Account Creation Improvement Project/Research/Results

This page is currently inactive and is retained for historical reference.
Either the page is no longer relevant, or consensus on its purpose has become unclear. If you want to revive discussion regarding the subject, you should seek broader input via a forum such as the Village pump.

You can get involved:

	Spread the word by signing up and getting your community involved.
	Help build new messages for the account creation process by clicking here.
	Discuss the open questions
	Participate in the discussion on the project's talk page
	Use the Toolserver if you have an account and analyse the existing data.

Terminology

So we don't confuse ourselves, here is a chart of the account creation process:

login page - where you end up after you click "login / create account"
create account page - where you end up after you click "create one", with the fields that you need to fill in
landing page - where you are sent after you have filled out the fields and clicked the Create account button

We will test various versions of the three pages against each other. This is called A/B testing. 'A' stands for the page we have right now, and 'B' stands for the new version. We can test as many B:s as we have time for, and measure how they fare against each other. We will do these tests on several language versions of Wikipedia, beginning with A/B-testing the landing pages (3), and then the create account pages (2) and lastly the login pages (1). For a visual representation of this, see this picture:

To make it easier to follow the research, it is divided into several pages. This page deals with the numbers of the account creation process. Then there are three pages for each of the steps in the process: step 1, step 2 and step 3.

Existing data

How many people click on "Don't have an account? Create one" and don't create a user account (i.e. don't submit the form via "create account" button)?
How many people create a user account per day on the various projects?
How many people create a user account and make at least 1 edit within the following 10 days?

How many people click on "create user account" and don't follow through?

From September 22 on, we're logging user clicks on the Wikimedia logging server to get a daily "account abandonment". We turned on clicktracking for the "Create account" page. We measure how many people land on this page and how many of those actually fill out all the information and click on the "Create account" button.

After running the clicktracking for about 6 hours on September 22, we estimate that about 21% of users follow through the current account creation process. This is for all languages and all Wikimedia projects. More data and project breakdowns to follow.

How many people create a user account in a given month?

The following data has been extracted from the Wikimedia Toolserver.^[1] All data is for August 2010:

Project	Number of new user accounts overall	New user accounts: average per day
English Wikipedia	169,706	5,474
Spanish Wikipedia	27,102	874
German Wikipedia	17,963	579
French Wikipedia	16,712	539
Russian Wikipedia	12,653	408
English Wiktionary	8,222	265
English Wikibooks	5,893	190
Dutch Wikipedia	5,743	185
Polish Wikipedia	5,393	174
Swedish Wikipedia	3,992	129
English Wikinews	2,801	90
Czech Wikipedia	2,686	87
Hungarian Wikipedia	2,602	84
Hindi Wikipedia	1,293	42
Tamil Wikipedia	807	26
Bengali Wikipedia	677	22
Malayalam Wikipedia	616	20

How many of these new user accounts are being used for editing?

The following data has been extracted from the Wikimedia Toolserver.^[2] All data is for August 2010:

File:Percentage of new user accounts with at least one edit (August 2010).png

Percentage of new user accounts with at least one edit within 10 days after account creation (August 2010)

Project	Number of new user accounts	New accounts with at least one edit within the first 10 days after creation	Percent- age
English Wikipedia	169,706	51,864	31%
Spanish Wikipedia	27,102	6,769	25%
German Wikipedia	17,963	6,221	34%
Russian Wikipedia	12,653	3,335	26%
Dutch Wikipedia	5,743	1,413	24%
Polish Wikipedia	5,393	1,255	23%
Swedish Wikipedia	3,992	1,190	30%
English Wikinews	2,801	2,798	100%
Hindi Wikipedia	1,293	121	9%
Tamil Wikipedia	807	82	10%
Bengali Wikipedia	677	74	11%
Malayalam Wikipedia	616	70	11%

How many new users make their first edit on the day of account creation?

The following data has been extracted from the Wikimedia Toolserver.^[3] All data is for August 1–15, 2010 (smaller sample than above):

Project	New accounts with at least one edit within the first 10 days after creation	New accounts with at least one edit on the day of account creation	Percent- age
English Wikipedia	24,881	22,885	92%
Spanish Wikipedia	3,396	3,022	89%
German Wikipedia	2,894	2,617	90%
Russian Wikipedia	1,515	1,332	88%
English Wikinews	1,336	1,336	100%
Swedish Wikipedia	548	510	93%
Hindi Wikipedia	50	41	88%
Tamil Wikipedia	34	30	87%

How many new users edit their user page at least once shortly after account creation?

The following data has been extracted from the Wikimedia Toolserver.^[4] All data is for August 1–15, 2010 (smaller sample than above):

Project	New accounts with at least one edit within the first 10 days after creation	New users who edit their user page at least once on day 1	Percent- age	New users who edit their user page at least once within the first 10 days	Percent- age
English Wikipedia	24,881	1,611	7%	2,045	8%
Spanish Wikipedia	3,022	113	3%	152	5%
German Wikipedia	2,894	200	7%	269	9%
English Wikinews	1,336	12	1%	13	1%

References

↑ select DAY(user_registration) as day, count(user_id) from user where user_registration IS NOT NULL AND user_registration>"20100800000000" AND user_registration<"20100900000000" group by day;
↑ SELECT COUNT(*) FROM (select DISTINCT user.user_id from user left JOIN user_daily_contribs ON user.user_id=user_daily_contribs.user_id where user_registration IS NOT NULL AND user_registration>"20100801000000" AND user_registration<"20100802000000" AND day>="2010-08-01" AND day<"2010-08-12") as dummy;
↑ SELECT COUNT(*) FROM (select DISTINCT user.user_id from user left JOIN user_daily_contribs ON user.user_id=user_daily_contribs.user_id where user_registration IS NOT NULL AND user_registration>"20100801000000" AND user_registration<"20100802000000" AND day="2010-08-01") as dummy;
↑ SELECT user_name, COUNT(*) FROM user, page, revision WHERE page_namespace=2 AND page_title=user_name AND rev_page=page_id AND rev_timestamp < '20100802000000' AND user_registration IS NOT NULL AND user_registration >= '20100801000000' AND user_registration < '20100802000000' GROUP BY user_name; and SELECT user_name, COUNT(*) FROM user, page, revision WHERE page_namespace=2 AND page_title=user_name AND rev_page=page_id AND rev_timestamp < '20100811000000' AND user_registration IS NOT NULL AND user_registration >= '20100801000000' AND user_registration < '20100802000000' GROUP BY user_name;

Survey 1

A statement from Wikimedia fellow Lennart Guldbrandsson:
Hello,

A couple of days ago, I asked admins on 7 language versions of Wikipedia (English, German, French, Spanish, Swedish, Bengali and Bahasa Melayu) to put a link to a very short survey on the page that new users "land" after they have created their new accounts. The survey asks them about what they expected when they created the account, or in essence, why create an account on Wikipedia?.

We intend to keep this survey running until February 4th and then analyse the results. This will be a helpful step in understanding what we need to change during the upcoming change in the account creation process.

But let's not wait until the middle of February to look at the results. Here is a small sneak peak of what people have chosen (for the full list of alternatives, see above):

out of 592 respondents, 67,4% on English Wikipedia say that the reason they created their account was... "I wanted to be part of Wikipedia".
this is, at 69,2%, also the most common answer on German Wikipedia
... and on Bengali Wikipedia (75,5%)
... in fact it's the most common answer across the board
the other two common alternatives are "I usually register for websites I like" (24%) and "I thought an account is required to edit articles" (26%)
even on Wikipedias where you don't have to have accounts to create articles, people assume that they do need accounts to do this

Some people have posted things in the comments field. Here are a few comments that are typical:

"I don't want to edit articles anonymously."
"Class assignment"
"I LOVE WIKIPEDIA"

And...

"Dunno. I was bored."

Right now, we have 1024 responses to the survey. And now, 1028. And in a few minutes even more. With the service we use (SurveyMonkey), it's really easy to track what people have chosen and said. After the 4th, I will do a quick analysis of the results, and hopefully have the results ready on February 7th. By then we should have many more responses, which means that the answers we get will be a very helpful guide for us. If you want to help change the process without doing the survey, you can help out in several ways, including by doing research or suggesting features of the new process. See the main page for the project for more information.

NB: The survey has not yet been launched on French and Bahasa Melayu. If you are an admin on those language versions, please contact me. And we would like to do the survey in even more languages, so post your translations of the survey on this page, and I will put them up along the others.

Finally, please sign up here to become part of the team that will change the account creation process - and spread the word about this. Our goal is to have changed the process by June and during that time increase the numbers of new users that start and continue to edit Wikipedia.

Thank you for reading this.

Best wishes//Hannibal 15:36, 26 January 2011 (UTC)[reply]

[1] select DAY(user_registration) as day, count(user_id) from user where user_registration IS NOT NULL AND user_registration>"20100800000000" AND user_registration<"20100900000000" group by day;

[2] SELECT COUNT(*) FROM (select DISTINCT user.user_id from user left JOIN user_daily_contribs ON user.user_id=user_daily_contribs.user_id where user_registration IS NOT NULL AND user_registration>"20100801000000" AND user_registration<"20100802000000" AND day>="2010-08-01" AND day<"2010-08-12") as dummy;

[3] SELECT COUNT(*) FROM (select DISTINCT user.user_id from user left JOIN user_daily_contribs ON user.user_id=user_daily_contribs.user_id where user_registration IS NOT NULL AND user_registration>"20100801000000" AND user_registration<"20100802000000" AND day="2010-08-01") as dummy;

[4] SELECT user_name, COUNT(*) FROM user, page, revision WHERE page_namespace=2 AND page_title=user_name AND rev_page=page_id AND rev_timestamp < '20100802000000' AND user_registration IS NOT NULL AND user_registration >= '20100801000000' AND user_registration < '20100802000000' GROUP BY user_name; and SELECT user_name, COUNT(*) FROM user, page, revision WHERE page_namespace=2 AND page_title=user_name AND rev_page=page_id AND rev_timestamp < '20100811000000' AND user_registration IS NOT NULL AND user_registration >= '20100801000000' AND user_registration < '20100802000000' GROUP BY user_name;

[1]

[2]

[3]

[4]