How do we get to amazing?

Here’s what everyone agrees on at Metro. We all want our website to be the best it possibly can be. We get millions of visitors a day and we want them to have a truly great experience. We want to get even more millions of visitors a day.

So how do we do that? There are two schools of thought.

The perfectionist school

Make everything we do as amazing as possible. Sounds good, right? There are a couple of catches though.

1. You spend most of your time optimizing the wrong things

For example, we spent months perfecting an Android app that no-one downloaded. We were optimizing the features rather than optimizing how to drive downloads. Totally ineffective.

It’s a similar story for “Metrovision”, timeline’s realtime updates, the Metro TV product, penalty shootout game and countless others. All ditched after months of hard work. Fail.

Similarly, we’re currently optimizing our new trending page even though we don’t know yet if anyone is going to visit it. (See later in this post for an alternative approach).

With each of these features we tried to make them perfect from the start. We ASSUMED this would make the feature a success, but that’s a self-centric approach.

We want to be USER-CENTRIC, not self-centric. That means getting feedback from users as quickly as possible. Do they care about the new feature or not?

Each month spent on a feature that flops, is a month we could have spent on something that really matters. By trying to make every feature perfect, you’re making it way slower to discover the features that really matter.

2. Over-complicating things…

You end up building things that are overly-complicated which means higher maintenance. Higher maintenance makes it harder to build amazing features.

One of Metro’s most impressive design achievements was making it possible to swipe between articles on the website. It made the website feel like a native app. Amazing! Of course, this wasn’t simple and it made the site higher maintenance. In fact, every feature we added took about four times longer to develop because of swipe.

In fact, despite being amazing, users didn’t care about it. Eventually we ditched it – the benefit of swipe wasn’t enough to justify the high maintenance.

By trying to make every feature amazing from the start, you’re actually making it harder to make the site as a whole amazing!

Let’s look at the alternative…

The “start simple” school

Don’t start with perfection, start with the simplest improvement you can think of. Get feedback from real users. Iterate.

How would that apply to our trending page?

We could build a simple trending page (like the BBC one, for example). It’s not perfect but it’s an improvement – it’s better than no trending page!

We could then promote the page (e.g. in the masthead) for 1% of our users. That’s 20,000 users a day – a big enough sample to get valuable feedback. If the feature isn’t popular with some people then at least only 1% of users have been exposed to it.

If not many users visit the page then we work on how we promote it (e.g. breakers). When (and if) that is successful then we can improve the trending page itself. We can measure the impact of each change we make to it.

In short, we can be rigorous and transparent about how we grow our product. We can move fast and ensure we’re being effective.

Don’t all these “imperfect” changes confuse users?

Every time you do a Google search, you are part of about 20 experiments. Does Google search confuse you? If so then it doesn’t seem to have hurt their product growth.

Does anyone else do “start simple”?

Apple started simple with the iPhone – iPhone 1 didn’t even support third party apps! Google then started simple and iterated faster to catch and overtake Apple. Android is now 80% of the smartphone market (detail).

In fact, we wouldn’t have smartphones at all without “start simple”. The technology underpinning smartphones is a product of science and a basic principle of science is… you guessed it, “start simple and get feedback” (aka experiments).


The perfectionist school is good for some problems but not complex ones, such as “how do we get Metro to overtake it’s rivals”.

MORE: What is so special about dev team “focuses”?

Leave a Reply

%d bloggers like this: