Deep Dive of Steemit Tags

in stats •  7 years ago  (edited)

Choosing tags is often a last minute after thought but well chosen tags can give your post increased visibility.



Today I will examine 4 popular tags and look at some of the relationships to post payouts.


Steemit Tags

On Steemit you can use up to 5 tags, they can be any word you choose but people expect them to be some way related to your post. Using tags just because they get high payouts may lead to flags or downvotes if your content is not related to the tag.
For example the introduceyourself tag has been a place where new users would post an intro and often would get big upvotes from Whales as a kickstart on Steemit. People have been abusing this tag by posting multiple intro posts or just tagging it with the hope that a completely unrelated post would get an upvote.

Today I am going to examine 4 popular tags:
Photogrpahy, Art, Photo, introduceyourself

Overview

The aim here is to identify variables that lead to higher post payouts on these 4 tags for newer users, as well as trends in payouts for these tags since December.

Before digging into different relationships we will begin by looking at some aggregate figures for these tags.

Categories

There can be some overlap between the tags which I will not analyse in detail in this post. For the following analysis each grouping will be based on the first tag of the post only.
This is the most visible tag on a post as it appears as the category and gets added to the permalink of the post and is likely to be the main tag used by newer users.
e.g. if photography was the second tag I would not be including the post

New users

For our analysis we wish examine the more regular experience for people on Steemit. People who have a lot of influence and have been here for a while are likely to have a higher Rep. For the rest of this analysis we restrict the data to people with Rep > 30 and < 50
The people posting on Steemit can be put into several different groups. There are people who have been on Steemit for a long time and may have achieved a lot of influence (as well as high Rep), and others who spam tags (probably have a low Rep).

The following graph shows the numbers per week for these 4 category tags since December.

Totals_30-50.png

  • There was increased activity due to the SBD bonanza, increases in the price of Steem as well as many new users signing up to Steemit over this period.

Averages (per post)

So how much does one of these Average Person make per post in these 4 categories?
Means_30-50.png

  • Pretty impressive?, between 1 & 1.5 Dollars for some weeks!

This is the average value. How does the median look like?
Median_30-50.png

  • This feels more like the experience of the vast majority of Steemians.

We will now divide the data into 4 groups based on the first tag.

Payout Per Post Over time

Do spammers get diminishing returns?
The next graph shows the payouts on posts ordered by the number of posts made by each author in this period.
i.e. If an author made several posts, the payout on the first post would be at position 1, the payout on the second post at 2 and so on.

You may expect spammers that create a lot of posts to get a diminishing return.

introduceyourself_Payouts vs Count.png

  • The introduceyourself tag is one of the tags that people generally are advised not to spam. I am very surprised to see so many posts by some authors, up to 30!

photography_Payouts vs Count.png

  • Payout on this tag seems to be greater for authors with a lower frequency of posting.

art_Payouts vs Count.png

  • There are smaller numbers on the art tag but again payout seems to be higher for posts where authors posted less. Maybe an indication of quality?

photo_Payouts vs Count.png

Average Values in Each Category

Next we will examine the Average values per week in each category and compare how much each author made in aggregate and per post

introduceyourself_post_averages.png

introduceyourself_author_averages.png

photography_post_averages.png

photography_author_averages.png

photo_post_averages.png

photo_author_averages.png

art_post_averages.png

art_author_averages.png


Correlations

It can sometimes feel like payouts on posts are random.
Is there any correlation between the variables such as number of votes, comments etc to payout?

introduceyourself_Post_Correlation.png
Blue represents a high Pearson Correlation and red a low correlation

photography_Post_Correlation.png

photo_Post_Correlation.png

art_Post_Correlation.png

Number of Comments seems to be the highest level of correlation but event that is a very low statstically

Payouts Per Author

What if instead of looking at posts if we group the data by Author to see is there a correlation to aggregate payout to author features?

Correlations


introduceyourself_Author_Correlation.png


photography_Author_Correlation.png


photo_Author_Correlation.png


art_Author_Correlation.png

Conclusion

There does not appear to be a strong correlation between any of the factors we looked at and the post payouts for this group of authors.

The first few posts by authors in these categories got higher rewards but there is diminishing return after that. It's likely that authors are moving on to other tags, or not using these 4 as the first tag after an initial period.

Does this analysis reflect your experience with using these tags?



Thank you for reading this. I write on Steemit about Blockchain, Cryptocurrency, Travel and lots of random topics.


Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

Nice post, thanks for the update, appreciated

that great analysis ! actually @introduceyourself tags are much uses and more and more people are upvote them and welcome him/her!
yaah 4 tags are really hidh payout in steemit and much uses!

discussion is an exchange of knowledge an argument an exchange of ignorance the aim of argument or of discussion should not be victory but progress. discussion is always better than argument because an argument is to find who is right and a discussion is to find what is right...

stay blessed and thanks for sharing keep it up broo......

Nice blog👌👌👌

Wow, this is good stuff @eroche. Interesting to look at this from in more scientific way :-)

Did you have any specific hypotheses upfront yourself or a strong hunch?

@cryptotem
Good question! The analysis was prompted my me. @eroche lent his time and deep expertise to conduct this analysis in a scientific way. i thank him once again for that.
Overall, while it was not the reason for the work, it is important to understand the role of tags. It will become increasingly important to get content sorted as the platform grows.

That is indeed quite generous of @eroche :-)

I agree that tags fulfil an important role on the platform. Some time ago I actually asked myself the question whether 5 tags isn't a bit too much. I think if we would only be able to use 3 tags, it forces you to really think well about them. It also would point out clearer winners (cryptocurrencies vs crypto for instance). Also, you could say it would make the tags more accurate.

Generally a piece of content has a certain focus on probably 1 or 2 subjects. By giving everyone the opportunity to pick 5 tags, usually the last few are related but less relevant or just used for visibility purposes.

What do you think @liotap?

Great analysis about tag . I think tag should be related to post in which man do't feel boring during any post. Your advice really impressed me . thank you for sharing your post with us .

I have come across using the introduceyourself tag, but my intension was good as it was a post about mistakes to avoid when joining steemit, but I did get warned by a bot and have learn't from it. Luckily I didn't get flagged. I still have difficulty choosing the correct tags but it is still all new to me. Have read your article and it makes a little more sence now. So I will see how my next post goes.

that s what it is all about. trial and error, social experimentation. you got it!

Nice Post @eroche...
Let's get used to using the correct hastag or tag according to post content...!!!

Thenk you

So, what should we do to get a better reputation and get a reward that worth it ...?

@eroche thanks for sharing.

I was wondering why you would want to show rewards in SBD as opposed to the number that appears on steemit next to the upvote button. In my view, it would be more comprehensive as it would capture both SP and SDB as most people do a 50/50 payout. SP counts!

Thanks for pointing this out @liotap, the values for payouts in the graphs represent the $ value of the posts as shown on the Steemit website. Because the SBD is so out of synch with the Dollar at the moment it represents a mixture of SBD and US $. I hope we get back to the peg soon and it will make this analysis more clear :)

thanks a lot for the clarification @eroche! You are the only user I know who hopes SBD goes back to $1! I'm pretty sure most people feel like the higher the better (i.e. people can earn more USD denominated value from producing content).

For the long term health of the platform I think it's vital for users to get some stake in Steem Power as part of each payout.

  • The higher the SBD price the less of each post that gets paid in Steem Power.
    Sure there is a short term benefit to individuals but looking out over a longer time horizon it may be less sustainable. What in effect is happening is value is being taken out of the system at an accelerated rate. You no longer have to wait to power down.
  • The second issue I have is (like all cryptocurrencies) it makes it very hard to engage in any form of commerce as you cannot set a price for a good or service over any time frame with certainty.
    I run monthly competitions and it makes it very difficult to set a prize fund as I don't know what its going to be worth in a months time.

this is a great post. thanks for sharing... i appreciate your post

Thank you for your work and sharing the results.

Thank you for that. Its so refreshing to see actually helpful and analysis and statistics to help people rather than the usual 'guide to steemians' written by people with reps over 60 whoi cant think of anything new to write.
Much appreciate your hard work, Cheers