Is it good to add some personal information (name, username or other) in success or error messages?
Igor. Content personalisation can be appropriate at times, like in an email or after login. Amongst a few, it makes the system appear more 'human', and can facilitate some personal 'bond' with the user. But one can argue that by attaching a name to a notification you won't achieve that (I see proper personalisation as real user-dependent content). Also, if overdone, you can annoy users, and repeating notifications with the user name in them can lead to exactly that. This is what I am trying to do by repeating your name in this answer, Igor.
You should ask yourself Igor whether:
One must ask Igor, what is the benefit in having someone's name in a notification. Interfaces are often designed to meet business goals - to incite some behaviour from users. Here is a place where it could be justified:
But how does this apply to notifications Igor? What is the goal?
You can argue that the use of personal names in systems follows the way we speak to people on a daily basis. I urge you to consider how we use people names in conversations.
Apart from in introductions and goodbyes, we typically say someone's name as a mean of emphasis or urgency:
Which one comes across as stronger?
You could argue, Igor, that in some cases saying someone's name is somewhat patronising.
Racheet has already covered this brilliantly. If you are to include names in notifications, for consistency sake you'll have to also include them in error messages:
A message as such can really annoy users (ditto for humor, but that's a different story).
Now Igor, I think the main priority with notifications is to deliver the information in the quickest, clearest, and most-concise way possible. A plain language expert would ask you to consider these options:
As you can see, Igor, all these options convey the same thing. Some nice experiment you can do is to toggle your eyes between the top and bottom options - it should take you around 3 seconds to get the top one (and your brain would need to work a bit, scanning and making sense and building models and stuff); but less than a second to get the bottom option (a really brief look would suffice and it feels as if you haven't read anything at all).
The middle option should take slightly longer than the bottom one with some brain work required. To really dissect, I should add that by introducing "John," you will be reducing the flow of the sentence. This is something commas do and style guides point that out. Dostoevsky was known for not complying with this, with stop-and-go sentences along these lines being customary:
"I was, by all means, ready, although not in the same way I was in my youth, to challenge this peasant, this dirty young man, which I met at the market."
Having said that, you can argue that 'Settings saved' is somewhat cold and mechanical. 'Your settings have been saved' is more personal than 'Settings have been saved' (because of 'Your' - See this blog on the power of 'you').
Depending on the user persona and the frequency of notifications, I would go with either the 'Your settings have been saved' version or the 'Settings saved' one.
"I'm sorry Dave, I'm afraid I can't do that."
That scene from 2001: A space odyssey is a good example for why this can be a dangerous practice.
Beware of anthropomorphising a computer to the point where the user starts ascribing malice to it. Error messages need to be non-personal to avoid the user feeling like the computer is complaining at them, or worse, actively trying to hinder them.
There are two big problems, from an internationalization perspective:
How sure are you that your Name data contains the name the user is called by? Getting your name data format correct is a classically difficult problem.
As soon as you add name wildcards to your error messages, they will become much more difficult to translate to other languages. Without knowing what the name is, the translator may have difficulty finding the correct level of formality, putting the wildcard in the right place in the string, or even using the right gender for verbs. Without context, it's much easier to translate "Your settings have been saved" than it is to translate "*, your settings have been saved."
The most relevant research I could find on this topic is a little dated. Fundamentally, adding user names to error messages deals with humanizing an interface. In a study of using human faces as part of computer interface  (admittedly, a step well beyond just including user names), the researchers found that increasing the humanization of an interface does not necessarily make users like it more. Another study comparing machine-like versus human-like computer feedback found a similar result, where error messages like "Incorrect date format - expected MM/DD/YYYY" were compared to "Sorry, [User], I don't understand that date format - I need it in MM/DD/YYYY" :
Human-like error feedback significantly increased the generation of negative cognitive responses by persons with high-self esteem but significantly decreased the negative cognitive responding of persons with low self-esteem...Subjects also rated the neutral feedback as being the most helpful.
Our data suggest that the terms "user oriented" and "user friendly" should not necessarily mean that the software contains human-like feedback messages for the user.
This is an interesting conclusion, suggesting that humanized system messages are better for users with low self esteem (or, perhaps, confidence in using the system).
Ultimately, I think you can take a look at the industry today and see that using personal information in system dialogues is not typical. Other answers here touch on some other possible reasons for this - users ascribing agency (and malice) to the system, etc.
: Lee Sproull, Mani Subramani, Sara Kiesler, Janet H. Walker, and Keith Waters. 1996. When the interface is a face. Hum.-Comput. Interact. 11, 2 (June 1996), 97-124. DOI=10.1207/s15327051hci1102_1 http://dx.doi.org/10.1207/s15327051hci1102_1
: Resnick, P. V. and Lammers, H. B. (1985). The Influence of Self-Esteem on Cognitive Responses to Machine-Like versus Human-Like Computer Feedback. The Journal of Social Psychology, 125(6):761-769.
Although 2001 SO is one of my favourite movie, here are a few things to keep in mind:
Users don't like to read, the shorter the better. Kudos if it smarts and eventually witty.
As a user, I would find quite redundant (and annoying) to have to read my name every 2 minutes. I'm already engaged in this 1 to 1 interaction with the interface.
In a conversation, we don't -or very rarely- repeat the other person's name when answering a question or making a statement.
Although being more personal is a great idea, it often fails when we have to dive into details: masculine and feminine first names, i18n, etc.
I think there is a place for this - but it's not about calling users by name. That reminds me of middle-school programming assignments, which have a habit of starting with asking a user to type their name so the program can echo it back in messages.
The more useful application of message personalization is to echo what has actually been done in a confirmation message, or what's needed in an error message.
Your password has been updated is clear and meaningful vs.
Your setting have been updated.
Your form contains an error isn't very helpful, but
Please enter your zip code is more clear and direct.
You have a new comment tells you more that
You have messages, etc. I may be mixing the idea of message personalization with customization, but in the end personalization is a subset of customization anyway.
Often we're used to copy and past a code error in Google (or in other help form) in order to find further details.
So, if there is a personal message some of this research could be empty.
Fot this, I think is not a good idea, and I don't see any important advantage.
This depends on the personality of your brand (site that is). MailChimp, OKCupid, etc. have done these things in the past where the error messages create an experience in there own. Example: Instead of "ERROR: Page unable to load" you might see "One of the Monkeys got lose, we will be back online as soon as we get him in his cage". It's the personality of the brand or site.
Meaning if this is an error message for a medical cliic I would say keep it simple and to the point. However, if this was an error message to an action movie lovers site you could make it fun and in tune with the theme of the site.
In the end will John not know it was his saving that were saved if you do not place his name there? Doubt it. If it was a multi player game and John's were saves while others were not, then yes. It just depends if you want to put in a litle dev time and have another dynamic piece to a site to deal with.
There are both practical and branding considerations here. One the branding side, you have to take into account the voice and tone of your website. Is it in line with your brand voice to address the user by their first name? Has this practice been consistently deployed throughout the site?
On the practical side, I would echo the comments made above about internationalization and add some other practical concerns. Just because you have captured a person's name doesn't mean you know the name the person likes as a form of address. Have you included a form field for "name you prefer to be addressed as" in order to have this information available for use? Different cultures may employ different forms of address, including using the last name as a familiar form of address and rules for when a familiar form can be used.
All in all, I would shy away from using personalization in error messages, but that is a judgment call.
I'll keep this short. A few reasons not to personalize error or confirmation messages are:
People scan information, I think less is more.. so in this example do not include the person's name. Add a tick mark instead :-)
Personalizing a machine - and by that I don't mean personalized search results or the internet bubble - is the most difficult obstacle we are about to face in the next few years.
Simple example? Robots are robots as long as they act like robots. If they start trying to act like humans but don't respond like humans (or the way a human would expect them to respond in a natural way), they get creepy. As in wrong facial expressions, undertone, responding to messages "between the lines" etc.
We have a million ways of interacting with each other. Trying to let a machine respond in the most natural way is a major task. Chances are high that machines will address natural human requirements in an unpleasant way.
To answer your question: It depends. It depends on your domain. If your talking to engineers, professionals etc. who talk tech, respond like a machine. "Done", "Saved", "Error. Please go back and..." etc. Be impersonal.
If you want to address etsy.com, deviantart.com or any other non-technic-affine users I would slowly try to introduce the one or other personal communication. I would keep a list of all those interfaces, where users are displayed with messages like these so you can grasp over it from time to time and have a feeling with which tone you are talking to your customers.
Very interesting question!
This would lead to Anthropomorphism, which is a bad idea (). Besides, that would creep many users out (*).
Users want computers to do as they ask, not to socialize with in an attempt to convince them to do as they want. When you address someone in a first name basis, you are communicating on a personal level, something which users don't expect computers to do.
(*) If your computer refers to you on a first name basis, wouldn't you feel that you have to reply in the same manner in return.
(**) Imagine if you used your face to log in and your computer commented the your hair looks good today.