Designing audio experience for digital publishing
Aftenposten — Norway’s leading online newspaper with always updated news in domestic, foreign, sports and culture.
Background
After doing extensive desktop research we can safely say the way people consume media is changing, with audio becoming a more central part of everyday life. Several publishers such as the Washing Post, News24, Nettavisen, Forbes Italia have started implementing text-to-speech services.
The Danish publishing house Zetland started having the journalist read up the articles. They wrote and report that about 75% of all articles are consumed via audio.
More and more major publishers and media creators are focusing on audio news and spoken word. And we expect this trend to continue.
So we had the hypothesis that:
Audio Articles will play an important role in news consumption and attracting more users to Aftenposten and increasing subscribers.
To test the hypothesis we asked ourselves the following question.
How do Aftenposten readers react to audio articles?
After desktop research we were asking ourselves:
Can we make audio articles good enough to attract new subscribers?
Can we create a habit for the users to use audio?
Can we create a superior voice experience compared to other publishing platforms?
Design Sprint
We wanted to understand how users respond to playback of pre-recorded human read articles, compared to regular text and image-based articles.
Based on design Spring problems & hypothesis we came up with a Solution: Static and sticky player. We tested the prototype with an article and audio recording of “Gutten som aldri fikk fred”, which was 26:57 min long.
During the interview:
- Context questions about news consumption
- Test prototype including follow-up questions
- Task questions, related to news and audio consumption.
- Deep-dive questions
Participants:
- 6 Norwegians with the following criteria (3 female & 3 male)
- Ages between 30 and 60 years old, living around Oslo.
- Read Aftenposten at least once a week.
Key takeaways:
- They wanted to listen to news not a feature story
- Article was too long and they didn’t like the theme of the article
- Users prefer consuming news through the phone and while commuting.
- Users confuse audio articles with podcast episodes.
- Article content determines whether users will engage with the article.
We considered User Feedback about player and usability, therefore created player matching with the Aftenposten brand.
1. We made sure the typography and icons are visible and easy to read
2. Made the player less intrusive — de-cluttered the view of the users who wanted to read.
3. Unfortunately we implemented only static player due to technical difficulties.
Engagement
How do we know that readers will not engage with this article?
We decided to implement an audio version of the article in production on the Aftenposten website.
Our initial hypothesis was that some users are not interested in the long feature article audio, however there was a significant amount of approval for the audio feature in these featured articles showing that many users may prefer to engage with the audio.
CTR is not the best metric, because people might be curious about a new feature or they clicked accidentally. That is why average listen length is really important.
Seeing the results it is clear that those who are interested in listening to the audio version will listen to 75% of the whole article.
How do we know that 3,6% is good?
We had a meeting with the Senior Product Manager of The Washington Post and he claimed that anything close to 1% is really good!
We consider 3,6% CTR really good, at the moment. We think it might change
over time because of Novelty effect.
Novelty effect, is when users sees something new on the website and specifically interact with the feature just because it’s new.
When it comes to behavior and content consumption, readers are typically inconsistent, but that doesn’t stop them from demanding/expecting a certain level of consistency in user experience. Providing a consistent platform will build trust over time and form new habits around consuming news.
Future ideas:
- Monitor growth and monthly return to the audio player and see who has
established a new habit to listen to the articles. - It would be interesting to find out in what context users prefer to listen to
audio articles.
Previous tests were conducted with pre-recorded human read articles, but we were wondering how many humans would we need to use per article?
How much more expensive is it to use human recorded articles than automated voice integration on articles?
That is why we decided to test significance between two voices. How do listeners respond to synthetic versus human-read voices?
Listeners prefer human voice, but from this test we can see the statistical significance is 14%, which is not large.
We expect that introduction of the custom synthesized voice would close the gap between human and robot voice.
“Fake door”
With the “Fake door” or “Smoke” test we wanted to understand the inclination of users who are willing to have this feature as part of the subscription.
The test was considered to be in production behind the paywall of Aftenposten.
- Only non-subscribers could interact with this test.
- The first thing users recognize is the player information panel, leading to a feedback form with an article audio sample. We considered that users would be annoyed to have a long feedback form, therefore we asked three simple questions about the audio.
- To avoid user disappointment, we considered providing a full audio article version after they submit the feedback form.
We planned to have this test before internship finishes, but unfortunately we did not conduct it because of technical difficulties.
We do hope Aftenposten will continue with this test and research further opportunities for audio articles.
However, regardless of whether or not we were able to do the backdoor-test. All the results from the design sprint, test of engagement and the A/B test point towards one possibility..
Audio has a future in Aftenposten!
Summary
After the tests we ran, we spoke with the Senior PM of Audio in the Washington Post who has implemented Text-to-speech as well.
And he concluded something that we predicted during our design sprint.
Which is to in order to solidify the role audio plays in article consumption, we should consider to create an experience to remind more of a podcast rather than reading the news.
And we agree. This entails for example getting a curated playlist which is based on the user preferences.
They can queue articles. And the front page is designed with audio in mind. By creating a more audio focused experience we can increase retention rate, increase subscription value and create a better experience for the users.
With this in mind while designing I had many questions about playlists and front page experience, it seems like Aftenposten would need to create a completely new way how readers/listeners interact with Aftenposten website and App.
