Forget Everything That You Read About Ideal Content Length

Data doesn’t lie but the interpretation of data does

Data doesn’t lie but the interpretation of data does.

There a lot of information about ideal content length. The most common content-length questions revolve around customarily considered long-form content pieces: blog posts and articles. It’s important not only from the point of view of content marketing but for bloggers and media too. Pursuing the own goals, everyone wants to stand out and do it in the right way.

Interest in the correct answer has led me to a set of researches. All of them are so different. Curious, I have attentively studied each of them. Let’s begin with me.

1. The seven-minute post collects the biggest
amount of reader’s time

One of the most known and quoted researches is “The Optimal Post is 7 Minutes”. It was published by Medium team 3 years ago but data have been obtained on the basis of the analysis of one million posts.

The total time spent reading a post equals the total number of visitors to that post multiplied by the average time spent reading per visitor.

How is it 7 minutes length?

7 minutes post length is a variable. It’s strange but many people understand that result as constant. For example Buffer or (TheNextWeb), MarketingLand, Snap, and many others mistakenly interpret these 7 minutes as 1600–1700 words. Probably, it is so if the article belongs to a genre prose or the non-fiction. Even though masters of speed reading techniques can do it faster. But if I read about JavaScript library for building neural networks I need a half hour and maybe more. 1600–1700 words in scientific or in official document stylistic will be able to be understood in 7 minutes only occasionally.

We can only to guess about numbers of genres and media file types. At that time in Medium, there were 5 million posts, behind an exception of articles with more than 1 million views and articles with more, than 20 minutes of the estimated reading time. Also, we must don’t forget about existence and numbers of types of images, video and other media files in posts. Chart and meme require a different attention. How long does it take?

7 minutes — is an indicator of the preferred time for engaging with the publication, not of content length. It’s important insight and we will return to it later.

What metrics matter?

At the beginning Medium analysts tried to determine optimal post length on the basis of the number of views. Then it was told:

We care less about clicks and more about actual reading. Time spent is a better reflection of this.

Total reading time is one of the most valuable metrics only for press. It displays opportunities for advertising campaigns. But no less important for press and content marketing is content performance. How deeply users engage with content and do they read up to the end? That’s not all. I asked experts about most important metrics for media.

Martijn Scheijbeler — Director of Marketing at The Next Web

Reach: most important for advertisers + create a sustainable business model in media. Then the secondary metrics are: Engagement / Retention + Loyalty. Making sure that a user is engaging with your media properties as much as possible so you can really stick the brand of the company in their mind.

Travis Bernard — Director of Audience Development at TechCrunch.

The most important metrics are your the number of users that regularly consume your content. Not the ones that come once a month, but the ones that come every day. How many people like this are there and how can you grow that number?

To read to the end a short article or not to read to the end long article?

Look at the chart below. It’s worth noting that at y-axis is written “AvgSesonds Spent per Visitor”, but in the context of whole research, it is exactly a time of reading. It determines by Medium technologies (recording scroll positions, pauses, an activity of tabs, etc.). Also, the x-axis is the estimated time of reading, not a real time of reading as on final chart with conclusion of about 7 minutes. Therefore Medium wrote about 7 minutes without mention the number of words.

Each dot represents an average of all posts with the same length, rounded to one second.

I have marked due places of the crossing of the estimated and actual reading time. As you can see two-minute articles (~485 words) have the highest coefficient of reading involvement. Whether the reading involvement correlates with a finished reading indicator is only possible to assume. The probability of finished reading much more if actual time of reading is more or equally of expected. But indeed it is possible to read attentively and long 30% of the article, then to doubt its usefulness and fluently to read only headlines up to the end.

Why are 7 minutes defined not absolutely correctly?

In a Medium research, there was no emphasis on the previous chat. After that was calculated the median of reading time per user for the y-axis, not average time. This median became the first component of final plotting with a conclusion of about 7 minutes. Pay your attention how strongly time for long reads has changed.

Each dot represents an average of all posts with the same length, rounded to one second.

Why the decision to be guided on a median has been made?

We often noticed outlier visitors who spent exceptionally long amounts of time reading a post. We already remove the time from idle browser tabs, but there are still outliers totaling several hours (sometimes spanning multiple visits). They raise the average enough to cause overestimates.

I consider that it’s not true. As said before, there are a lot of publication types. I was reading Medium research more than 1 hour, analyzing together with them and checking all facts many times (because it’s Medium). As result in Medium’s opinion, I became an outlier visitor who spent exceptionally long amounts of time reading a post. I haven’t got to the math range. Although there are already excluded and forgotten a huge number of people, who attentively read articles in Pocket, Instapaper, Evernote.

What about content marketing?

The main goals of content marketing are registrations and sales. From the point of view of these aims, even single-use readers matter too. It’s noteworthy, I didn’t say that recognition of the reader isn’t important. But I can read Buffer’s post about ideal content length (I hope you understand what I meant) and will start to use their product (Buffer I really love you).

Double–edged sword. I can read digital agency’s blog to improve my skills and will never order their services. Double–edged sword again. Maybe their strategy of content marketing isn’t correctly built or, for example, their content marketing is directed to recruiting.

Let’s return to the stats. The average number of visitors per post was the penultimate chart in the Medium research. I already said that median component for y-axis doesn’t consider correct, but it’s data too.

From the chart, it is visible that four-minute articles on average gather ~600 visits, and seven-minute ~750 visits (sorry, I plotted it as I could in the Photoshop). In the aspect of marketing and all business, it is important not to forget about ROI (return of investments). It is obvious that 2 four-minute articles potentially have the bigger value and moreover less cost.

That’s all about Medium research. It would be interesting to look at the same chart, but with displaying the number of words at y-axis.


2. Long-form content is better ranked in search engines

Second of the most popular research was published in the 2012 year by Serp IQ. The chart reflects content length multiplied by the search engine top results. How does it made — nothing is told. I had looked for the answer and found two articles dated in 2009 about correlation in-links and SERP.

Guys, we live in 2016. Search engine results depend on hundreds of factors. The length of article has humiliating low value on this list. Three years ago my collection of best Sublime plugins was on the 3rd position of Google after good sharing on Reddit. It was giving me 200–300 clicks per day for several months. Length is 1000 words. But I am not SEO expert, please correct me if I am wrong.


3. Finally the viral effect directly depends on content length.

The most objective insights aren’t most popular over the internet. In 2015 BuzzSumo teamed up with Moz to analyzed the shares and links of over 1 million articles. They removed videos and quizzes from their initial sample to analyze the impact of content length. This gave a sample of 489,128 text based articles.

Seems, I have nothing to say against it. But my research has arisen not only because of interest in math sequences. I was pursuing “golden ratio” for content marketing. Based on the data from the above table it is possible to tell that three articles with 1000 words will be more effective and easy to produce than one article with 5000 words.

And now compare a difference of average and median values. Moreover, in a research, it is told that about 50% of all analyzed articles had only a few shares. Thus final huge numbers, for the most part, are got by the largest media companies. Therefore insights from Moz and BuzzSumo so differ from Medium.

I recommend to read the full 30-page research report. You will find there a lot of useful insights. Also, there is a very interesting work by data champion HubSpot, but they analyzed only 6,192 blog posts.


It turns out that better to write more?

Let’s remember the 7 minutes of Medium. The most expensive currency in the world is your time. Financial Times, TheGuardian, HufingtonPost, TechCruch, TheNextWeb and others are the famous brands. You trust them. You “pay more” for their information, but you aren’t ready to give over 7 minutes to the unknown author. That’s how I realized Medium research.

So what is the best content length?

There is not. Forgot everything that you read above. I only wanted to expose incorrect information for you. Today’s technologies couldn’t answer this question. It is necessary to determine text stylistic, a reading rhythm, reading involvement, finished reading, kind of content and many other things. You may do a great research with huge datasets, but as a result, two words — name and surname of the new president will break your math range.

Mike Sall — Head of Product Science at Medium:

What it does mean is that it’s worth writing however much you really need. Don’t feel constrained by presumed short attention spans. If you put in the effort, so will your audience. It’s just math.

Write so much as you want and don’t put yourself in borders of the math
probabilities. Your feelings, your thoughts, your worldview couldn’t be calculated. These still matter in this technical world.

This post is part of our contributor series. It is written and published independently of TNW.

Read next: Learn to Detect What WordPress Theme a Website is Using