Jakob's Law of Internet User Experience states that since users spend most of their time on sites other than yours, straying too far from established standards creates an unnecessary cognitive load. It’s a simple idea that can be immensely helpful for creative work.

I remember when browsers and bandwidth had improved enough that website owners rushed to add HD videos behind H1 text. Or when design agencies coded cursor overlays to make their experiences “stand out from the rest.” What they were were distractions. A reason to hesitate before clicking “Sign Up.” 

The writing world has its own version of Jakob’s law. Readers expect certain conventions, formats, and tones from SaaS websites. And the person who can master those, “hacking away at the unessential” as Bruce Lee put it, has an unbeatable edge.

My job is to create helpful, interesting explanations, walkthroughs, and guides for tech companies. If I wrote them in a vacuum, without doing my homework and surveying industry norms, my writing would flop. So I need to periodically venture out and see what others are doing in the space. 

This time, I kept it small. Manually cataloging attributes of the 10 most recent posts from 12 somewhat arbitrary blogs (I was going to stick with Mike Sonders’ The 50 Biggest Public SaaS Companies but ended up adding those that I interact with most):

• Adobe

• Salesforce

• Zapier

• Shopify

• Datadog

• Slack

• MongoDB

• HubSpot

• Stripe

• Jotform

• MailChimp

• Sentry

The sample size would’ve been larger if all of the variables I wanted were unambiguous (e.g. word count, publishing frequency, and Flesch Kincaid reading ease). But categorizing articles by format, calls to action by location, and whether or not an article counts as a guest post were all subjective measurements that I had to do by hand. Even so, a sample size of 120 recent articles led to a few clear trends that writers can learn from.

Longform is less popular than I would’ve thought

Half of the articles I surveyed were under 1,250 words. That runs counter to most of the SEO advice I see nowadays. HubSpot, for example, says “For SEO, the ideal blog post length should be 2,100-2,400 words, according to our data.” And to their credit, the average word count of the HubSpot articles I cataloged was 2,273.

Then again, anyone providing SEO advice on article length will inevitably caveat it with “though you should always write for humans above all else.” Personally, I’d like to see more posts like Sentry’s < 1,000-word We Just Gave $500,000 to Open Source Maintainers, which racked up 140+ points on Hacker News, earned ~100 backlinks in a matter of weeks, and included more than simple self-promotion.

For broad overviews and technical deep dives, 2,000 - 2,500 words is still a safe target (app roundups might go longer). But for opinions, observations, hot takes, and predictions, I think we’ll start to see the advice swing toward brevity as the age of SEO is replaced by new forms of discovery and distribution. 

Less than 20% of articles were written in the first person

I recently saw some SEO talking heads claim that writing in the first person boosts expertise and experience signals to Google’s ranking algorithm. Others have made similar comments about adding author bylines and making detailed profiles more visible. There doesn’t seem to be any way to verify or debunk these claims, but to me, they’re worth implementing nonetheless, if only because it incentivizes careful writing.

When someone can publish without worrying about public backlash, writing quality suffers. If, however, you take the “make engineers sleep under their bridges” approach, authors have skin in the game. They have to be careful before putting things out into the world. And readers know this. 

Every blog I checked provides author bylines, except MailChimp. Only six out of 12 SaaS blogs had detailed profiles for authors though, providing information on their backgrounds and portfolios. 

Even more rare were articles written in first person, of which I counted only 21. That feels like a half measure and a wasted opportunity to make the writing feel more direct and personal. As AI writing ramps up and imitates what came before it, now seems like a perfect time to ditch the corporate “We.” 

The average SaaS blog is considered college-level reading

This Flesch Kincaid Calculator gives a reading-ease score primarily based on two numbers: [# of words / # of sentences] and [total # of syllables / # of words]. An article with shorter sentences and shorter words receives a higher score than one with run-on sentences and big words. 

Obviously, certain topics have an upper limit on reading-ease. Find and remediate identity risks with Datadog CIEM, for example, would be hard to score above its graduate-level score of 21.75. But should Jotform’s Salesforce Surveys Pricing article really be at 26.82, or a post-undergrad level? No. It’s a broadly useful and straightforward breakdown. That, and I got its score over 50 after about 15 minutes of editing. 

If your goal is to encourage conversation and collaboration, your writing should be as easy to read as the topic allows. You’d reach the point of diminishing returns very quickly when trying to raise the Flesch reading score of an article about configuring Stripe’s API. And your target audience would barely notice. 

Henrik Karlsson recently wrote, “Writing for a general public, you need to be broad and a bit bland. I didn’t want a general public. I wanted a specific set of people, the people who could help me along as a human being obsessed with certain intellectual problems. I didn’t know who these people were. I only knew that they existed. Hence my writing was a search query.” When complex, challenging topics attract the people you want to connect with, a higher reading-ease score may actually be detrimental.

Almost half of the articles didn’t have a direct call to action

When I first started writing for the web, I was told to always end with a direct call to action. Every single time. An imperative verb followed by “now,” “today,” or “before it’s too late.” Veteran writers described doing otherwise as leaving money on the table. I hated that take when I was coming up, and I still hate it today.

Only 25 out of 120 articles I cataloged this go-around had no obvious CTA. Another 29 had a generic, “separate” call to action in the footer, often one that applied sitewide (not just to blogs). Of the remaining 66, 34 were invitations to take a “transitional” or in-between action, like downloading an eBook or signing up for a newsletter. That left 32 that leaned on a traditional “in-line” CTA—one that was placed within the article, aligned with the topic, and invited readers to sign up for the app itself. 

Almost all of the No-CTA articles were from three sites: Adobe, Slack, and MailChimp. Apps that are arguably so dominant in their respective spaces that they don’t need to shove people further down the funnel. But this is still a hill that I’ll die on for apps of all sizes. Because when you write about cool use cases, inserting interesting screenshots and detailing relatable case studies, those are the highest-performing CTAs. When they aren’t working for you, that’s not a sign you should be more heavy-handed. It’s a sign that you need better writing.

Feature updates and listicles are the most frequent formats

While there are no clear definitions of article types, this was one of the graphs I was most excited about. The number of product announcements surprised me but there might be a bias toward larger teams having more frequent feature rollouts. And although I don’t love listicles, most of the ones I saw (Zapier and Shopify accounted for most of them) were useful. 

It’s great to see such a good showing for human interest (interviews, case studies, etc.) and thought leadership articles like Sentry’s A Story About HTTP Status Codes. Especially considering how time intensive and SEO-unfriendly they can be. 

Comparison articles showing up last on the list is the one that surprised me the most. They’re relatively easy to produce, aligned with readers who are primed to buy, and relatively easy to rank for if one of the two subjects isn’t well known. MailChimp’s RFQ vs RFP ranks in the top five results for 88 keywords that, in total, have well over 10k searches per month.

Where to go from here

After all of the copying, pasting, counting, and averaging, I’ll be pushing clients to get on board with more more CTA-free comparison articles (seeing as there’s less coverage than I I thought) and trying to get more comfortable with listicles (since they’ve endured, and succeeded, in spite of their reputation) written in the first person. Unless a client wants to show up on Hacker News. Then it’s back to contrarian anti-capitalist takes with Flesch scores under 10.