My co-host Rana spent three days trying to give Render money.
Not because she didn’t want to pay. She’d already built two apps on it: a stock tracker and an eBay reselling agent that classifies her clothes from photos and writes the listings automatically. Real products, live, with friends asking to use them. She was ready to upgrade. She just couldn’t figure out what she was buying, and the product fought her the whole way.
That’s what we pulled apart on this week’s teardown. Worth saying up front: Render is a great product. Rana hosts her side projects on it. One of my clients runs on it. People will crawl over broken glass to use the thing. Which is exactly why its monetization flow is such a waste.
Start with the button. When you finally find the screen to upgrade, the call to action says “Save changes.” I’m handing over my credit card and the button reads like I’m submitting a pull request. It should say “Check out.” It should say almost anything else.
Then there’s where that button lives, buried at the bottom of a long scrollable page, under a warning that your database will be deleted if you don’t pay. The thing your eye gets pulled to is the threat, not the action. It reminded me of changing a Gmail setting and forgetting to scroll down to save. Easy to think you upgraded, walk away, and come back wondering why you’re still on the free tier.
It gets worse before payment is even on file. Picking a plan throws an error code instead of sending you to add a card. The plans don’t auto-select. Behavioral economics 101 says you steer people toward the plan they’re most likely to pick. Render makes you choose blind across four options that don’t match the four tiers on its own pricing page. Pro claims Postgres is included, yet Rana still pays six or seven dollars a month for the database. There’s no annual plan, so no way to lock her in even though she told us she’ll be on this for a year.
Here’s the part that matters. Rana and I have something like forty to fifty years in tech between us. The easy way to wave off this critique is “you two aren’t engineers, you don’t get the terminology.” We get it. It was still confusing. That’s the tell.
This is where vibe coding shows its seams. The product is excellent at the engineering and reads like it was built only by engineers. The upsell, the checkout, the moment a happy customer wants to pay, got none of the same care. It’s not a bad product. It’s a great product leaving money on the table.
And the people walking through that door are changing. Rana hadn’t written code in two decades. Now she ships apps with Claude Code and posts them on LinkedIn, and the response is “wow, how did you do that.” She’s the front edge of a wave of builders, not engineers, who don’t know a web server from a persistent disk and shouldn’t have to. They want an app that tracks their kid’s sleep, working in ten minutes.
Most of these tools are still built for Silicon Valley. The winner won’t be whoever has the most powerful infrastructure. It’ll be whoever hides it. The version that works looks less like a spec sheet and more like Home Depot: here are the five things you need, these boards and these screws, go build.
The bottleneck used to be writing the code. That’s mostly gone now. The new one is everything around it, and the least glamorous screen in the product, the one where someone is trying to hand you money, is where it’s hiding.
If you’re a PM reading this, go look at your own checkout. Then go fix it.
