This can vary from project to project, historian to historian, subject to subject. But I can tell you about my research. Below I am talking mostly about papers; a book can be 5-20 times the work of a research paper, and there are differences of scale in that (like how long it takes, and how one manages a research project that large), but the process is pretty similar.

Historical research is not usually about "new information," though it can sometimes be. More often it is about putting together information in new ways, making new interpretations. So my most significant research papers have generally had no "new information" in them, in the sense that nothing in them was in any real sense "undiscoverable" by other historians. But they are based on sources that other historians hadn't looked at towards answering the same questions that I was interested in asking, and many of the sources were, shall we say, "neglected" — they weren't part of the stories that other historians were telling, because they were asking very different questions than me. The actual sources were in archives (like the National Archives), though sometimes these are available in easier-to-access sources (a lot is online these days, in large databases, some of which are free to the public, some of which require an institutional subscription).

When I do get "new information" it is usually supplemental to these other sources. In my own case, "new information" can come from the Freedom of Information Act, which compels the US government to check if its once-secret documents are still secret. This can result in documents been released that had never been released to the broader public. This can sometimes be quite useful though I think people overestimate how important it can be — very rarely does it do more than supplement what I already know.

I also can get "new information" from interacting with "historical actors" (people who were involved in the things I am studying) because my work is largely focused on the late 20th century. Sometimes that can mean documents, sometimes that can mean interviews. Again, I think it can be easy to over-estimate how useful an interview or a document can be, but sometimes it can really help get at aspects that are not preserved otherwise in the historical record, or (in the case of my topics) things that are still too classified to find much documentation on (but a person might be willing to speak vaguely about).

Historians have different schedules of work. I think I'm on the slow side of things; it takes me about 4-5 years per major research project I work on, though I usually am working on several at once, so my rate of "production" is about 1 per year or two. Some historians can produce much more than that (e.g., a book per year) and still do good work. We each have our own modes of working on things. I like taking my time and really learning the ins-and-outs of the subject, and making each project a sort of self-contained, all-encompassing entity (if I want to write smaller things, I do that in non-academic settings).

For me, the process of using the documents, interviews, secondary sources, and other evidence largely comes down to taking lots of notes on them. I sometimes use database software for this, I sometimes just use a large Word file or two. It depends on my own inclination and the complexity of the topic. My goal is to piece together a narrative that makes an interpretive argument. Which is to say, I'm trying to tell a historical story that answers a historical question that other historians might find worth answering. Sometimes these questions are ones that non-historians would find interesting without too much explanation ("what did Harry Truman know about Hiroshima prior to the atomic bombing?") and sometimes not ("what role did patenting play in the Manhattan Project?"). All of them are more than just the questions themselves — a "good question" is one that leads you to all sorts of other interesting questions, and whose answer has larger implications than a simple answer. (So for Truman, the answer I came up with was, "he wasn't aware it was a city," which raises all sorts of other issues about what he thought it was, why he would have been confused on this point, and how this changes our understanding of the atomic bombings. For the patenting one, the answer gets into how secrecy worked and did not work during World War II, and what they were imagining the post-WWII approaches of controlling nuclear weapons would look like.)

Coming up with the "question" is the hardest part. It requires quite a lot of mastery of the subject matter you're studying, to know if you're asking a new question, much less an interesting one. Often one gets a whiff of an interesting question while working on other research, and only later does one start to flesh it out. Often the fully-formed version of "the question" isn't clear until you're near the end of the research process (sometimes trying to answer a different question initially). This is the hardest part of the process to write out how it works — it's not an orderly process, it's associative and intuitive. It's the sort of thing you learn from doing over and over again, over a long period of time. It's what separate the historians from the history buffs, the ability to do this well.

The actual writing process for me is pretty quick compared to the researching process; by the time I get to the place where I'm ready to write things down, I've been thinking about the topic and the documents and everything else for a long time, so there is by that point a natural "structure" to how to write the paper. (It helps that I've written many papers at this point — knowing how to write a history paper is something you get better at over time.)

Prior to actually writing it, though, I give the paper as a talk (sometimes several times) at conferences or workshops. This gives me an opportunity to organize my thoughts about the work and the argument concretely, to share the work with other scholars in a preliminary way, and to signal to people that this is what I am working on (which sometimes results in other scholars sharing documents or information with me that I might not have come across). It also lets me see what kinds of questions other scholars ask about the work, which is useful for the writing phase, since I want my ideal scholarly reader to be satisfied (or, at least, I want to know what objections they might have ahead of time).

Eventually I do write it up — this takes maybe a week or two at most for a paper (it would take less if writing papers was all I had to do as a historian, but it is not!) — and then one submits it to a venue for publication, like an academic journal. Then one waits a long time (maybe 4-6 months if you are lucky) to get peer reviews back. Then one makes revisions along the lines of the reviews and the editorial comments, then sends it back in. Maybe they publish it then, maybe they show it to more reviewers. If they are happy, then it gets copyedited and typeset, and then you make one last attempt to find any errors that have crept into the process (on my Truman paper I found one huge error — the word "not" had been somehow removed, which changed a key sentence), and then, finally, maybe six to twelve months after that, the thing emerges in print.

At that point, you want people to know about it, not only because it's nice to have someone actually read and engage with the work at that point (it's been so long!), but also because the "currency" of internal academic promotion is being cited, and people won't cite you if they don't know about your paper. So you then e-mail your paper to colleagues who you think might be interested in it, and mention it if you give talks, and maybe upload it to your website (if you can get away with that), and generally try to make it easier to access. And you might engage with critics and so on. This is the "getting accepted" phase of things. Especially of use is sending it to colleagues who teach courses on the subject you're writing about, with the hope that they'll assign it to their grad students — it's a slow payoff, but this is ultimately what establishes which articles "matter" in a field, if they are commonly read by students early in their careers.

As you can see the whole process is very long and very slow for any individual article. A book can be even slower (add a decade onto all of that, perhaps). But again we are not usually doing things serially; we are doing several things at once, so there is a regular rate of "ripening" of topics and papers (a "pipeline" if you will). Altogether it is not a fast discipline, and very unlike the sciences in that respect (in which a paper's contents can be out of date in a month or two).