Everyone seems to be shipping faster than ever. Scroll through your feed long enough, and it’s hard not to feel like you’re falling behind. Someone merged 100 PRs last week. Someone else built an entire app over the weekend. Meanwhile, you’re still working on the same feature you started on Monday. But falling behind what, exactly?

AI has dramatically increased how quickly we can build software. Work that used to take weeks can now take days, sometimes hours. That’s a remarkable improvement, and there’s plenty to celebrate. But somewhere along the way, we started treating the amount of software we ship as a measure of how successful we are. So when you feel like you’re falling behind, you’re usually measuring yourself by volume, and volume says very little about whether anyone is better off.

What do 100 PRs actually tell you?

Consider someone announcing that they’ve merged 100 PRs in a week. What do you know about what they’ve accomplished? Not much, really. Those PRs could be critical bug fixes or features customers have been asking for. They could just as well be dependency updates and refactoring that nobody will ever notice. Some could be tiny, surgical improvements, while others could introduce entirely new capabilities. Without knowing what those changes accomplished, the number itself tells you very little.

Yet we compare ourselves against it. Someone shipped 100 PRs, and we shipped 10. Surely they’re ten times more productive? Of course not. But when everyone around you celebrates how much they’re shipping, it’s easy to start feeling that you should be shipping more too.

Customers aren’t moving at the same speed

There’s another side to this story that we rarely talk about: customers. We’ve sped up how quickly we can produce software, but have we sped up how quickly customers can adopt it? I’m not convinced. Think about how larger organizations adopt software changes. They need to understand what’s changed, evaluate the impact on their processes, assess potential risks, prepare their employees, and sometimes update internal documentation or training. That takes time, regardless of how quickly the software was built.

Now imagine the product they’re using releases updates every few days instead of every few months. By the time they’ve evaluated one update, several more have arrived. And that’s just one product. Multiply that by all the software an organization uses, with each vendor shipping faster because AI makes it possible. We’ve made producing changes easier without making them any easier to consume.

It’s tempting to assume that more features mean more customer value. But a feature that customers don’t know about delivers none, regardless of how impressive it was to build, and the same goes for a feature they can’t adopt or have no use for. You could ship 10,000 features, but if customers only benefit from five, those five are what matter to them. The other 9,995 won’t convince them to stay if their needs aren’t being met.

What are we doing with the time AI saves us?

Suppose a customer reports a bug that previously would have taken your team two weeks to fix. With AI, you can fix it in a day and get the solution into their hands. That’s worth celebrating, because you’ve solved a real customer problem faster. But what happens with the time you’ve saved? Do you spend it understanding what else your customers struggle with? Do you improve documentation or simplify onboarding so customers can adopt the features you’ve already built? Or do you immediately start building the next thing, because everyone else is shipping and you don’t want to fall behind?

Building the next thing is tempting because it gives us tangible results. We can point to merged PRs and shipped features. Helping customers adopt changes is harder to quantify, and the results often take longer to appear. But the fact that something is easy to measure doesn’t make it the right thing to optimize.

So instead of comparing how many PRs we’ve merged, we should talk about how long it takes to solve a customer problem. How quickly can we turn a reported bug into a fix the customer can use? How long does it take to deliver a requested capability? How many customers are benefiting from what we’ve already shipped? There isn’t one universal metric for customer value, because different products solve different problems, and every organization needs to figure out what matters to its customers. That’s harder than counting PRs, but it tells you whether your work made a difference to the people you build for.

You don’t have to keep up

There’s nothing wrong with shipping fast. And there’s nothing wrong with being proud of what you’ve built. AI has given us capabilities that would have seemed extraordinary just a few years ago, and we should make the most of them. But speed only counts for what it lets you do for your customers. If you’re talking to customers about their problems and helping them get more value from your product, you’re not falling behind someone who merged another 50 PRs. You might be doing exactly what your customers need.

The next time you see someone celebrating how much they’ve shipped, be happy for them. Then ask yourself a different question: what have I done lately that made my customers’ lives better? You might find that you’ve made more progress than you think.