Why anyone would take UX advice from Google (or Facebook, Twitter, Amazon, Microsoft) is beyond me. I have this rule that I don't take UX advice from companies that have shitty/dark/manipulative UX.
Well..Facebook originally was utilitarian data dense UI..which didnt win in all categories but appealed to me as the prickles person I am. The rest of those names are terrible UI that say "Im unique and wizzy!!" far too unsubtley.
Conversion is exactly the metric there and btw have little to do with UX in this specific situation and everything to do with the quality of the content.
No it is not. You haven't provided supporting arguments, but I will anyway: you can sell newspapers with sensation and lies. Look at PCC vs Guardian on the Mark Duggan case.
> There is only one way to measure UX IMO and that is conversion
Are you just saying that outside of a sales funnel there's no way to measure UX? Like, on a day-to-day product usage basis it's not possible to measure my users' experience?
I am saying that the purpose of good UX is to lead to some sort of conversion whatever that might be. Whether that's buying, using, submitting etc doesn't matter but you can't measure something you don't put measurable metrics around.
I discuss UX all day long and have all sorts of discussion around various approaches, have principles and strong opinions on various UX related issues, but I would never claim it's a framework that can be used to judge the UX in itself unless I am willing to put it to the test of something externally and quantifiable.
That's a little overly simplistic. You can increase conversion rates at the expense of user experience, increasing churn and lowering long term LTV. Depending on the metrics you're measuring it can go completely unnoticed.
So when you sell UX to a client you would tell them to not worry about conversion because you give the user a better experience?
I am not against having personal opinions about what good UX is, but the idea that you can measure UX if you don't have measurable metrics is nothing but a marketing idea.
Its about tricking the user vs not tricking the user, and everything between. Churn can be as simple as a user aborting or clicking back right after they commit a certain action. If you don't measure confusion you'll fail to reduce it.
There is no need to make that leap. You're just measuring conversions, period. And why not? If that's what's most important to you or your job.
The UX could be shitty, but if it converts, you're good. There are plenty of dark patterns that fall into this category, including clickbait article titles. Clicking a title just to find out you were baited is horrible UX. But everyone get's paid.
There's so many factors that come into play with conversion, though. And this will vary wildly between B2B, B2C, and enterprise.
I think marketers have invaded the UX space a little too hard. We should be thinking of more inventive metrics. Stuff that isn't generic content marketable acronyms for cheap blog thrills, but more deeply ingrained metrics that correlate with the problem you're solving.
Are you solving for conversions? Then sure, use conversions and call it a day.
I would love to apply this simple yet effective framework for a Non-Enterprise application.
The EAR of this hEARt has no significance in this equation.
Engagement for Enterprise Application is Forced, Enterprise execs Buy Products and make the people under them live with it.
Adoption: on similar lines to Engagement.
Retention: Enterprise, the retention is not to be questions, users are locked-in till the time the application is paid for.
Where do qualitative methods come into the conversation?
Data-driven UX is better than gut feeling alone, but it's certainly not a replacement for qualitative studies. They complement each other very well. In fact, when gut feeling is reinforced or produced by both qualitative and quantitative studies, you're in a good spot.
If your (or better: the team's) intuition, deep interviews and data all point in one direction, then it's very likely to be the right one. By my experience, if you have support from two of the three sources (intuition—quantitative—qualitative), it's often good enough to pick a direction.
However, picking a direction based on the _same two_ sources for every decision you make is not a good idea. The larger framework you employ should include all of them, even if smaller decisions are based on an "incomplete" perspective based on fewer information sources.
Processes based solely on quantitative measures lean towards being reactive. You're viewing the future in the light of data from the past. Often, the past is a reliable predictor of the future. Sometimes, crucially, it's not. And sometimes, key behaviour is captured, but buried in the cheer volume of data.
Qualitative studies add a predictive element since you are able to capture elements that you didn't think to measure for beforehand, or didn't think to look for in the data. Once captured, those elements can sometimes be measured quantitatively, possibly even retroactively.
There is this disturbing trend of mixing the service/product and the interface and calling it "the experience". This is good for marketing maybe (of UX designers probably), but when designing a product, you're actually making the problem incredibly difficult. You're merging different problems into one, then trying to find silver bullets. Hence, everyone reinvents the UI for a "unique" UX. But that's the exact thing users do not want, and that makes everything harder for everybody.
The UI needs to disappear. No one should even "see" it. The perfect interface is still an invisible one. If you go with this principle, then once the UI task is done, we've eliminated UI from UX. Your first task is to eliminate UI from the UX problem.
What is left is the product/service in it's finest, purest, most concise form. And now, as this is adjusted, the UI may need adjustment. But this is the proper order of business.
Take the missing audio jack. Everybody notices it. Hence, this is an obvious UX failure. The designer's "opinion" or "gut" is not more accurate or more important than the user. UX is not about forcing experience. And the UI itself is about non-experience. Just follow this principle, and the best UX will follow.
Of course, even the data driven UX designers won't put the audio jack back despite the "unhappiness". And that is what is most disturbing for me, at least. It seems these UX designers have no principles. I sense arrogance and apathy. But this makes sense if you consider a UI that wants to be noticed as being self important, and a UI that is data driven to lack any personal or emotional touch.
No single user is a general statistic. Basing design on statistics is not design; it's giving up on designing; it's anti-design.
Anyway, I know HN is hardly the place to push POVs but I will keep trying.
Hear hear on anti-design! It is avoidance of design. Designing well is hard. Easy escape route: metrics!
By the time you ship a design with these metrics, much like usability testing, it is too late. You have just spent a bunch of time building and QAing something to a shippable state, only to find out some major usability issue. But the team thought they were almost done.
No single user is another great point. Complex software has multiple personas (types of users) using it. They have different needs and sometimes they are not served well. This HEART score needs to be at least segmented by persona to start being useful.
27 comments
[ 4.2 ms ] story [ 79.9 ms ] threadJust as an example: try to add a new property to an analytics account. The whole ABCD naming of the menus also bugs me, like really?
Good UX should help conversion rate, but so might dark patterns, obnoxious popups, lies, etc.
Otherwise, it just becomes an art discussion which can be amusing but hardly something you can build a business on.
For a newspaper, you could also have whether users feel more informed, whether they are more informed, etc.
Are you just saying that outside of a sales funnel there's no way to measure UX? Like, on a day-to-day product usage basis it's not possible to measure my users' experience?
I discuss UX all day long and have all sorts of discussion around various approaches, have principles and strong opinions on various UX related issues, but I would never claim it's a framework that can be used to judge the UX in itself unless I am willing to put it to the test of something externally and quantifiable.
I am not against having personal opinions about what good UX is, but the idea that you can measure UX if you don't have measurable metrics is nothing but a marketing idea.
The UX could be shitty, but if it converts, you're good. There are plenty of dark patterns that fall into this category, including clickbait article titles. Clicking a title just to find out you were baited is horrible UX. But everyone get's paid.
I think marketers have invaded the UX space a little too hard. We should be thinking of more inventive metrics. Stuff that isn't generic content marketable acronyms for cheap blog thrills, but more deeply ingrained metrics that correlate with the problem you're solving.
Are you solving for conversions? Then sure, use conversions and call it a day.
Data-driven UX is better than gut feeling alone, but it's certainly not a replacement for qualitative studies. They complement each other very well. In fact, when gut feeling is reinforced or produced by both qualitative and quantitative studies, you're in a good spot.
If your (or better: the team's) intuition, deep interviews and data all point in one direction, then it's very likely to be the right one. By my experience, if you have support from two of the three sources (intuition—quantitative—qualitative), it's often good enough to pick a direction.
However, picking a direction based on the _same two_ sources for every decision you make is not a good idea. The larger framework you employ should include all of them, even if smaller decisions are based on an "incomplete" perspective based on fewer information sources.
Processes based solely on quantitative measures lean towards being reactive. You're viewing the future in the light of data from the past. Often, the past is a reliable predictor of the future. Sometimes, crucially, it's not. And sometimes, key behaviour is captured, but buried in the cheer volume of data.
Qualitative studies add a predictive element since you are able to capture elements that you didn't think to measure for beforehand, or didn't think to look for in the data. Once captured, those elements can sometimes be measured quantitatively, possibly even retroactively.
The UI needs to disappear. No one should even "see" it. The perfect interface is still an invisible one. If you go with this principle, then once the UI task is done, we've eliminated UI from UX. Your first task is to eliminate UI from the UX problem.
What is left is the product/service in it's finest, purest, most concise form. And now, as this is adjusted, the UI may need adjustment. But this is the proper order of business.
Take the missing audio jack. Everybody notices it. Hence, this is an obvious UX failure. The designer's "opinion" or "gut" is not more accurate or more important than the user. UX is not about forcing experience. And the UI itself is about non-experience. Just follow this principle, and the best UX will follow.
Of course, even the data driven UX designers won't put the audio jack back despite the "unhappiness". And that is what is most disturbing for me, at least. It seems these UX designers have no principles. I sense arrogance and apathy. But this makes sense if you consider a UI that wants to be noticed as being self important, and a UI that is data driven to lack any personal or emotional touch.
No single user is a general statistic. Basing design on statistics is not design; it's giving up on designing; it's anti-design.
Anyway, I know HN is hardly the place to push POVs but I will keep trying.
By the time you ship a design with these metrics, much like usability testing, it is too late. You have just spent a bunch of time building and QAing something to a shippable state, only to find out some major usability issue. But the team thought they were almost done.
No single user is another great point. Complex software has multiple personas (types of users) using it. They have different needs and sometimes they are not served well. This HEART score needs to be at least segmented by persona to start being useful.