Tackling big redesigns with humanity

A talk from Bo Ren at Canvas 2017 - lessons from big redesign projects.

Bo Ren describes her career as a ‘comet’s journey’. A comet orbits different planets temporarily; Bo’s gone from Law School, to a utilities company working on behavioural sciences, to product management at Sigfig.

More recently, she’s worked on new features at social giants Facebook, Instagram and then Tumblr, where she was part of the team that redesigned the mobile expressions toolkit.

Her talk focused on the tricky subject of redesigns at scale; how do you know when you should consider one, and how you should approach such a mammoth task.

When should you consider a redesign?

1. Technical constraints

Firstly, you might need to redesign because of a big change in your product – in other words, the legacy stack can’t support the new features needed.

2. Design paradigms change

As UI trends accelerate and designers create new languages, sometimes there is a need to keep up with visual changes in a UI. A good example would be iOS going from skeumorphic to flat.

3. Business evolution

Perhaps, the model at the business has evolved. Bo gave the example of SigFig, which began as a portfolio tracker, but pivoted to a portfolio manager. This was a big change of model, so the old designs no longer translated to the future direction.

4. Keep up with competitors

As brands evolve and competition changes, redesigns are sometimes necessary. Bo cited the need for Instagram to introduce the ‘Stories’ feature to compete with Snapchat. The brand sometimes has to evolve to keep up.

Iterative vs Overhaul

Bo then went on to describe the two typical approaches to redesigns and the trade-offs you get with both; ‘iterative’ and ‘grand overhaul’. The iterative route is often advocated by tech companies – in other words, you’re constantly testing, measuring and prototyping. This is usually a great approach – but the trade off with this is that you can sometimes ‘science the shit’ out of a product! Too many tests mean the approach is imperfect.

It’s usually designers who favour the ‘grand overhaul’. This is the rip the plaster off approached where you simply do big change, ship it and hope it works. The trade off here is that you don’t know the causality of any impact you can – many metrics can change at once.

Which is best? It depends – neither option is right, or wrong. Bo shared a chart tracking ‘Intuition’ at one end, with ‘Data’ at the other. She said that most Series A companies who operate using MVPs work at the intuition end. Series Bs companies tend to fall in the middle where they can use their own user data, twinned with intuition. Big tech companies like Facebook operate at the data end – because of their sheer number of users and differentiation in their circumstances it’s impossible to use intuition.

However, when re-designing the Notes product at Facebook, Bo had to operate at the intuition end, which is maybe not what you’d expect at a big company. Why? There was no need to be iterative as there was no real legacy users for a long-form text based product; Facebook had evolved by this point to be much more focused on visual rather than pure text posts.

Despite a shift to more visual posts, long-form text is not dead.

They simply developed a migration plan where any old notes would immediately convert to the new product when they were next edited (and Bo goes into more detail in her talk about they made sure the migration plan was a success). With low usage, they could basically transform overnight into the new version rather than a longer iterative process (a huge luxury at Facebook).

The new version was much more immersive, flexible and design-led (as well as fulfilling a requirements to be mobile-first, which wasn’t there first time round), and became an alternative publishing outlet for many users versus the previous short text post focus.

As a PM, Bo gets most excited about these nebulous problems where products have to go from zero to 1 overnight, rather than an iterative 1 to 1.5.

Do you really need a redesign?

That’s the big question teams often ask themselves, and Bo encouraged anyone thinking that to dig deep into the why; what issues are you trying to solve with a redesign, and why?

Tumblr fell more on the data/iterative end of the spectrum above, and had the challenge of having a core product that people felt very passionate about; so they knew a redesign would have a huge impact.

They also had the issue that the composer feature on mobile was beholden to old architecture (which sucked) and was generally an archaic experience compared to the web. Tumblr’s mission is all about self-expression, so a redesign was crucial to help its mobile experience fall more in line with that.

Final takeaways

1. Don’t see people as numbers but rather as people

Product people have an ethical duty to see the human problems behind the numbers – it’s not just about employing design thinking, it’s about being a good person. Especially larger tech companies can fall into the trap of thinking they need to ‘science the shit’ out of everything – whereas really the key is to humanise technology.

2. Test your assumptions through user testing

That said, user testing is crucial to making sure you’re not building things on assumption. It doesn’t need to be fancy – Bo often used 2D storyboards with users at Tumblr, which were very effectively for seeing the whole picture of a product experience without having to prototype excessively.

3. Do gradual rollouts

Bo’s final takeaway was around gradually rolling out new features or products to your userbase, rather than having a big ‘ta-dah’ moment to all of your audience. Do incremental rollouts to 1, 2, 5, 10% of your users, then you have the ability to rollback if something’s not working.

You can watch all of Bo's talk on redesign at Canvas 2017 below:

If you didn’t get a chance to catch Bo at Canvas 2017, we have an awesome line-up of speakers for you at this year’s Canvas, held on October 4th at the Birmingham Rep. You’ll get to hear from the likes of Dropbox, Booking.com, Ends and many more.