Upvoted because I think this is a good point, but the statement "nice error pages matter" isn't backed up with any evidence. I would have loved it if there had been statistics showing how people reacted to different error messages, even though I know that kind of information might be difficult to gather.
I would be interested in this as well - but IMHO the effort involved in putting up something other than plaintext error (like Borders) is so trivial that it's practically a crime to not do it.
I think it's hard to argue that there isn't some gain to be had with a proper error page - and implementation is just so darned simple.
Quite - and there's lots of literature to say that negative interactions are the ones that have the most impact on customers. Smart and responsive branding at failure points strikes me as one of the most valuable investments you can make. Of course I'll be annoyed if a page doesn't load, but if you can put a smile on my face, reassure me about the temporary nature of the problem, and offer me a useful path back in, I'm far more likely to follow through rather than go elsewhere.
It's quite astonishing that a large company like Borders would be unaware of this and fail to communicate it to their web staff.
Agreed. I'd also argue that the "cuteness" of such an error page actually can have a major impact on how much the visitor will forgive you. Twitter's Fail Whale is the perfect example :)
Anything is better than Borders' page (except possibly a blank one)... even a rather cold, drab, boring error page that at least describes the error (we are over capacity, sod off) would serve better. I don't think we really need to do any extensive focus group testing to agree on this much...
Amazon has repeatedly said that page load times matter. They decreased them once by 0.1s, and revenue dropped 1%. Google is also fanatical about speed (ever noticed how much GMail caches things?). Linus Torvalds said that because Git is so fast at what it does, it changes the way you use SCM.
