I Spent ~$3,000 on Ads for a B2B SaaS Nobody Wanted. Here's What I Learned.
Hey everyone. I wanted to share a proper failure story because I think we don't do enough of that here. This isn't a "failure that secretly led to a $10k MRR product" narrative. It's just a failure, and then a different project that's still too early to call a success.
What I built
Between April and September 2025, I built Unify-Live — a multi-channel messaging widget for small businesses. Think one inbox for WhatsApp, Instagram DMs, Facebook Messenger, email. The backend was Python (FastAPI) with PostgreSQL, the frontend was Vue 3 + Pinia + PrimeVue. I ended up with 87 commits in the backend repo, 66 in the widget. Five months of real, focused work.
To keep costs down, I self-hosted the entire production stack on a Raspberry Pi on my desk. No AWS bill. I thought this was clever.
What happened
I had no audience, no blog, no social media presence. So I went straight to paid acquisition — Facebook ads and Google ads, running simultaneously. By my own estimate, I burned roughly $3,000 over those months.
People signed up. Almost nobody stuck around. Retention was basically nonexistent. CAC kept climbing while retention stayed flat. When I eventually paused the ads, traffic went to zero. Not "low." Zero. There was nothing underneath.
I killed the project in September 2025. The code was fine. The business wasn't.
The five mistakes (as I see them now)
1. I never talked to potential users before building. I assumed the problem was obvious and my solution was good enough. Classic.
2. I ran paid ads before proving the product could retain anyone. Paid acquisition without retention is just paying to watch people leave.
3. I had zero organic distribution. No blog, no Reddit presence, no LinkedIn, no content. When paid stopped, everything stopped.
4. I optimized the wrong thing. The Raspberry Pi saved me maybe $30-50/month. The ads were burning hundreds. I was proud of the hosting setup and should have been panicking about retention.
5. Time was always working against me. Without organic compounding in the background, every day was just clock ticking on paid spend.
What I did differently next
In early 2026 I started FlipperHelper, an iOS app for people who resell at car boot sales and flea markets. I deliberately tried to invert every Unify-Live mistake:
- User research first. I scraped and read 2,000+ Reddit posts from reselling communities before writing code. The product spec came from their words.
- Organic distribution from the start. Months before launch, I started participating in reselling subreddits. Not posting links — actually being a community member.
- No server. Everything stored locally on the device, syncs to Google Drive. Zero hosting cost.
- No paid ads. Every download is organic. Reddit, blog, word of mouth.
FlipperHelper is at about 50 downloads and 25 TestFlight users as of mid-April 2026. Tiny numbers. But the shape feels different. People are finding it through content and community, a few have come back with real feedback and feature requests. That never happened once with Unify-Live.
The actual takeaway
Cheap to run doesn't matter if retention is zero. Fix retention before you scale distribution. That's it. That's the whole lesson.
I don't know yet if FlipperHelper will work as a real product. But at least this time I'm building from evidence instead of assumptions, and I'm not paying rent on a treadmill while I figure it out.
Anyone else burn through paid ads before finding PMF? What finally made you stop?
