We're trying out a feature that asks the user how they feel about _____ (Saving money, investing, other banking products), and the user is prompted to tap on either a sad face or a happy face. Whether they answered smily or frowny, the user is brought to the same page, (in our department's case, they go to a savings tool) but the copy changes depending on the user's answer.

I'm looking into alternatives to happy/sad, because we're uncertain as to whether the design will go through. (Color and tone don't match brand)

Alternatives to Happy/Sad

  • A) 1-10 scale
  • B) Likert Test
  • C) Thumbs Up/Thumbs Down
  • C) Other
  • D) These are all bad, and none of them should be used as entry points

I am personally leaning towards D, These are all bad, and none of them should be used as entry points because some users were expecting to take a survey but instead, they were directed to a product.

I don't think a product entry point disguised as a survey is the best option, but I do like that we're trying to recommend products based on user emotion/confidence, and I want to convince everyone here that we can't just assume what our users are feeling, even based on their financials.

So what works? How can I get an accurate emotional choice from a user without getting them lost?

  • Sorry, but I on't understand what you ask. Could you explain how are entry points related to scales or metrics in your case?
    – Devin
    Commented Mar 8, 2017 at 1:18
  • Is this some sort of marketing campaign? Get the user to express an opinion about something and then help them understand why that thing might be important to them? Either way, getting users to do work (thinking about and idea and then expressing an opinion counts as work for users) could leave your users with a bad feeling about your product unless there is some reward for their work. Commented Mar 8, 2017 at 9:09
  • Hi Devin, in this case, entry points are not related to scales/metrics. I think that the person who first designed them (in a different department) brought the happy clickers one place, and the sad clickers to another.
    – WushuDrew
    Commented Mar 8, 2017 at 14:49
  • Andrew, we're trying to get the right product for the right user, but in this case we're doing it by emotion. (Example: How does pineapple on Pizza make you feel? Happy = Pineapple Pizza page, Sad = Pepperoni Page). I had a feeling that this isn't the best approach, so I'm looking for better ones. I'll probably just explain to the stakeholders that this isn't a good practice.
    – WushuDrew
    Commented Mar 8, 2017 at 14:56
  • 1
    All those item you call alternatives are scale systems, hence why I ask. Rating scales and entry points have no relation at all (maybe tangentially in some very specific case), so I'm guessing you're asking for some kind of system to display conditional data based on user choices, where choices are influenced by emotions, is that correct?
    – Devin
    Commented Mar 9, 2017 at 18:28

1 Answer 1


Feelings Are Hard!

Trying to understand feelings about a topic with happy/sad faces is always going to be a difficult task, and probably not a very valuable one in this case due to the complexity of the content. So the idea of understanding feelings and acting on them to improve UX is a good one, but I think you need to do some more in-depth research first.

This is mostly because you'll only ever get responses to the questions you knew to ask and you can't know all of the feelings that users may wish to express. It's also extremely difficult for a user to express mixed feelings about the topic (eg: this bit's good, but that bit's awful) with a simple "happy/sad".

Instead, I'd suggest you start off by doing a bit of user research to try to understand the feelings your customers have towards your products. You can then use the outcome of that to better design what user input you ask for in your UI, and how to act in response to that user input.

A Research Approach

My approach would be to start off by looking at qualitative metrics for a relatively small group of test subjects from a variety of backgrounds and situations which you are attempting to cater for. It may be impractical to do this online and unsupervised, so I'd expect you to have to compensate these users in some way, and to either visit them or do this research remotely.

The format I'd try to start with is something like this:

  1. Recruit - Recruit a number of test participants who fit the personas you are aiming for.
  2. Capture each user's feelings - Ask each of them to use or review the content you're trying to match to emotional responses, and briefly describe their feelings regarding each section of the content. Encourage them to give an initial gut reaction, but also press for a bit more detail.
  3. Generate a *master feelings list - Identify all of the feelings expressed about any of the content by any participant, so you have a full list of feelings experienced - not just happy or sad. You might need to do some normalization work, but be careful when you do this. You may well lose some nuance from the research - so if you do normalize the responses, keep the originals as well so you can check back if things don't go as you expect!
  4. Add any feelings you want to see - If there are feelings you want users to experience, but which have not been mentioned, add them to the list. You may need to do this for positive feelings, or feelings you suspect your users won't think about unless prompted.
  5. Match feelings to content / products - For each product or piece of content, ask each participant to sort the master feelings list in order of how well they match their feelings - putting the best match first, the second best second, and on down the list. They don't have to place every feeling, but you'll have more to work with if they do.
  6. Find commmon placements Work out the percentage of participants who put each feeling in first place, the percentage who put them in second place, third place, etc. so that you have a record that says "50% of users found this confusing, 30% found it reassuring, 20% found it scary, no other feelings matched significantly".

Once you have done this, you'll have some statistics to work with about how a user's feelings match up to the various banking products and the information about them.

Acting on the research

You can use those statistics to determine better options than happy or sad faces, and perhaps make the options into something that's a little bit more meaningful, nuanced and related to things which your users actually feel.

If you have a product that 60% of your test participants said made them feel nervous, and some content explaining it which a 70% of your test participants said made them feel reassured, a user who chooses an option for "this kind of product makes makes me nervous" would be a good person to show the reassuring content to.

It may still boil down to "comfortable" vs "uncomfortable" or "clear" vs "confused", but at least you'll have a better idea how to ensure that the questions you ask about feelings relate in some way to how the content or banking products are perceived!

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.