If the previous article is the most important research work I do in the short term, then building a research database is the most important research work I do in the long term.
When I say “database,” I’m using a deliberately vague version of that term. What I mean is that you want to have an organized place in which you can store research findings over time. You want your research work, most of which will be on a per-article and per-project basis, to accumulate toward something larger and greater.
The end goal is to slowly free yourself from having to do all of your research from scratch. Eventually, you’ll almost always have at least a starting point, and in some cases, you’ll either have much of the research done or have information and insights you couldn’t have realistically found otherwise before you even open Google.
Your database can take many different forms and I recommend experimenting to see what works for you. To guide you, though, I’ve identified three stages. And I want to note up front that this work has diminishing returns: You’ll achieve a majority of the benefits from the first stage, another good but smaller chunk from the second stage, and depending on your niche, pretty narrow benefits from the third.
Highlights will form the foundation of your research database but even on their own, they provide significant benefits.
Instead of reading books in paperback or articles on the web, read eBooks via Kindle (or another eReader) and read articles via Pocket (or another “save it for later” app, such as Readwise Reader, Instapaper, or Matter).
As you read, highlight sections that seem especially interesting or relevant—not only to the article you’re working on currently but to work you might do in the future. You don’t have to save everything nor read everything extensively, but highlights give you anchor points so that in the future, you can remember “Hey, didn’t I read something about [X] once?” and you’ll actually be able to find it.
Here’s an example of how this could work: A year and a half ago, I worked full-time at a company that built developer tools and as part of a particular content pillar, we were going to write articles about technical debt. I purchased and downloaded the eBook Kill It with Fire: Manage Aging Computer Systems (and Future Proof Modern Ones) by Marianne Bellotti because it sounded like it’d have a deep as well as wide-ranging perspective on technical debt.
And it did! I enjoyed the book more than I thought I would and made many highlights.
Cut to a few weeks ago, more than a year after I finished the book, a startup called Authentik contracted me to write an article about a shifting technology trend—a movement from SaaS delivery models back to self-hosted delivery models. This topic had nothing to do with technical debt but I recalled an interesting argument in Kill It with Fire—something about technology cycles.
Within a few minutes, I was able to find the passages I was remembering by going to my Kindle highlights and skimming.
I threaded the quotes into the argument I was making and in the final article, I think it all comes together in a compelling argument.
Note: the payoff not only came more than a year after reading the book but the cause for the citation wasn’t even the primary topic of the book in question. That’s why I recommend having a trigger finger for highlighting: It’s always easier to skim and CMD+F than to reread.
The primary weakness of the highlights system is that it’s decentralized and unwieldy. It can work well, but it gets cumbersome if you don’t remember where an idea comes from. Worse, it’s almost impossible to research in the other direction—if you want to, say, look up everything you’ve ever read about technical debt.
That’s where the database comes in. A database can take many forms but for these purposes, it has to, at a minimum, centralize your highlights from across all sources and give you an effective search function to search through them.
I’ve iterated on this a lot over the years but my current database consists of Notion, as my primary database; Readwise, as my connective tissue; Readwise Reader, as my article reader (only a week ago, this was Pocket); and Snipd, as my podcast player (previously Airr).
Automation is an essential component because manually copying and migrating text is so laborious that you will inevitably stop doing it. I pay for Readwise so that I can highlight in any of my reader apps and automatically port highlights to Notion.
Note that I’m neither particularly technical nor a pro Notion user, so this setup is pretty out of the box.
Now, if I’m writing an article about technical debt, I can put in a keyword and see what other research I’ve done and highlights I’ve made that reference the concept.
The benefits of centralization are surprisingly exponential. The more narrow your research field and the more research you do, the more results this work can yield.
In the grand scheme of things, I haven’t spent a lot of time working in my current niche nor contributing research toward that niche—and yet, doing this search reminded me of a few sources that mention technical debt that I would never have remembered doing so otherwise.
Full disclosure: I’m not at this stage and I’m not convinced it necessarily benefits a content marketer to reach this stage.
Learning systems, part of a broader field called personal knowledge management, are when you take the tools and processes above and add procedures for reading, distilling, and summarizing the research you’ve done so that you can really learn everything you read.
(Source—and yes, though this looks complicated already, there are four more infographics in this series).
There’s lively debate among personal knowledge management fans about the benefits of these models with the primary question being: Are you spending more time making notes than reading or writing?
But I believe these systems are especially likely to waste a content marketer’s time because most content marketers aren’t learning to become an expert, so we don’t need to memorize or internalize much of our research in the way a founder might want to internalize a new mental model or a scientist might want to memorize emerging data.
That being said, all these systems are worth looking into because they all, at a minimum, offer techniques we can learn and steal from. My introduction to many of these ideas was via Forte, even though I put a few of his practices into use these days.