Tonight, I reluctantly return to a topic that was part of my everyday existence for at least 5 years: Twitter. Earlier today, the Verge and New York magazine published their collaboration about the timeline and inner workings of the recent Twitter takeover. And while I think it’s a valuable use of your time even if you’re not a former Tweep, I want to focus on one paragraph in particular toward the end of the piece:
Four days later, Twitter crashed. More than 10,000 users, many of them international, submitted reports of problems accessing the site. Some got an error message reading, “Something went wrong, but don’t fret — it’s not your fault.”
See that error message there? That last sentence there with the em dash? The one on the quotes? I wrote that! And, as I mentioned in a Slack channel earlier today, if that’s my Twitter legacy, I’m fine with that. But I want to give you a little history about how that particular error message came about. It’s a combination of content strategy and #LoveWhereYouWork ethos, which is a great example of the zenith of my time at Twitter.
In 2018, our batch of interns was stellar. We consistently had great groups come through, but that year, there was something special in the air. And I’m not just talking about the ones assigned to the Design and Research team. Throughout the org, every intern I worked with not only made the platform better, they taught me a ton, too. It was a pretty great summer.
Each intern had a mentor assigned, and specific tasks they were assigned to work on. But they also got to pitch and complete a project on their own. One of that year’s interns, Vanessa, wanted to reimagine the design of our error messages. Visually, they had changed very little from the days of the infamous Fail Whale. And that illustration didn’t even come from a Tweep. So her plan was to update our errors, aligning them with our then-current visual brand guidelines, which gave me an opportunity to update the kinds of messages which were available for our engineers to use.
That’s where this post turns from watching the current Twitter car crash (all layers of innuendo intended) and more towards an error message how-to, at least in regards to how we did it on my team back in 2018. First, we asked ourselves what scenarios could cause an error to pop up for people. From what I can remember, we came up with four broad scenarios we needed to account for:
User error- This should be self explanatory, but it’s basically something like a typo, usually in a search term or URL.
Twitter error- Sometimes, things break. And when we knew it was our fault, we needed to tell them that.
Connectivity error- Not all of Twitter’s features were available offline, so we had to account for moments when we needed more bandwidth.
Mystery error- And then there were the unexplained gremlins, which may later have an explanation, but in the moment, this was the equivalent of a shruggie.
After we had mapped out all the instances which needed messages, then we had to decide what to tell people. And whether or not we needed them to take an action. So all four of the types of messages we mapped out had a version with a call-to-action (CTA) and a version without. If we knew a refresh might fix things, for instance, we’d include a button reading “Try again”.
Lastly, we made sure that each of the messages, whether they had CTAs or not, were using our updated brand voice. But we had to make sure that our tone was appropriate for the emotional state of people when they saw it. Remembering that nobody really wants to see an error message — no matter how witty it is — was a key component in crafting a clear message rather than a clever one. We did this by developing a template so that we could balance information with empathy, and roll them out in all 42 languages we were using at the time.
If we look at that example from earlier, “Something went wrong, but don’t fret — it’s not your fault,” I can tell you that was one for the Twitter error scenario. And it was probably followed by the “Try again” CTA. We reused that “Something went wrong…” piece for all of them, I think, and then tailored the ending depending on which error type it was and what we wanted people to do next.
Honestly, it’s been a while since writing those, and I don’t remember them all. I wish I had a list of all of them, but I’m sure if you’re still using Twitter, you’ll most likely see them pop up more and more the longer the current regime retains control. While I’m glad these posts aren’t revisiting the Twitter drama as often as when I first started writing them, I was glad for this trip down memory lane today. And happy to revisit some of the content design problem-solving that consumed most of my waking hours back then. I miss that work a lot. But I’m glad to have my life back, too.
Rusty Cage
17 January 2023
Whale watching.
Tonight, I reluctantly return to a topic that was part of my everyday existence for at least 5 years: Twitter. Earlier today, the Verge and New York magazine published their collaboration about the timeline and inner workings of the recent Twitter takeover. And while I think it’s a valuable use of your time even if you’re not a former Tweep, I want to focus on one paragraph in particular toward the end of the piece:
See that error message there? That last sentence there with the em dash? The one on the quotes? I wrote that! And, as I mentioned in a Slack channel earlier today, if that’s my Twitter legacy, I’m fine with that. But I want to give you a little history about how that particular error message came about. It’s a combination of content strategy and #LoveWhereYouWork ethos, which is a great example of the zenith of my time at Twitter.
In 2018, our batch of interns was stellar. We consistently had great groups come through, but that year, there was something special in the air. And I’m not just talking about the ones assigned to the Design and Research team. Throughout the org, every intern I worked with not only made the platform better, they taught me a ton, too. It was a pretty great summer.
Each intern had a mentor assigned, and specific tasks they were assigned to work on. But they also got to pitch and complete a project on their own. One of that year’s interns, Vanessa, wanted to reimagine the design of our error messages. Visually, they had changed very little from the days of the infamous Fail Whale. And that illustration didn’t even come from a Tweep. So her plan was to update our errors, aligning them with our then-current visual brand guidelines, which gave me an opportunity to update the kinds of messages which were available for our engineers to use.
That’s where this post turns from watching the current Twitter car crash (all layers of innuendo intended) and more towards an error message how-to, at least in regards to how we did it on my team back in 2018. First, we asked ourselves what scenarios could cause an error to pop up for people. From what I can remember, we came up with four broad scenarios we needed to account for:
User error- This should be self explanatory, but it’s basically something like a typo, usually in a search term or URL.
Twitter error- Sometimes, things break. And when we knew it was our fault, we needed to tell them that.
Connectivity error- Not all of Twitter’s features were available offline, so we had to account for moments when we needed more bandwidth.
Mystery error- And then there were the unexplained gremlins, which may later have an explanation, but in the moment, this was the equivalent of a shruggie.
After we had mapped out all the instances which needed messages, then we had to decide what to tell people. And whether or not we needed them to take an action. So all four of the types of messages we mapped out had a version with a call-to-action (CTA) and a version without. If we knew a refresh might fix things, for instance, we’d include a button reading “Try again”.
Lastly, we made sure that each of the messages, whether they had CTAs or not, were using our updated brand voice. But we had to make sure that our tone was appropriate for the emotional state of people when they saw it. Remembering that nobody really wants to see an error message — no matter how witty it is — was a key component in crafting a clear message rather than a clever one. We did this by developing a template so that we could balance information with empathy, and roll them out in all 42 languages we were using at the time.
If we look at that example from earlier, “Something went wrong, but don’t fret — it’s not your fault,” I can tell you that was one for the Twitter error scenario. And it was probably followed by the “Try again” CTA. We reused that “Something went wrong…” piece for all of them, I think, and then tailored the ending depending on which error type it was and what we wanted people to do next.
Honestly, it’s been a while since writing those, and I don’t remember them all. I wish I had a list of all of them, but I’m sure if you’re still using Twitter, you’ll most likely see them pop up more and more the longer the current regime retains control. While I’m glad these posts aren’t revisiting the Twitter drama as often as when I first started writing them, I was glad for this trip down memory lane today. And happy to revisit some of the content design problem-solving that consumed most of my waking hours back then. I miss that work a lot. But I’m glad to have my life back, too.
See you tomorrow?