⛏️ What's the last thing that broke while building a Product?

Many makers have shared the same experience: posts about what broke are 1000x more interesting than posts about polished things that are working perfectly =D

And sometimes, nothing is actually broken.
You just discover something that makes you change direction.

For us, we built a CMS where users could create their own pages:

userpage/ourdomain.com

Everything was working fine. But then we started thinking about potential bad players and content we wouldn't want associated with our main domain.

So we changed the approach and bought a separate domain: ourdomain.site

Now user pages will live there, while our main domain stays clean.

Not a disaster... Not really something that broke! Just one of those small discoveries that changes the direction of what you're building.

What's the last thing that broke or simply made you change direction while building?

143 views

Add a comment

Replies

Best

We've built a website optimisation audit tool. When we did our first round of user testing, we very quickly discovered that too many audits running at the same time caused it to crash and do nothing. When I say "too many audits" that was 2 at once. In our second round of user testing, we identified a new use for the platform. We originally built it for non-technical small business owners, but technical and large companies were interested. Now we have 2 use cases that we cover!

 Wow, inspiring story! Haha "too many audits" that was 2 at once =D
What’s the second use case that interested tech larger companies?

 haha yeah we just didn't know the system would reach its cap at that point, but that's why we test right?!

The second use case we learned is that agencies can bring a scored, ranked UX report to every pitch. Scope work from the findings, prove the result with a re-audit, and give clients a reason to keep them on. It provides a faster first layer of analysis to identify opportunities, support research, and prioritise areas for deeper investigation.

 Great! Are you planning to launch it here?

 Absolutely! We're gathering and creating the content needed and planning the launch.

For us, one unexpected turning point was buying the .ai domain. We were originally using a .co domain while another old and established company with similar sounding name owned the .com. Even though the businesses were completely different, it created a real identity problem and made it harder to position the brand clearly. Moving to .ai felt like a small naming decision at first but it meant migrating the website, blog, redirects, analytics and search indexing. It was a fair amount of work but so far it has turned out to be one of the better decisions we made.

 Choosing the right domain is definitely a must-have from the beginning.

I remember we spent weeks debating what to do with ours =D

But I’m curious - what made you feel it was a better decision?

multiple things - less brand confusion in Google Search, clearer positioning as an AI-first platform & better recognition by AI engines - our traffic has also improved consistently - though this cannot be only attributed to domain switch but I am sure it has played a major role in this.

 Cool ! It’s nice to hear that things are really changing for the better.

Buying the second domain also means those user pages stop building anything for your main domain, which is the half nobody mentions. From where I sit doing SEO the trade is still worth it, because shared reputation cuts both ways and you cannot unwind it once bad content is indexed. Do the user pages link back to the main domain?

 Good question. We never expected those pages to build much authority because serious users usually connect their own domains. The free-tier pages link back with a small “Made with Pagenth” badge, but we're after brand awareness and clicks, not SEO credit.

The unexpected benefit was security: a separate domain creates a cleaner cookie boundary between tenants. And honestly, acme.pagent.site- simply feels like a namespace people want to claim. We went looking for a safety fix and found a better product direction.

  So the badge is the part worth watching. On the old setup those links were internal and worth nothing, but from a separate domain they point back at you as external links, which is the SEO credit you thought you had given up. Worth checking whether the badge is nofollowed, because a site-wide footer link at that scale gets discounted anyway.

 You're right, and we hadn't thought about it that way. The same badge was worth nothing while it was internal, and the split is what turned it into an actual link. We haven't nofollowed it, and now we definitely won't. Fair on the site-wide discount though. We never planned around that link carrying weight. The badge earns its place on referral clicks and people seeing the name, so anything Google credits on top is a bonus we didn't budget for. Came in thinking we traded SEO for safety, turns out the trade was better than we understood.

Mine is embarrassing in a very specific way: I broke my own server and then spent half a day investigating the "attack".

I was setting up a new staging box and tested SSH access the lazy way - with password auth, a few wrong attempts, just to check the error messages. Then the server went dead.

Connection refused. Firewall rules looked fine, the console showed sshd running, tcpdump showed my SYN packets arriving and getting nothing back. I was deep into "is this the hosting provider? is this a routing issue?" territory.

It was fail2ban. Doing exactly what I had configured it to do, protecting the server from the only attacker it ever met: me.

The direction change: a written rule that access never gets tested with password auth, ever, and the unban command now lives in the runbook next to the words "check this FIRST, you idiot". The number of mysterious outages dropped by exactly one hundred percent.

 Haha =DDD Laughing out loud!

Never thought a technical issue could be described this funnily and so artistically! "D

Thanks for sharing it!

It might be useful to let users set their own priorities before comparing products, like price, reliability, or specific features. That could make the results feel much more personal.