UXperiments – 24/07/2017

Helping users discover new content on BBC iPlayer for mobile.

In today's world, we've never had so much access to new content. The average newspaper has more information in than someone would obtain in their entire lifetime a few centuries back. That's a newspaper, now think about how many lifetimes worth of information there are on the internet.

Having this amount of information at our fingertips is, in my opinion, one of humanity's greatest achievements. But it's made it bloody hard to decide what to watch on an evening. With this in mind, I set about finding a way to help users discover new content on BBC iPlayer; following the 5 steps of the design thinking process outlined below.


1. Empathise

James Mellor Task Presentation-4James Mellor Task Presentation-4

Defining Personas

As the difficulty in finding and choosing what to watch was a problem I'd identified myself, I built up a core persona around this; "20 somethings". This also meant I could gain first-hand user insight from people I knew within this group; and gain fast feedback when testing.

James Mellor Task Presentation-5James Mellor Task Presentation-5

User Research

To start, I asked a group of 8 friends within this persona what bugged them with iPlayer and discovering new content in general. Using Watsapp to do this allowed me to get fast responses, and a conversation going within the group.

The most useful insight was around the algorithms used to recommend content. These were described as "repetitive and single minded"; hindering discovery by recommending content too similar, and not accommodating for multiple interests.

James Mellor Task Presentation-6James Mellor Task Presentation-6

Competitor Analysis

As multiple streaming services were mentioned in research, I took a look at how competitors help users discover new content.

Of the most popular video services, the user was overwhelmed with choice; little of which was labelled as personal recommendations. Streaming services seemed to prioritse overall popularity and current shows when landing on their home screen.

One service that did stand out, however, was Spotify. Spotify's Friends option allows you to see what your friends are listening to, as well as their own playlists. Recommendations from friends feel more human than those from an algorithm, and the social needs of users are likely to spark interest in these.


2. Define

Defining the problem(s)

What makes discovering new content on iPlayer difficult?

How might we make discovering new content more human?

After defining the problems with others within the user group, the following 3 key themes emerged to drive the Ideation phase.

1. Media Relationships

The Blumler & Katz Uses & Gratifications theory examines how and why people consume media. A key aspect is Media Relationships, where media is consumed to help build personal relationships with others (think talking about last night's Corrie in the office). This social need was an area I thought might influence a more human discovery experience.

2. What shall we watch tonight?

I'm sure most people can relate to this. You're sat at home with your partner or friend, you open iPlayer or Netflix; half an hour later you're either still deciding or have just watched the titles of 5 different programs — and decided to just go to bed.

3. Beyond Algorithms

Algorithms are great, but the machines aren't taking over just yet. As previous research identified, recommendations from algorithms are often reptitive, and don't encourage exploration outside of what you've already watched.

3. Ideate

Initial Ideas

How might we make discovering new content more human?

After exploring a range of potential solutions, I decided to take the following ideas forward:

1. A recommend feature linked to social media

2. A shared online viewing experience

3. Content structured on what friends recommend

1. Recommend Feature linked to Social Media

My first idea was to use the mobile device as a remote while watching shows, with a scene sharing button to push to social media. From here, the user's friends could pick up recommendations. As I began exploring this with sketches, I found it was moving more towards Social than Discovery. There were also some issues raised with how the scene sharing button might interrupt viewing.

2. Shared online viewing experience

Drawing on the insight of the core persona generally living away from friends and family in different cities, I explored the idea of a shared online viewing experience. This was inspired by how Twitter often blows up when a popular show is aired. Creating an in-app feed of comments and reactions would build a social aspect around a particular show, making people all over the country feel connected. In a similar way to the first concept, this idea felt like it had moved too far away from discovery; and if anything, would become a distraction from the show itself.

3. Content structured on what friends recommend

When everything starts to get a bit messy and chaotic on the sketchpad, I know I'm onto something. I started to work up some ideas about an in-app "Friends' Picks" feature. This allowed the user to recommend shows to specific friends, or their full list. They can then see what their friends have been watching; both narrowing down and influencing their decision when discovering new content. This brings a more social, human aspect at the beggining of the decision process; which doesn't create a distraction from the shows. I decided to take this forward into a prototype.

4. Prototype


Flagged as a new feature at the top of the home screen, the Friends' Picks feature uses icons to denote how many new and personal recommendations there are. This instantly positions the content as fresh, personal and relevant; sparking user interest. To encourage early adoption of the feature, I planned to introduce Guest Picks; where famous BBC actors/personalities list their favourite shows.

Marvel Prototype


Friends Picks Landing

To reduce cognitive load, just the top 5 picks from friends would be displayed at the top; with friend-to-friend recommendations prioritised. With live TV not being the preferred choice of my target persona, I chose to show genre rather than channel above the title. An indicator to show many friends have watched the programme further appeals to the Media Relationships need to be involved.

Marvel Prototype


Friends List

A simple tab at the top allows the user to view their friends list. Looking for a specific friend creates a more human feel; seeking recommendations from others based on what you know about them. It also encourages further conversation around the shows offline – "Oh, you watch that too?!".

Marvel Prototype



Friend Profile

The individual profile page prioritised any personalised recommendations for the user. Comments add a more human touch.

Marvel Prototype


Recommend Feature

To ensure the Friends' Picks feature incorporated with the current user journey, I added a simple Recommend button to the programme page.

Marvel Prototype


Recommend This Show

To make adding a show to your picks as easy as possible, I simplified the final step into the key components: Adding a message, selecting specific friends, and a "Send to All" call to action

Marvel Prototype


Select Friend

While the call to action defaults to Send to All, if the user selects specific friends it begins to count them up

Marvel Prototype



A simple confirmation page uses a conversational approach, displaying the show that has been recommended. Some interaction design would bring this to life in development.

Marvel Prototype


5. Test

James Mellor Task Presentation-24James Mellor Task Presentation-24

User Testing

I'd usually like to get out there and conduct guerilla testing at this point; but with limited time, I picked 3 users I knew within the core "20 Somethings" persona. Each person was selected based on their suitability; and  for the honest, qualitative feedback I'd get back from them.

I devised a test scenario for them to follow, and have collected the results below.


James Mellor Task Presentation-25James Mellor Task Presentation-25

Test Results

I used the same test scenario for each subject; observing Sarah moving through it, and gaining remote feedback from James and Rob.

The feedback was generally quite positive. Users found the journey easy to navigate, and didn't feel like the feature needed any onboarding to understand it.

A key finding was that it was difficult to find the personal friend-to-friend recommendations without having to click into individual profiles. Being able to recommend a show from a thumbnail was a suggested feature. And the feedback on the friends list was that new recommendations were more interesting that total numbers — sorting friends by shows in common was also suggested.



6. Hi-fi Prototype


Hi-Fi Prototype

Working within the BBC's GEL guidelines, I began to work the wireframes up into a Hi-Fi prototype. Following the previous user feedback, I introduced the personal picks and comments on the main landing page. I also added New and Just for You to the Friends List. I chose to use circles for the profile pictures as this is what users are familiar with on social media; it also clearly distinguishes them from viewable content.

Using Marvel, I added a few simple interactions to the prototype. I also created a slide to help pitch the concept to key stakeholders below.

View Marvel Prototype


James Mellor Task Presentation-27James Mellor Task Presentation-27

7. Summary


How has this excercise addressed the original task of "helping users discover new content on iPlayer for mobile"? Through exploring the idea of a more human discovery process, I found that recommendations from friends could help refine choices and influence decisions in a world of content overload.

Users have a social need to be involved through media, sharing certain shows they watch; and building relationships out from these. Friends Picks add further confirmation that the user may like the content, as they trust their friends. It also adds a more human touch to the process that a cold, repetitive algorithm. The algorithm could be better used here to determine friends with common interests, rather than the interests themselves.

What I'd do next

While my research and testing has shown demand for a Friends' Picks feature, I'd like to look at a wider audience sample to determine if this would work if launched. Quantitive research about how people recommend and discover new shows would better inform my design decisions. Guerilla testing would provide feedback on the user journey, and the usefulness of the feature to people.