Sam Selbie from our UX team, explores how to handle design for failure to develop a more helpful and supportive Customer Experience, as measured by our PulseCX benchmarking tool.

Your product will fail, your customers will make mistakes and break things, and your servers will crash. Even Amazon acknowledges this[1].

Your product's ability to handle malfunction and misbehaviour is a crucial, but often overlooked part of your customer experience. For examples of the best, and the worst responses, let's look at error pages, plasticine, and power plants.

ERROR 404: The requested resource cannot be found.

You have seen this message countless times. If you run a website, you need to have one so people know where to go if something breaks. So, consider what you can do to make this a pleasant interaction instead of an abrupt stop in a users’ digital journey.

Users arrive at this digital dead end for any number of reasons: a broken link somewhere, a mistyped search query, a cat walking over a keyboard…Now that they are lost – take the opportunity to help them find their way, promote a product they might be looking for, or even insert a funny one liner. Pixar, the National Trust for Scotland, and Glasgow Film Theatre all have lovely, on-brand 404 pages. Google Chrome even built a game into their browser so you can play when your wifi goes down! (To play it, turn your wifi off and open a new tab.)


A screenshot of the Pixar 404 page featuring the blue character Sadness, crying from the film Inside Out


Treat the inevitable issues as opportunities to engage with your customers and point them in the right direction.  


Things break all the time


You should hunt for these vulnerabilities, and make them as smooth as possible. Take my experience ordering plasticine as an example of what not to do.

I placed an order for some modelling putty from Amazon three weeks ago, but it never arrived. I hadn’t heard anything from Amazon, so I logged on to my account to track the parcel.

At the very top of the page, there was a big message saying that ‘It looks like your order hasn’t arrived, would you like to get a refund?’.

This annoyed me, because they knew that my product hadn’t arrived, but forced me to check on it before telling me what I could do about it. If I had forgotten about the order (which I have done many times in the past!) then I might never have received a refund.

Instead of doing this, follow the example of ATG Tickets. When my tickets for Book of Mormon got changed because of COVID-19, they emailed me to let me know they had moved my booking, but my seats were the same, and that I could cancel or exchange them if needed. It was a much nicer experience, and I didn’t have to do anything.

When things break, try and fix them first. If you can’t, show the user there is a problem, and tell them how to fix it.


Danger, danger. High Voltage


And that leads me nicely onto nuclear power stations.

On the 28th of March, 1979, at 4am the turbine system at Three Mile Island nuclear power plant in Dauphin County, Pennsylvania shut down. As heat started to build in the reactor core of the plant and things began to fail more than 100 alarms went off in the control room. Imagine being the operator, trying to figure out what’s going wrong when there’s a cacophony of bells ringing in your ears and more than 750 lights flashing on a dashboard in front of you.

Overwhelming doesn’t come close.

There’s no telling what might have happened had the system selectively displayed the relevant alarms, or flashed just the most important lights, but I suspect it might have helped.

You can use the same principles on your websites. If a user fills out a form incorrectly, give them clear instructions on what to fix, like Business Gateway does on their contact form:  


A screenshot of a Business Gateway error message


They show the errors in a place where users are most likely to see them, they tell users what to fix, and each error message links a user up to the right field, where they show the message again.


Find your points of failure  


Although neither you, nor your users run nuclear power plants (I hope), you may have a website, possibly that sells products on it.   

In either case, users will behave unexpectedly and products will break. URLs will be misspelled, forms will be filled out incorrectly and parcels may go missing. When they do make sure you think about what happens next. Inform the user that there is an issue, identify the issue, and give them instructions to fix it.    

Looking at points of failure in a customer experience is just one of the ways we identify if your service is Helpful and Supportive as part of our PulseCX evaluation. To use our PulseCX benchmarking tool contact Managing Partner Phillip Lockwood-Holmes on


[1] Amazon’s SLA gives you 99.99% uptime, but that still leaves a .01% allowance where things can go wrong.

After we published this article, approached us to show off their own, very cool 404 page. We liked it so much we wanted to share it with you as well. You can check it out at