Building in Public #2: The user who broke my quota system (and the bug almost killing my ads)
Hey everyone,
When I committed here to building AnyDub in public, I promised the raw, unvarnished truth of being a solo dev. I just didn't expect reality to hit this fast or this hard.
If you’ve ever built a software product completely on your own, you know that weird mix of imposter syndrome and adrenaline when the first paid users roll in. You celebrate for two seconds, and then the panic sets in about whether the system will hold up.
Well, 48 hours after launching our paid tiers, a user signed up for our highest plan. I felt a huge wave of relief. People are actually willing to pay for this! But two days later, my dashboard went red.
This user had exhausted their entire monthly quota in just 48 hours.
Because AnyDub handles real time audio translation on the fly, every minute converts to direct API and server costs. I sat there looking at the numbers thinking: "If I don't release a top up minute system right now, this user gets blocked, or my bank account gets drained." It forced a high stress weekend sprint to code an extra minutes add on. Messy and intense, but seeing someone use the tool so heavily was strangely validating.
Then came the real plot twist with my marketing experiments.
I’ve been learning Meta and Google Ads completely from scratch, burning budget, tweaking targeting, and trying to convert desktop extension installs into paid users. The metrics looked awful, and I was starting to accept that my ad funnel was just failing.
Then, I did something simple: I reached out directly to an un-converted user to ask for feedback.
His reply blew my mind. He wrote back saying he loved the product and wanted to upgrade, but every time he tried, the checkout blocked him saying his "24 hour welcome discount had expired."
It turns out I had a logic bug in my dynamic pricing workflow. Once the 24 hour initial discount timer ended, instead of reverting the price back to full, it completely blocked the user from buying any plan.
I was literally paying for ads, driving high intent users to the product, and then closing the door on their faces when they tried to give me money. If I hadn't sent that single, personal feedback email, I would have thrown my entire ad budget down the drain thinking "ads don't work for this."
Amidst the chaos of fixing quotas and unblocking checkouts, I did manage to land a major feature win I’ve been chasing for months: Frictionless Two Way Translation.
In the past, real time two way voice translation required both participants to install extensions, set up accounts, and sync. The friction killed it. Earlier this week, something clicked. I shifted the processing load entirely to the person with the extension. Now, the host hears the incoming speakers dubbed live into their language, while broadcasting their own voice dubbed into the other party's language. The other side doesn't need to sign up, pay, or install a single thing. Testing it live felt like pure magic.
Being a solo founder really is a continuous roller coaster between "I'm burning cash on broken funnels" and "Holy crap, this code actually solves a real problem."
I'd love to hear your thoughts and stories:
How often has a direct email to a user saved you from a silent, fatal bug?
For API heavy tools: How do you structure safety nets when a power user suddenly consumes 10x what you planned for?
Check out what I'm building: https://anydub.pro
Appreciate this community and all the support!
ezlabs
