Tags: test

I was doing some accessibility work with a client a little while back. It was mostly giving their site the once-over, highlighting any issues that we could then discuss. It was an audit of sorts.

While I was doing this I started to realise that not all accessibility issues are created equal. I don’t just mean in their severity. I mean that some issues can—and should—be caught early on, while other issues can only be found later.

Take colour contrast. This is something that should be checked before a line of code is written. When designs are being sketched out and then refined in a graphical editor like Figma, that’s the time to check the ratio between background and foreground colours to make sure there’s enough contrast between them. You can catch this kind of thing later on, but by then it’s likely to come with a higher cost—you might have to literally go back to the drawing board. It’s better to find the issue when you’re at the drawing board the first time.

Then there’s the HTML. Most accessibility issues here can be caught before the site goes live. Usually they’re issues of ommission: form fields that don’t have an explicitly associated label element (using the for and id attributes); images that don’t have alt text; pages that don’t have sensible heading levels or landmark regions like main and nav. None of these are particularly onerous to fix and they come with the biggest bang for your buck. If you’ve got sensible forms, sensible headings, alt text on images, and a solid document structure, you’ve already covered the vast majority of accessibility issues with very little overhead. Some of these checks can also be automated: alt text for images; labels for inputs.

Then there’s interactive stuff. If you only use native HTML elements you’re probably in the clear, but chances are you’ve got some bespoke interactivity on your site: a carousel; a mega dropdown for navigation; a tabbed interface. HTML doesn’t give you any of those out of the box so you’d need to make your own using a combination of HTML, CSS, JavaScript and ARIA. There’s plenty of testing you can do before launching—I always ask myself “What would Heydon do?”—but these components really benefit from being tested by real screen reader users.

So if you commission an accessibility audit, you should hope to get feedback that’s mostly in that third category—interactive widgets.

If you get feedback on document structure and other semantic issues with the HTML, you should fix those issues, sure, but you should also see what you can do to stop those issues going live again in the future. Perhaps you can add some steps in the build process. Or maybe it’s more about making sure the devs are aware of these low-hanging fruit. Or perhaps there’s a framework or content management system that’s stopping you from improving your HTML. Then you need to execute a plan for ditching that software.

If you get feedback about colour contrast issues, just fixing the immediate problem isn’t going to address the underlying issue. There’s a process problem, or perhaps a communication issue. In that case, don’t look for a technical solution. A design system, for example, will not magically fix a workflow issue or route around the problem of designers and developers not talking to each other.

When you commission an accessibility audit, you want to make sure you’re getting the most out of it. Don’t squander it on issues that you can catch and fix yourself. Make sure that the bulk of the audit is being spent on the specific issues that are unique to your site.

You can listen to an audio version of Weighing up UX.

This is the month of UX Fest 2021—this year’s online version of UX London. The festival continues with masterclasses every Tuesday in June and a festival day of talks every Thursday (tickets for both are still available). But it all kicked off with the conference part last week: three back-to-back days of talks.

I have the great pleasure of hosting the event so not only do I get to see a whole lot of great talks, I also get to quiz the speakers afterwards.

Right from day one, a theme emerged that continued throughout the conference and I suspect will continue for the rest of the festival too. That topic was metrics. Kind of.

See, metrics come up when we’re talking about A/B testing, growth design, and all of the practices that help designers get their seat at the table (to use the well-worn cliché). But while metrics are very useful for measuring design’s benefit to the business, they’re not really cut out for measuring user experience.

People have tried to quantify user experience benefits using measurements like NetPromoter Score, which is about as useful as reading tea leaves or chicken entrails.

So we tend to equate user experience gains with business gains. That makes sense. Happy users should be good for business. That’s a reasonable hypothesis. But it gets tricky when you need to make the case for improving the user experience if you can’t tie it directly to some business metric. That’s when we run into the McNamara fallacy:

Making a decision based solely on quantitative observations (or metrics) and ignoring all others.

The way out of this quantitative blind spot is to use qualitative research. But another theme of UX Fest was just how woefully under-represented researchers are in most organisations. And even when you’ve gone and talked to users and you’ve got their stories, you still need to play that back in a way that makes sense to the business folks. These are stories. They don’t lend themselves to being converted into charts’n’graphs.

And so we tend to fall back on more traditional metrics, based on that assumption that what’s good for user experience is good for business. But it’s a short step from making that equivalency to flipping the equation: what’s good for the business must, by definition, be good user experience. That’s where things get dicey.

Broadly speaking, the talks at UX Fest could be put into two categories. You’ve got talks covering practical subjects like product design, content design, research, growth design, and so on. Then you’ve got the higher-level, almost philosophical talks looking at the big picture and questioning the industry’s direction of travel.

The tension between these two categories was the highlight of the conference for me. It worked particularly well when there were back-to-back talks (and joint Q&A) featuring a hands-on case study that successfully pushed the needle on business metrics followed by a more cautionary talk asking whether our priorities are out of whack.

For example, there was a case study on growth design, which emphasised the importance of A/B testing for validation, immediately followed by a talk on deceptive dark patterns. Now, I suspect that if you were to A/B test a deceptive dark pattern, the test would validate its use (at least in the short term). It’s no coincidence that a company like Booking.com, which lives by the A/B sword, is also one of the companies sued for using distressing design patterns.

Using A/B tests alone is like using a loaded weapon without supervision. They only tell you what people do. And again, the solution is to make sure you’re also doing qualitative research—that’s how you find out why people are doing what they do.

But as I’ve pondered the lessons from last week’s conference, I’ve come to realise that there’s also a danger of focusing purely on the user experience. Hear me out…

At one point, the question came up as to whether deceptive dark patterns were ever justified. What if it’s for a good cause? What if the deceptive dark pattern is being used by an organisation actively campaigning to do good in the world?

In my mind, there was no question. A deceptive dark pattern is wrong, no matter who’s doing it.

(There’s also the problem of organisations that think they’re doing good in the world: I’m sure that every talented engineer that worked on Google AMP honestly believed they were acting in the best interests of the open web even as they worked to destroy it.)

Where it gets interesting is when you flip the question around.

Suppose you’re a designer working at an organisation that is decidedly not a force for good in the world. Say you’re working at Facebook, a company that prioritises data-gathering and engagement so much that they’ll tolerate insurrectionists and even genocidal movements. Now let’s say there’s talk in your department of implementing a deceptive dark pattern that will drive user engagement. But you, being a good designer who fights for the user, take a stand against this and you successfully find a way to ensure that Facebook doesn’t deploy that deceptive dark pattern.

Yay?

Does that count as being a good user experience designer? Yes, you’ve done good work at the coalface. But the overall business goal is like a deceptive dark pattern that’s so big you can’t take it in. Is it even possible to do “good” design when you’re inside the belly of that beast?

Facebook is a relatively straightforward case. Anyone who’s still working at Facebook can’t claim ignorance. They know full well where that company’s priorities lie. No doubt they sleep at night by convincing themselves they can accomplish more from the inside than without. But what about companies that exist in the grey area of being imperfect? Frankly, what about any company that relies on surveillance capitalism for its success? Is it still possible to do “good” design there?

There are no easy answers and that’s why it so often comes down to individual choice. I know many designers who wouldn’t work at certain companies …but they also wouldn’t judge anyone else who chooses to work at those companies.

At Clearleft, every staff member has two levels of veto on client work. You can say “I’m not comfortable working on this”, in which case, the work may still happen but we’ll make sure the resourcing works out so you don’t have anything to do with that project. Or you can say “I’m not comfortable with Clearleft working on this”, in which case the work won’t go ahead (this usually happens before we even get to the pitching stage although there have been one or two examples over the years where we’ve pulled out of the running for certain projects).

Going back to the question of whether it’s ever okay to use a deceptive dark pattern, here’s what I think…

It makes no difference whether it’s implemented by ProPublica or Breitbart; using a deceptive dark pattern is wrong.

But there is a world of difference in being a designer who works at ProPublica and being a designer who works at Breitbart.

That’s what I’m getting at when I say there’s a danger to focusing purely on user experience. That focus can be used as a way of avoiding responsibility for the larger business goals. Then designers are like the soldiers on the eve of battle in Henry V:

For we know enough, if we know we are the kings subjects: if his cause be wrong, our obedience to the king wipes the crime of it out of us.