I reverse-engineered how 10 SaaS companies do content marketing

I spent the better part of three weeks reading every blog post on ten SaaS company websites. Not skimming. Reading. I made a spreadsheet with one row per post, columns for length, format, topic, “what is this trying to do,” and a score from 1–5 for “would I genuinely recommend this to someone.” I also tracked publish date (to spot cadence patterns) and whether each post linked out to other posts on the same blog (a quick way to see whether internal linking was deliberate or auto-generated). Two to three hours per blog, give or take.

The ten companies, in no particular order: Linear, Ghost, PostHog, Buffer, Intercom, Drift, Zapier, Stripe, Basecamp, and Fathom (the analytics one, not the AI notetaker).

I picked them because they’re all interesting in different ways and because I had opinions about all of them going in. By the end, half of my opinions had changed.

The four content strategies SaaS companies actually run

Every blog I read fell into one of four buckets. I started naming them halfway through and I’m going to stick with the names because they’re useful.

1. The Flagship strategy

Run by: Stripe, Linear, Basecamp, Fathom.

These are companies that publish less, but each post is a flagship — long, opinionated, well-edited, and built to be referenced for years. Stripe’s Atlas guides are the canonical example. Basecamp publishes a few times a quarter and every post has a point of view you could disagree with. Linear’s posts about how they build product are genuinely useful and have outlasted three SEO trend cycles.

The tell of a flagship strategy is that the latest post on the blog isn’t necessarily the most-trafficked post on the site. The compounding asset is the back catalogue. They write evergreen, and the catalogue earns links and rankings on its own time.

This strategy is hard. It requires good writers, good editing, and the patience to publish less. It is also the only SaaS content strategy that consistently produces work I want to read more than once.

2. The Help-Centre-as-Blog strategy

Run by: Zapier, Intercom (sort of).

These companies treat their blog as a discoverability layer for their help docs. Almost every post is structured like a tutorial — “How to do X with Y” — and the goal is to rank for the long tail of “how do I…” queries that the target user is searching.

Zapier is the best example of this in the category, by a huge margin. Their content is famously SEO-driven and famously enormous in volume. It works for them because their product is “the glue between other tools,” which means every other tool’s name is a relevant keyword for them. Almost no other SaaS has that breadth of natural keyword universe.

Most companies that try to copy Zapier fail. Their product doesn’t have the keyword universe to support it, and they end up writing 200 thin posts about topics that are tangentially related to what they sell. That’s the next category.

3. The Content-Mill strategy

Run by: Drift, Buffer (some of the time, not always).

These are blogs that publish a lot, on a lot of topics, with a lot of repetition. The posts are designed to capture top-of-funnel SEO traffic from broad terms in the category. Drift’s blog has dozens of posts about “conversational marketing,” because conversational marketing was their category, and they were going to own the keyword space if it killed them.

Some of these blogs work as SEO machines. Buffer’s blog ranks for an enormous number of social media-related queries and brings them meaningful traffic. Drift’s worked when conversational marketing was an emerging category and got progressively less interesting as the category matured.

The problem with the content-mill strategy is that it produces forgettable work. I read about 80 posts across the two of them in the spreadsheet phase. I could not, today, name a single one of them. I could rattle off five Stripe posts from memory. That gap is the point.

4. The Magazine strategy

Run by: PostHog, Ghost, Buffer (some of the time, in their better moments).

A small but interesting group: blogs that are not trying to do SEO and are not trying to do flagship-and-evergreen. They’re trying to be a magazine — a regularly-updated publication with a voice, multiple authors, opinions, a sense of humour, and a clear editorial sensibility.

PostHog’s blog is the strongest example I read. They write about engineering, about product analytics, about how their own team works, about their hiring philosophy. The posts are clearly written by humans who like writing. They link to each other freely. They sometimes disagree with each other across posts. The whole blog feels like a thing you’d subscribe to.

Ghost’s blog is similar — it’s mostly about publishing and writing and the indie web — and it works because Ghost itself is a publishing tool, so the audience self-selects. The posts read like a personal blog, not a corporate one.

The magazine strategy is much rarer than the other three, and when it works it produces the most loyal audience by a wide margin. The ratio of “people who would defend this brand on the internet” to “people who would defend Drift on the internet” is not even close.

What the good ones have in common

After reading roughly 600 posts across the ten companies, I started seeing patterns in the ones I scored 4 or 5. Five things showed up repeatedly:

A clear point of view

The good posts always took a position. Not “here’s what some people think about X.” Not “here are the pros and cons of Y.” A position. “We think this. Here’s why. Here’s what we’d do differently from the conventional wisdom.”

This is the single biggest difference between content I rated highly and content I didn’t. Most SaaS blog posts read like they were written by a committee that was afraid of being wrong. The good ones read like they were written by one person who didn’t mind being wrong if it meant being interesting.

Most SaaS blog posts read like they were written by a committee afraid of being wrong. The good ones read like they were written by one person who didn’t mind being wrong if it meant being interesting.

Specific numbers and specific examples

Stripe’s posts use real numbers from real Stripe data. Linear’s posts about how they build product reference specific decisions they made and their actual outcomes. PostHog publishes their actual hiring funnel statistics. Buffer used to publish their salaries, which is a level of specificity almost no other company has matched.

The opposite move — writing in vague generalities — produces the kind of post where every paragraph could appear in any article on any blog. That’s the killer. If a sentence is generic enough that another company could use it word-for-word, it isn’t doing any work.

Internal links that aren’t just SEO

The flagship and magazine blogs link to their own older posts a lot, but the links read like an editor said “this is genuinely related, you should read it.” On the content-mill blogs, the internal links read like an SEO tool generated them, because they often were generated by an SEO tool. The reader can feel the difference.

A good internal link is a recommendation, not a backlink.

Long is fine; bloated is not

The best-rated posts in my spreadsheet ranged from about 800 words to about 4,500. Length wasn’t the variable. Density was. A 4,500-word post that earned every paragraph beat a 1,500-word post that padded the introduction.

The worst posts I read all started with an inflated intro — a paragraph about “in today’s fast-moving digital landscape” or some version of that — before they got to anything specific. You could delete the first 200 words of every Drift post I read and the post would be objectively better. The same was true for about half the Intercom posts.

The flagship blogs almost never had inflated intros. They started with a claim or a question and earned their length by saying things.

A consistent editorial voice

The blogs I scored highest all had a recognisable voice. You could read three Linear posts back to back and notice they sounded like the same company. Same with Basecamp. Same with PostHog. It wasn’t because they were written by one person — it was because the writers had a clear sense of what the blog was and what it wasn’t.

The content-mill blogs were the opposite. Three Drift posts in a row felt like three different companies. Some were chatty, some were corporate, some were SEO-stuffed lists. There was no editorial centre.

A consistent voice is not something you achieve by writing a brand guidelines document. It’s something you achieve by editing every post against an idea of what the publication is. Most SaaS companies don’t have that idea, which is why their blogs feel rented.

What the bad ones have in common

The blogs I rated lowest all shared three things, in some combination:

• An obvious SEO motive that wasn’t hidden well. The post was clearly written to rank for a query, not to answer one. You could feel the keyword stuffing. The headings were keyword variants of each other. The internal links were optimised, not editorial.
• Stock-phrase openings. “In today’s fast-paced digital world.” “Whether you’re a beginner or an experienced marketer.” “Let’s dive in.” The instant tells.
• A vague middle and a CTA at the end. The post would set up a question, fail to answer it specifically, and then ask the reader to “book a demo to learn more.” This is the laziest move in content marketing and it shows up everywhere.

I’m not going to single out specific posts because the people who wrote them were almost certainly under pressure to publish more, faster, on smaller budgets. The bad posts are usually a process problem, not a person problem.

The Stripe outlier

A note about Stripe specifically. Their content output is unusual because they have a lot of money and they spend it on writers who can actually write. The Atlas guides, the Stripe Press book series, the engineering blog — these are produced at a level that almost no other SaaS company can match. It is unfair to compare a startup blog to Stripe’s. I included them in the spreadsheet anyway because it’s useful to see what the ceiling looks like.

The interesting thing is that Stripe’s content rarely sells Stripe. It sells the idea of being the kind of company Stripe wants to work with. They write for startup founders. The implicit message is “we get you, we’re for you.” The content is the whole pitch.

What to take from all this

If you’re running a SaaS blog, the practical things:

Pick one strategy. Do not try to be a flagship publisher and a content mill and a magazine. They require different writers, different editing, different cadences, different metrics. The blogs I rated lowest were the ones trying to do two strategies at once.

Publish less, edit more. Almost every SaaS blog I read would be better if it published half as often and edited twice as hard. The content-mill blogs especially. Cutting the bottom 30% of your output does not lose you traffic — it concentrates it on the posts that were doing the work anyway.

Hire a writer with a point of view. Not a content marketer who can write. A writer who cares about writing. There’s a difference, and it shows up on every line.

Don’t fake the voice. A consistent voice has to come from somewhere. If you’re trying to write a magazine blog, you need an editor who has actual taste and is allowed to enforce it. Without that, you’re just publishing things.

Write the post you’d want to read. This sounds like a cliché. The reason it sounds like a cliché is that almost nobody actually does it. Every time I asked myself “would I have read this to the end if I hadn’t been making a spreadsheet?” the answer told me whether it was a good post.

If you want a related piece in a different shape, how Notion’s landing page actually converts is the same exercise applied to a single page instead of an entire blog. Or, for the inverse — what an unsexy part of marketing that nobody writes about looks like — the email sequence nobody talks about: post-purchase is about the same kind of thing in a different channel.

The best SaaS blogs I read had one thing in common that nothing else did: they were obviously written by humans who liked their job. That sounds soft. It’s actually the only signal that matters.