Should a button communicate the current state, the intended behavior, or both?

Ben Halpern on June 17, 2019

When I magnify a window in Chrome, I see an icon like this: When I reduce the size of a window below it's normal 100% I get this: Clicking ei... [Read Full]
markdown guide

In my experience, the button should convey the action. Although the button says Following, that does not indicate what it should do. If we take a page from Twitter's UX, hovering over the button when it says Following, changes the text to Unfollow, the intended action. So I think it's OK to show the current state potentially, but in the end before it's pressed, it should convey the actual action about to happen.

Demo of Twitter Unfollow button

Update... Just adding this comment here as I forgot to mention mobile

Good point about touch/mobile with hover. I should have took a bit more time before answering. 🙃

Looks like Twitter on mobile just shows Following and when you click it, you get the same prompt as desktop prompting you to make sure you really want to unfollow. I agree having just unfollow text would have made sense here like you suggest.


What about touch devices? They don't have hover effects.


This is a huge part of mobile UI that I think a lot of people overlook if they're more focused on desktop. I think you should always design as though hover is not an option, only using it to add extra clarity, but not relying on it to convey the meaning of a button, etc.


I answered it here.

Good point about touch/mobile with hover. I should have took a bit more time before answering. 🙃

Looks like Twitter on mobile just shows Following and when you click it, you get the same prompt as desktop prompting you to make sure you really want to unfollow. I agree having just unfollow text would have made sense here like you suggest.


What is interesting is that with Twitter, even as something as non-destructive as unfollowing someone, they prompt to confirm so even if you didn't understand the meaning of the button, you've got a second chance with a more complete description.


The unfollow on Twitter can be pretty destructive, in that if you unfollow somewhat who is private, you have to request permission to follow again (so it's very clear to that person that you unfollowed). I'm a fan of any action that can be difficult to redo easily (or may have unintended consequences) getting a two step confirmation.

Well, at least they prompt you before you unfollow.


That's no surprise, as Twitter focuses on maximizing engagement from its users.
They really don't want you to unfollow anyone. 🙈
They now even prompt you what the handles you follow like and follow themselves... Quite annoying. 🙄


Here's an example of the worst possible combination of things, courtesy of Quora:

Screenshot of the follow and notification icons from Quora

It says "Follow", but that's not an action, because the tiny tick indicates I'm already following. I have no idea what happens if I click on "All notifs" and don't want to try.


I thought exactly the same example about Twitter. I think it is always good to convey the action. When I see a "Following" button I somewhat feel that I spend a split of second to understand that it is a button and it may be used to unfollow someone. On Twitter if you hover over it you have no doubt about it. However, this is not true for the app which makes me think what we can do to convey the action and state at the same time on mobile apps.


I expect a button to tell me what will happen if I click. For example, as I write this, I have two choices: Preview or Submit. Both of those are very clear as to what the action will do. (Although clicking preview changes the button to markdown which I actually find confusing. I expected something more like edit. )

To be honest, I didn't know that Following was a button that would unfollow. 😳 There's nothing about it that indicates to me visually that I can click on it. The + Follow is more clear, mainly because of the +.

The unsave hover over I find really confusing. What does that mean? I don't know how you unsave something, unless it means it's going to be deleted.


This is an interesting topic. I had an argument with a designer regarding this sometime ago. Personally I think if its a simple toggle which turns something on/off indicated by an icon, I would prefer showing the current state.
If its something which requires a call to action and includes some text (which you could/couldn't cancel in the future) , I prefer showing the intended behavior


Based on how confusing I find my iPhone mute toggle button, I am leaning toward the idea that buttons that have no text (are they always toggles?) should be different from text-based buttons. Icon toggle buttons seem to make more sense to me when they are, as you say, showing the current state. Then again, play/pause, if it’s one button that toggles, would make more sense to me in the normal way, where the play button means I want to play, not that it’s already playing.


I remember seeing something similar about toggle states on the UX Stack Exchange site. That question/answer talk about on/off options and indicating state etc.

While it is super convenient in one way to have the button be dual purposed, it is probably better to give more distinct actions and have them be more discoverable.

It might take up more space but maybe more:

Saved. Undo?

("Saved." would be text, "Undo" would be a link)

That way you give both the current state and the desired action.

With the following button, I do think you have more flexibility with it simply because of the ubiquity of that type of button across other systems (Twitter etc).


A toggle should show a current state. A button should show an action.

When the button shows a current state, it stops being a button, and becomes a sort of a tag. Whether alternating between a tag and a button on hover is good user experience - is quite subjective. Users might not realize that a tag could be actually clickable or would become a button on hover. Moreover, it is even less intuitive on mobile, because there is no such thing as hovering.

Most of the time the actions can also convey the current state. For example, Unsubscribe button would imply that the current state is Subscribed. I don't personally think that the users really need an extra confirmation describing the state in that part of the UI. Perhaps another UI is needed to list all user subscriptions and allow them to manage them all in one place, maybe something like a "SUBSCRIPTIONS" section on YouTube.


What we have here is a blend of different paradigms: button and checkbox. In the cases you mentioned here, the button is actually not representing a singular action, but instead a dual-state or in case of the zoom a multi-state switch.

The ideal solution would be a component that conveys both its present state and its future state in an understandable way. In case of the follow-button, this would probably be something like a Checkbox button



The checkbox conveys that this element has a state that could be toggled rather than singularly selected.


There's a tension between screen real-estate and clarity of function. The "state button" takes up less space than the switch or checkbox, but it can be unclear what the current state is, because of the blending of paradigms that you mention.


The button should describe the action to be performed. Use a separate label or some other indicator to show the current state.

As a demonstration, most video conferencing apps confuse me with their "mute" button. I'll see a full mic and I'm never quite sure if it means "my mic is currently on" or "click me to turn your mic on". Since they almost always act as both the action and the indication of state, I'm left always worried that I misread the UI and am actually unmuted when I think I'm muted.

Another thought: if a button is intended to convey both state and action, it should be designed as a switch. A switch cognitively makes it clear whether it's on or not, and also indicates an action can be taken. Here, I would expect the switch's text label to indicate the current state, with the switch's own appearance, together with the label for the current state, implying what will happen when I click the switch.


I don't think the label for switches should change between states. That is always something that confuses me:
[ON] Do something
[OFF] Don't do something
When seeing the first it's clear: right now "do something" is on, so it'll be done. But the second isn't: "don't do something" is off - does that mean it will now be done? Or not?
If the text were to stay constant, I can switch "Do something" on or off, which makes it really clear in my opinion.

Also, always use the positive clause to avoid double negations.
Do: [y/n] do something
Don't: [y/n] don't do something.
The second could lead to "not don't do something" when read out, which doesn't make much sense. "Yes, do something" and "Not do something" make more sense. :)


I would say that I am on team "action". I think a button should indicate what will happen if you click it. For the "Following" button, you could rename it "Unfollow".

I think the hover behaviour @nickytonline mentioned is okay for desktop, but that wouldn't really work for a touch interface. I'd suggest the default should be "Follow" and "Unfollow". Then you could optionally add the fancier "Following" with "Unfollow" when hovering over the button for desktop devices.

One tricky case I've run across though is a toggle for muting. On my iPhone, it shows a speaker with a slash through it when the sound is not muted, and a normal speaker when the sound is muted. That's consistent with what I've said above, but it's actually very confusing to me in practice. It seems to make more intuitive sense to me that the speaker with the slash indicates that the sound is currently off.


Good point about touch/mobile with hover. I should have took a bit more time before answering. 🙃

Looks like Twitter on mobile just shows Following and when you click it, you get the same prompt as desktop prompting you to make sure you really want to unfollow. I agree having just unfollow text would have made sense here like you suggest.

  • Hover states should be optional as they don't always exist (mob/kiosk/disabled/TV etc). Active state.
  • Combination of icon & text. Icon should "be more substantial" once activated. Think filled hearts vs outline. Finger presssable (44px+ says HIG).
  • It's not save, it's "bookmark". You are not saving anything. There is no organizational element. Icon on the left shows bookmark (that's "tag" really, but looks like a bookmark-ish). A variation of that works better.
  • Use correct casing. trendy lower case & EASY DESIGN ALL UPPPER reduce readability. Harder to pull off.
  • Avoid 'hurdling' (distractions that have to be seen 1000x times but rarely used - save btn distracting from titles & activity level). The instinct is "more saves because it stands out" but that is not aways correct. If the content is good, people will find the way to "save" it.

a.. Now:

  1. Your btn says "Follow'
  2. I press, it says "Followed" (or following). Supplement it with the appropriate symbol (or color). Behavior is implied. Done.

b. Otherwise:

  1. Your btn says "Follow?'
  2. I press, it shows my next action. "Unfollow".

In this case, I am subliminally being told to unfollow you so your current choice seems better.

Like, Unicorn & Save/Bookmark are pretty much 1-2-3 rating system where all three btns do almost the same thing (they could be merged, with a built in bookmark aspect). Then you can have other features.

Follow/unfollow on your profile listings, along with descriptions would be useful (so you can follow or unfollow without having to check each profile).


Something like this seems like a correct behavior to me. Straightforward and simple, though it may seem a bit flashy at some point.


This is the way to go, you may want to add the "result" in the button.
When clicking "+ Follow", jump to the "✓ Following" state.
If you hover again display: "x Unfollow"

This Implementation does not consider Mobile UX, in this case you may want to prompt the user for an action.


Don't. Make. Me. Think!

By this I mean, if it's a "button", always make it clear what pressing it will do.

Find another way to convey state if you can't make the button do both things easily.

Also, if pressing the button is trigging an action that might take time, make it obvious I pressed it and they I'm now waiting.

And if you don't want me to press it again while I'm waiting, disable it.

While you are at it, remember to enable it again if the action was successful and can now be done again or if it failed/times out and can be retried.

Where possible, don't make the button change after I click it to be doing another thing. Find room for another button.

And finally, any actions I might want to preform a lot, let me do them against multiple instances of the item, using some kind of multi-select. Bonus points for a "select all".

One obvious place for this, I would like to be able to find all the people who I can "follow back", and do they all in one go! Not one at a time.

Thanks for listening. As always, my opinions are my own. I am open to updating then if presented with a solid argument.


The place this confuses me the most is with twisties (the little arrows on accordion items or menu expanders). The can either show the current state or the action to transform the state and there's no consistency between providers.

I like the way Apple did it back in the day where they indicated the collapsed (vertical) state of an item with a horizontal arrow. i.e.

closed: ⇦
open: ⇩

Especially with a transitional animation, this helps alleviate some of the confusion. I think Apple have gone to the dark side on this since though. Switching between styles on a Mac and on a Google leads me to not trust any control I see and click everything at least once to discover what it does.

It's like when you search a form to see whether there's a save button or whether things are just going to update by magic. Sites and apps are so inconsistent in this that nobody is ever really sure if their changes are going to stick, and you end up closing and reopening the settings pages just to make sure.

Where I'm going with this is that I'm not sure at this point that there is a correct answer that's compatible with keeping terse buttons or pictographic icons, because the well has been thoroughly poisoned.


I'd lean towards showing current state for consistency not just across this app but across most apps 'cos that's how my minds expects the UI to be across apps these days.

It could be Save <=> Saved instead of Save <=> Unsave


Jakob's Law indicates users expect and prefer a similar pattern across the internet.

For example:
1 billion people on instagram are used to seeing this functionality with that saved icon, and possibly majority of us experienced instagram UI. We can make it easier for users to not guess the UI of same functionality and make it quick for them to start.



Totally just my personal opinion, but I think a button should indicate the intended behavior.

The Following example doesn't completely offend me, only cause I know it was Follow 'before' I clicked it (however long ago it was).

The Save example here doesn't bother me a ton, mostly because of the hover state + the know the Save button used to be there so it makes sense thats where un-save lives too

I think I would change both to their intended action, and maybe use something else to display the state? :shrug:


I second this. In most cases, it seems more clear to show the intended action rather than the current state. My thought on this falls back to what feels natural to the average user. For instance, as I'm typing this on my phone, I see a "Submit" button. I know that clicking it will submit this post. But what if it showed "Not Submitted" instead? Would I know to click it to send it? Probably not. At least not right away. But it's a matter of context as with most things and highly subjective as well. I think I err more on the side of "what buttons are considered the most important/most used, and how do I make it most obvious what their function is? " Keeping them simple and easy to understand, I believe most people will pick up the more intricate buttons through experimentation on their own. Ya know, give them enough to understand basic functionality, but not so much that they lose curiosity. DEV seems to balance that line very well so far.


I definitely think buttons should indicate expected behavior, not state. I think we can apply the single-responsibility principle here! While it is usually not used for UI/UX, it was invented to make code easier to understand for humans. So I think it works just as well for UI/UX.
Therefore I think buttons should say "Save"/"Unsave", "Follow"/"Unfollow",... It is possible to add another item to the UI indicating state: "You are following USER123"/"You are not following USER123." Be careful though to make these really clear.

I think of buttons as function invocations, or requests for action. Something else can then tell me the result.

There might be good exceptions though - as there always are.


Oh, also: I don't think hover effects are enough. If someone is using a touchscreen they don't work, and to others it might not be obvious. When using a keyboard instead of mouse they also don't work. Rather have two different UI elements.


I think the button should convey what action will be performed rather than the current state. It's a common pattern I see when looking at websites that are in 2 languages in Canada (french + english).

For example on Kijiji (It's like Ebay), if you're navigating the site in English, you'll see "FR" displayed in the navbar to switch to French. If navigating in French, you'll see "EN" displayed in the navbar to switch to English.


Man, that's confusing like hell... 😳


I have to agree for this case. I think "switch to FR" would be much better.
The way shown basically shows the opposite state - neither an action, nor the current state... 😅

It's quite common on web sites, though. Sometimes you'll see "English" or "Français" instead of just the first two letters. Sites that support more than two languages will usually have a dropdown to pick your preferred language, usually showing the currently selected language. Can't say I've ever had any problem with either approach.


We have this conversation over and over at work and never come to a consensus >_>. My opinion mirrors the physical world. We’re all used to controls telling us what they do. If they need to show state, there’s usually a separate indicator of some sort. Given that we all use physical controls all the time, it makes sense to try and replicate that behavior. That’s why buttons exist in UI to begin with right?

More important than that though, make sure whatever decision you make you write it down in your style guide so you can stop having the same discussion.


I prefer the button tell me the exact action that will occur when I click it. I didn't realise that clicking the "Following" button actual unfollows someone :) I rather would prefer a text label Following followed by a button stating Unfollow.


As others have said, the "problem" is that the buttons are doing double-duty. Once you're putting two things in one place, I'm not sure there's a "right" way to juggle them.

(It's a bit like mutable data structures -- when space is at a premium, the place gets prioritized over the stuff that's in it. But if you have to, you make do.)

I think this is one place where skeuomorphism make sense. If something behaves like a toggle, making it look like a switch helps get the point across.

Short of that, IMO ideally the status and the actions have separate homes. Short of that, having a hover state seems better than not having one, but it still excludes mobile and screen reader users (I think).


My thought is that they should follow their real-life counterparts. A button should indicate what actions you can take now. Like below here, I can click "preview" or "submit". If I can follow you, there should be a button I can click that says "follow".


I feel that the button should state what it will do. Having button text change on hover/focus seems like an accessibility nightmare. It could also cause confusion for faster mouse users.

I would like things to clearly communicate their intentions, color changes and stuff to achieve this are great.


well both, there could be 2 ways to get each value such as:
When the button is clicked do this...
Button.getState(); //JS oriented, but can be changed for the language
which can always use a loop to check the state of a button...


Hi Ben... GREAT question, and I've actually learned a ton from reading through the responses.

At the end of it all, to me I go back to my machinery/engineering roots:
A "button" is a "do" thing, It's an action, as mentioned, but at the core, it represents a verb.
The text on a button fills in the blank of this statement : "When I press here, it will do _____"
Follow, Save, Accept, Submit... they're all verbs.
When I reach for a button, my expectation is that it will "do" the verb that it says on it.

"Following" is not a verb (action), it's a state/condition.
( "All Notifs" in one example... not a verb (I wouldn't have clicked on it either ;) !)

When a button text gives me state, it's not very clear or distinct..
It requires more mental overhead, in that I have to think about different states, and translate to action.

When a button gives me a verb, it just seems more simple, natural, intuitive.

Be safe, be kind... wash your hands :)


This dilemma is common due to trying to do too many things with one element (communicate a state AND describe an action).

The proper solution is to have text that comminates the state (like a checkmark & the word Following) in addition to a button that says Unfollow.


This is my personal opinion, A button should only communicate the action. In this case, the button should only say unfollow or unsave. maybe we can use some other icons to display that you are following that user or saved the post. That would be less ambiguous for the user.


I would say, if it is some kind of toggle button it should tell the user the current state. For instance, "Following".
If it is a button to perform an action (which is not supposed to toggle some state), it should tell the action with a verb


I think it depends on how it's framed. The follow button, for example, functions more as a checkbox than a button. That is, it has two states, and no real side effects that can be immediately determined (passive state). The save button is similar, but there's an immediate action or change in state associated with it (active state). As for the magnifying glass, it's not really a button. It's an indicator of state, but clicking it doesn't really trigger any action or state change other than showing a UI pane that has buttons for triggering state changes.


Too many buttons are trying to act like a push-on-push-off switch.

Imagine your game controller buttons being replaced with switches... that would be tedious and no fun.

Therefore, a button should be labeled for it's action, not it's state. Only checkboxes or switches should be labeled for their current state.


A hover or hold shouldn't change the action: an electrified fence should always shock you even if you take your time to touch it.


It can be both depending on the priority of the context and whether words, icons or styles can clearly differentiate the actions. Also, for critical actions there should be confirmations, especially if its only icons


You are confusing two closely related UX elements - A button and a toggle button.

A button doesn't have a state.

It merely provides a way to perform an action - just like a link, or the physical buttons on your mouse - It's there to be clicked and therefor should convey the action it will perform - Just like the "WRITE A POST" button on the top right corner of this page (BTW, I'm not a native English speaker, but shouldn't that be "Write a post"?)

A toggle button, on the other hand, is a different kind of button. It's like the on/off switch and has two distinct modes - but in my humble opinion - it should still convey the action rather then the state.

Let's look at the follow toggle button: Its state is either "following" or "not following" - and I imagine everyone can understand what happens when you click a button entitled "Follow" or "Unfollow", much more intuitive than "following" and "not following" (What the hell should "not following" mean on a button, anyway?)


In my opinion - If you want to convey state, neither a button nor a toggle button is the way to do it - they should both convey action.

current state should be conveyed by other UX elements - texts, checkboxes and radioboxes comes in mind.


The button should convey the action


Following the principle of least astonishment, I'd say the button should describe what happens when clicked.


I think buttons should convey current state by default, and intent when they are hovered over.


I don't have enough UX skills to answer, but I would be curious to see what our folks in the domain think about through a survey?


I don't think there's a true best practice for this. This should be done depending on the context.


I personally believe that the intended behavior should be marked. As it makes understanding more clear and less-technical people can also understand it easily.


I always find Instagram labels following and follower confusing


Consider tooltips to describe about the current state or purpose of the button:

  • Follow (Click to follow)
  • Unfollow (Click to unfollow)

Hi, Ben. I think this is correct and useful, I'd love to see it implemented.


Difficult question tbh, but I would say the intended behavior is the way to go. A button indicates that an action will be performed when clicked.


A button should state the intended behaviour, a user needs to see what a button does, before he's using it.


BOTH as per the github FOLLOW and github UNFOLLOW button/s

code of conduct - report abuse