Why big redesigns fail

One mystery that’s always tugged at me is why big redesigns tend to flame out. They take forever to ship. They rarely end in more money or happier customers. Then I saw Stripe’s new nav redesign, and it finally clicked.

A redesign often makes things harder for customers… even if it’s technically “better”. And with all the big moves in AI, redesigns may be a thing of the past. Let’s unpack what this means.

As a customer of too many software solutions 😅, there are things that bother me now that was never on my radar as a product manager. The first is the gravity of habit. Once I figure out how to use something, almost any change to my workflow is painful.

In product reviews, we compare old vs. new, but we often forget the elephant in the room: the old is familiar. The new has to not only be better, it needs to be overwhelmingly better that it escapes the gravity of habit.

When a product team sees a crappy design, the first thought is Low-hanging fruit! But the customers that have stuck around beat the odds and figured it out. They’re not psyched to re-learn everything — even if it’s 20% easier this time around.

A crappy old system that’s familiar holds a quiet edge over the new and unknown. So the job of a redesign is not only to make the new undeniably better, it’s to understand and bridge the old habits. The devil lies in re-onboarding existing customers, especially for a bigger business.

Let’s take Stripe as an example. They recently moved navigation from the top of the page to the left side, along with some edits to the hierarchy. Given the size of the org, I appreciate this change was a big investment. As a customer though, I felt instant dread.

In the old design, the places I needed most were buried in unexpected corners. The new design buries them yet again, only now they’re in new corners that I have to remember. 🥲

Out of the hundreds of possible paths in Stripe, I only need 3. My greatest wish is to have those 3 paths be easy to find.  Why bury SaaS metrics for a subscription business? I would love to see that on the homepage. Or if there’s a failed payment, help me retry processing, don't hide it 2 clicks down.

It’s clear they sweat over the pixels. But not enough attention is paid to paving clear paths within the product. This problem isn’t unique to Stripe though. It's a common reality, especially if the builders are not the customers. Everyone sees pixels, but only real customers experience the paths.

Naturally, Stripe gets a huge pass. If you've built a robust payment processor, I’ll put up with some UX papercuts, but most businesses are far less resistant.

I also now understand the customer who resists all change. I use PgAdmin to edit my databases. It’s powerful but has a crazy dense interface. It took a long time to learn the ropes, and now I refuse to get updates… I’m about 10 versions behind.

I swear I like progress, but I’m paranoid they won’t support the features I need, or they’ll be buried in some obscure corner. Both of which are good questions for any redesign: does it visibly support all core features? And how will old people find their way around their new home?

I admit it’s far easier to critique someone else’s product. There are countless invisible trade-offs on every screen. I know my own use case, but I’m ignoring all the others that have to be supported.

To even things out, I always return to myself and ask: in what ways am I making the same mistake? Being attentive consumers can make us better creators.

Sometimes a product is complex because it does so much. Here’s a clear sign: when it’s faster to ask ChatGPT or YouTube for guidance than tinkering with the product. The old-school answer to this problem is: redesign! These days, there's a bigger opportunity on the horizon. Instead of relying on ChatGPT or YouTube to demystify the product, bring it in-house. Build an AI agent to help me find what I need, or take action on my behalf.

Most software is some version of DIY (do it yourself) that beats doing it by hand. The next leap forward would be some version of DFY (done for you) that turns intent into outcome.

Google is DIY, ChatGPT is DFY. Even when DFY is imperfect, the magic of getting an instant answer is enough to keep people hooked. I'm betting that the best software will evolve from organizing pages for customers to act on, to taking actions for customers.

If I’ve learned anything being a customer, it’s that we care surprisingly little about how pixels are arranged. We just want to get a job done. For that, DFY will be the holy grail.