groups for expenses/income (where they’re appropriate, of course). Also, I have a goup “documents” for contracts and the like.

In German, I’d answer that with “im Prinzip ja”, which is a polite and diplomatic version of: you can try, but be prepared for it to not work in all cases. You might want to dig around in this forum and in the automation sub-forum – this question arises regularly and there does not seem to be a satisfying answer to it. Again, as @rmschnee, I use Hazel for some documents that arrive often and have a reasonable structure (notably, bank statements and telecom invoices). For everything else, I just add the date from the document to the file name itself and don’t bother with creation date. I then have smart groups to select all documents for one year. Given that date formats vary widely and that there’s no simple way to automatically select the correct date from a document containing several ones, this is causing less pain than trying to automate it.

1 Like

With regard to tagging, I use tagging extensively and use smart groups (based on tags) to organise my information. The ability to set a time limit in a smart group is great for ensuring you have the documents needed for a meeting (and only those documents) available.


Tags are just another type of group (folder) holding document replicants. It makes sense to me focus my organizational efforts on the database’s main folder hierarchy, which I think you’re saying you do also. The tags for senders, recipients, categories, etc., can definitely have their own hierarchy as groups in Tags. So, yes, you can have a top-level hierarchy such as:

… …Jones
… …Smith

and so on. But after a while, that might become very tedious if you have a lot of senders / receivers / categories. Perhaps rely more on compound searches when you need to find emails that Jones sent, rather than try to tag everything that comes in.

No hard and fast rules, but before committing yourself to a method think about the cost/benefit tradeoff of spending hours tagging in order to minimize a few minutes of search time. Tagging can quickly become a massive time sink.

1 Like