Word Count Data-Mining Study

I went data mining (means to drill figuratively into a database to analyze data or data about data (meta-data)) to try and prove my gut feeling that articles less than 500 words outperform articles with more than 500 words in terms of total page views of traffic earned per article.

Over time, I found my hypothesis to be true, but it was only true over time and almost never true in the short-term. Check out the 6 charts below that cover my 27 day analysis period of 2005 vs. 2006 for various word count ranges:

Here is a chart for the first 27 days of September 2005:

ime Period 2005-9-1 to 2005-9-27
Word Amount Articles >= 500 Articles < 500
Total Articles 4,522 4,574
Total Views 2,481,420 2,743,964
Average Views 548 599

Compared to the first 27 days of 2006:

Time Period 2006-9-1 to 2006-9-27
Word Amount Articles >= 500 Articles < 500
Total Articles 8,439 10,857
Total Views 616,624 730,602
Average Views 73 67

Mini-Conclusion based on 500 word analysis: In the short-term, shorter articles receive less traffic than larger articles, but over the long-term, shorter articles outperform longer articles. However, the difference is negligible.

750 Word, 27 days of Sept 2005 vs. 2006 analysis:

Time Period 2005-9-1 to 2005-9-27
Word Amount Articles >= 750 Articles < 750
Total Articles 1,981 7,115
Total Views 1,030,876 4,194,576
Average Views 520 589

Notice that the 2005 articles less than 750 words outperformed articles more than 750 words.

Time Period 2006-9-1 to 2006-9-27
Word Amount Articles >= 750 Articles < 750
Total Articles 2,842 16,463
Total Views 199,280 1,148,447
Average Views 70 69

Mini-Conclusion based on 750 word analysis: In the short-term, articles in the 750 word count range perform the same, but over the long-term, articles less than 750 words outperform articles with more than 750 words.

This same conclusions held true as I continued to run tests with higher word counts.

The only time the conclusion reversed itself is when I drilled into the 250 word count range:

250 Word, 27 days of Sept 2005 vs 2006 analysis:

Time Period 2005-9-1 to 2005-9-27
Word Amount Articles >= 250 Articles < 250
Total Articles 8,667 429
Total Views 4,983,966 241,561
Average Views 575 563

And here is the same stats for 2006:

Time Period 2006-9-1 to 2006-9-27
Word Amount Articles >= 250 Articles < 250
Total Articles 18,079 1,230
Total Views 1,263,572 84,897
Average Views 69 69

Mini-Conclusion based on 250 word analysis: In the short-term, 250 word articles perform the same as articles with more than 250 words, but over the long-term, articles greater than 250 words outperform articles with less than 250 words by a very insignificant margin.

I’m not sure I’ve proven anything here as the differences in results appears to be statistically insignificant, but perhaps I’ve proven that writing shorter word count articles (500 word count range), you won’t be penalized in traffic vs. longer articles and you might as well write (2) 500 word articles instead of (1) 1000 word article.

What conclusions did you pull from this data-mining exercise?

8 Comments »


1
Max writes:

Interesting study. I’m confused about the pageviews though…was traffic really that much better 1 year ago?

Max

Comment provided September 27, 2006 at 11:25 AM

[Reply]

2

Max,

It’s not that page views were better a year ago (hardly), but rather the view count is for all of the articles submitted during the Sept 1-27, 2005 time period, but with the total aggregate page views through today.

We never kept track of the time frames that articles received page views — While we can track which articles got added or approved in a given time frame, we can only review how many total page views it has through today. This is something we’re evaluating adding to the database for future studies.

2005 page views are really total page views for the articles from Sept 1-27th 2005 through today.

Comment provided September 27, 2006 at 11:31 AM

[Reply]

3
Edward Weiss writes:

Copywriters are continuously debating whether long copy is more effective than short copy. They say if it interests the reader, 2000 words or more is best because you’re engaging them and drawing them more into your “web” so to speak.

I tend to agree and I think this also can apply to article marketing. If something is interesting and reads well, I’ll most likely read it.. or at least scan it and read it later.

However, I won’t read short articles that are poorly written and have no worthwhile content. If I read an articles headline and am intrigued, they’d better deliver some substance in the body or I won’t read it.

Comment provided September 27, 2006 at 12:15 PM

[Reply]

4
Lance Winslow writes:

On the 250 word articles it appears to be close and if someone wrote a 750 word article and it out performs a 250 word article that makes sense over time. But with (3) 250 word articles to (1) 750 word article the (3) Swarm past the (1). In my estimations many of the articles I originally wrote with 900, 1500, 2000, 2500 or more have lots of traffic consistently, whereas some of the smaller articles 275 to 325 range died out around 350-450 article views over time. However, even those with 350-450 article views they still get some traffic and when I multiply those by 3 it seems that they are pull better than articles three times as much.

Additionally, I am interested in the article pick-ups ratios, because if each article picked up gets 300 article views somewhere else on average over a year or more then, what size articles are being syndicated more? That is a question I need to know because right now at; 52,759 article pickups, I would like to figure out how to get that to lets say 100,000 by mid-2007. And at the current 2,620,379 article views, it would be nice to see if that can get to 4,000,000 article views by mid-2007. I think both goals are possible based on the current growth rates. Do you have any data on # of article pick-ups per size of article?

Comment provided September 27, 2006 at 4:27 PM

[Reply]

5

Lance,

Currently, we do not have any stats on pickups in relation to article word count…but I’m sure we could add that to query report… I’m not sure it’ll provide any meaningful info?

Comment provided September 28, 2006 at 1:45 AM

[Reply]

6
Lance Winslow writes:

It matters to the article marketer who is targeting their articles to reach maximum syndication rather than merely article views probably, so this is why I ask. Of course a survey off all your Ezine Publishers who have been with you for so long could easily tell us their preferences too. It appears that those who pick articles up for websites might even have a different preferrence. I noticed the self help articles I have written really kick butt with the Ezine Pick-ups, but I imagine you have more Self Help motivational Ezine types and super industry subsector niche consultants who really are super positive people and they well, I just suspect they really like those positive articles on things like success, motivation and inspiration. Indeed just judging by you and your teams comments on this Blog I kind of get that from you too and so I assume that your entire staff is like that, like you that is and that is why you have attracted such a great group of people, vendors and even authors over the years? Well that is my impression, I just really get that feeling and seems to me that is why the ezine pickups are greater with positive self help articles. My question is I just wonder what size articles they are looking for. And I see you point about the use of Bullets and Numbers too, yes I see that in my articles, they tend to be the first ones picked up. Interesting. Anyways I am rambling, but I think that data would be interesting and worthy too for article marketers on a mission! The final frontier of article marketing at Flight Level 1O or 10,000 AGL.

Comment provided September 28, 2006 at 5:42 AM

[Reply]

7
Adam writes:

Would it not benefit you more if you took several of those long articles and broke them into several short articles. Would that not have better results?

Comment provided October 3, 2006 at 2:03 PM

[Reply]

8

Adam,

Yes, that is the conclusion that I was trying to further prove through studying the history of various word counts compared against the traffic generated.

In terms of traffic per article, the larger ones do perform better, but not statistically better… meaning, it’s clear that a higher volume of smaller articles will net you more results than a low volume of larger articles.

This study has influenced me to begin encouraging our authors to move towards the 400 word min. word count rather than skimming the rim with 250 words.

Comment provided October 3, 2006 at 2:09 PM

[Reply]

RSS feed for comments on this post.

Leave a comment

Please read our comment policy before commenting.