Wikimedia Deutschland/new-editors/Key Findings and Results/2020 Key Findings and Results

Overview

Activities

Associated Research

Available Materials for Adoption

Go back to 'Key Findings and Results' overview page.

'Thank You!' Banner Campaign 2020

What and why?

With the 'Thank You!' Banner Campaign 2020 ("Dankesbanner-Kampagne 2020"), we are once again addressing those users who contributed financially during WMDE’s Fundraising Campaign 2019 – we are sincerely thanking them for their support.

In addition to this ‘Thank You!’ gesture, we aim to recruit donors as long-term contributors to Free Knowledge and motivate them to become actively involved with the Wikipedia community. Concretely, we invite users who are viewing Wikipedia, to either become a supporting member of the nonprofit organization Wikimedia Deutschland, or to register as a Wikipedia editor via the campaign page ‘Learn Wikipedia’ (“Lerne Wikipedia”). As in the previous years, our Fundraising Team (FUN) and Volunteer Support Team (IF) are coordinating the campaign across teams. Within the context of the ‘Thank You!’ Banner Campaign 2020, IF and FUN are communicating their two main concerns – “Recruiting supporting members” and “Recruiting new editors”, but this time this is done for the first time using two independent banners (hereinafter called ‘FUN Banner’ and “IF Banner’).

Basic Parameters

• Type: Central Notice DE-WP
• Campaign Start: 1 January, 2020
• Campaign End: 19 January, 2020
• Playout: 8x max., 100 % of guest readers (not logged in)
• Prioritization: a. Playout: at least 50 % FUN Banner b. Timeline: 4x display of FUN Banner, after that 4x display of IF Banner

Results

Summary

The ‘Thank You!’ Campaign, which ran between January 1 and January 19, 2020, can be rated a success regarding the effect of the message and the landing page. This year, the call for recruiting new editors was run on a smaller scale compared with last year. Therefore, the absolute numbers for newly recruited editors were relatively low. The conversion rates lay within the average of the previous campaigns, nonetheless. With a number of tests, we were able to gain interesting insights into the effects of banners.

What did we want to know?

We wanted to answer the following questions with the analysis:

1. Registration: Is the message of the banner and the landing page appealing? (method: page views, registrations, and respective conversion rates)
2. Training modules: Do new volunteers use the training modules, and if so, do they also complete them?
3. Edits: Do users begin to edit Wikipedia after finishing the training modules? Is there a difference between those users who completed the training modules and those who did not?

Registration

Effect of the banner

The banner for recruiting new editors was displayed a total of 13,478,295 times. Of those, 5,409,335 impressions (40.13%) were displayed on mobile devices, 255,416 impressions on iPads (1.89%), and 7,813,544 impressions (57,97%) on desktop computers.

Analysis of banner interactions

The closing rate for the banner was on average 5.87%. For 2.71% of banner impressions, the banner was subsequently unfolded, and thus the banner text and the button for referral to the landing page were displayed.

TABLE 1: Unfolding rate for the banners by variant and platform:

Platform	Banner Variant	Unfolding Rate
Desktop	CTRL	0.51 %
Desktop	VAR	0.55 %
iPad	CTRL	3.18 %
iPad	VAR	3.54 %
Mobile	CTRL	6.57 %
Mobile	VAR	5.8 %

TABLE 2: Closing rate for the banners by variant and platform

Platform	Banner Variant	Closing Rate
Desktop	CTRL	3.37 %
Desktop	VAR	2.75 %
iPad	CTRL	5.29 %
iPad	VAR	5.38 %
Mobile	CTRL	11.63 %
Mobile	VAR	10.85 %

From these figures, it becomes apparent that there are fewer interactions happening on desktop devices compared to mobile devices – both fewer closings and fewer unfoldings. In addition, there is a slightly higher closing rate for banner variant 1 (CTRL).

Inside the banners, there were two links leading to the landing page: one in the text and one in the button with the call-to-action ‘Try now’ (“Jetzt ausprobieren”). The analysis shows that 346 page views came from the link in the text (15.63%) and 1,867 page views came from the button (84.37%).
Back to top

Evaluation of the A/B Tests

This year, we conducted an A/B test, in which two different variants of the banner text that appears after unfolding the banner were tested. The first variant of the text (Banner CTRL) was different from the second one as it included an additional paragraph with a three-step explanation on how to participate for interested users:

Participating is very easy:

Step 1: Create a user account in Wikipedia.

Step 2: Learn what is important for editing Wikipedia.

Step 3: Find help and get to know the community behind the project.

This explanation is also included on the landing page ’Learn Wikipedia’ (“Lerne Wikipedia”). The second banner (Banner VAR) did not include this paragraph.

Regarding page views for the landing page ‘Learn Wikipedia,’ the following picture emerges: The conversion rate for banner impressions to page views is higher for the Banner VAR with the shorter text. This means that when the banner with the shorter text was displayed, more users clicked on it and viewed the landing page.

TABLE 3: Conversion rate for banner impressions to landing page views by banner

Banner CTRL	0.0128822%
Banner VAR	0.0198188%
Banner Total	0.0164190%

From these results, the following hypotheses can be deducted:

• The shorter the better: Users prefer shorter texts
• The simpler the better: Users are scared away by complex-sounding activities (three steps)

Regarding the second hypothesis, the opposite could be argued, too, i.e. that it can be more beneficial in the long run to openly communicate all necessary steps to users from the beginning, so that only those who are strongly interested in participating click the button. This would be reflected, for example, by a higher level of registration and editing activity by those users who came through the first banner (CTRL).

Differences in Registrations

The analysis of the registration and editing behavior in relation to the banner version does not suggest that the banner text influences the decision to register. While ten people registered via the first banner (CTRL), eleven people registered via the second banner (VAR) with the shorter text. All edits were made by people who came via the second banner (VAR). As only four people in total made any edits, this number is, however, too small to draw any conclusions from it.
Back to top

Effects of the landing page and registrations

On the landing page ‘Learn Wikipedia’ (“Lerne Wikipedia”), interested users are called upon to qualify for participating in Wikipedia in three steps and to become active as an editor:

Creating a user account Completing training modules Editing pages that need revision The landing page was viewed a total of 2,213 times, that is equivalent to a conversion rate of 0.016419 percent.

While this conversion rate is very low, it is still higher than the conversion rate of last year’s ‘Thank You! Campaign’ (0.01220 percent). The low conversion rate might be caused by the fact that the banner for recruiting new editors was only displayed on the fifth view. This means that users had already seen the ‘Thank You!’ banner four times in all cases, and had possibly already interacted with it, which might have reduced their interest in an additional banner interaction.

Throughout the duration of the campaign, a total of twenty-one new Wikipedia accounts were registered via the landing page ‘Learn Wikipedia’ (“Lerne Wikipedia”). This amounts to a conversion rate for page views to registrations of 0.9489 percent. This conversion rate is slightly above the average of other banner campaigns. It has to be noted, however, that due to an error in the assignment of tracking tags, from day two onwards, registrations that were made on desktop devices, could not be tracked. It has to be assumed that more registrations took place. From previous campaigns we know that most registrations are made on desktop devices while only a few are made on mobile devices.

Training modules

The following training modules were advertised on the landing page: Editing Basics and Wikipedia Basics, Evaluating Articles and Sources, and Discussing with Source Code. Only one new editor completed a training module – the module on editing basics.

Edits

Due to the low number of new registrations, the total number of new users who made edits was very low. Overall, four users made edits. Percentage-wise, the numbers are slightly below the average of other banner campaigns. The edits from the newly-registered Wikipedia accounts were analyzed from the start of the campaign on January 1, 2020 until January 27, 2020.

TABLE 4: Edits by newly registered editors

Category	Number of Newly Registered Editors	Percentage of Newly Registered Editors
0 - 1 Edit	1	4.76 %
2 - 4 Edits	1	4.76 %
5 - 9 Edits	1	4.76 %
10 - 49 Edits	1	4.76 %
Total	4	19 %

Back to top
Go back to 'Key Findings and Results' overview page.