I spent months building MarketingNow, watched it flatline in real time, and then built the right product in a fraction of the time. This is the story of what went wrong, what I actually learned from it, and why killing your project can be the best decision you make.
What MarketingNow Was
The pitch was simple: an AI-powered tool to help founders handle their personal brand and marketing content. Think of it as a marketing co-pilot that could generate social posts, draft newsletters, suggest content strategies, and help indie hackers who are terrible at marketing (most of us) actually show up consistently online.
It sounded like a great idea. I believed in it. I got early interest. People signed up for the waitlist. A few early users got access. And then… nothing happened. Not a dramatic crash. Just a slow, quiet flatline where nobody was really using it and nobody was paying for it.
What Went Wrong
I want to be honest here because I think most “pivot stories” gloss over the ugly parts. They frame the failure as some noble learning experience that was always leading to the next thing. The truth is simpler: I made real mistakes, and they cost me months.
The product was overengineered. I built features for users who did not exist yet. Content calendars, multi-platform publishing, analytics dashboards, template libraries. I convinced myself that a complete suite would be the differentiator. In reality, nobody was asking for any of it. I was building for an imaginary power user instead of solving one painful problem for one real person.
The validation was fake. This one stings the most. I did “validation” the way every indie hacking guide tells you to. I talked to people. I posted about it. I collected signups. People said encouraging things like “this looks amazing” and “I would definitely use this.” But here is the thing: saying nice things is free. Paying for something is not. When I looked closer at the early outreach numbers, a significant chunk of the “interest” came from AI bots interacting with my posts and landing pages. I was measuring engagement that was not even human. The real founders who did sign up mostly poked around once and never came back.
The costs did not make sense. The AI infrastructure needed to power all those features was expensive. At the price points that made sense for indie hackers and small founders, the unit economics were brutal. I could not charge $49/month for something targeting people who are careful about every dollar, but I could not run the product sustainably at $15/month either. The operational model was broken from the start, and I just did not want to see it.
The problem was too broad. “Help founders with marketing” is not a problem. It is a category. It is like saying you are building a tool to “help people be healthier.” Where do you even start? I tried to start everywhere, which meant I was solving nothing deeply. The product had no sharp edge. It was a Swiss Army knife for a user who just needed a really good screwdriver.
The Moment It Clicked
The realization did not come from a book or a podcast or a Twitter thread about product-market fit. It came from my own daily routine.
I was spending two to three hours every day on X. Writing posts, engaging with replies, trying to figure out what worked, what the algorithm favored, what time to post, how to structure threads. And I was not unique in this. Every founder I talked to had the same complaint: growing on X feels like a part-time job, and most of us already have a full-time one.
That was the problem. Not “marketing” as a vague category. Not content creation in general. One specific, painful, time-consuming thing that indie hackers deal with every single day. The kind of problem where if you could save someone even an hour a day, they would pay for it without thinking twice.
MarketingNow tried to be everything for everyone. The real opportunity was being one thing for one group of people, and being incredibly good at it. I wrote more about why distribution matters more than building, and this experience is a big part of what shaped that perspective.
What XPilot Was
XPilot was an AI autopilot for your X presence. You connected your account, told it about yourself and your goals, and it handled the rest: content generation, scheduling, and continuous adaptation based on what actually performed well for your audience.
The interface was chat-first. No complex dashboard to learn, no 47-tab navigation, no onboarding flow that takes 20 minutes. You talked to it like you would talk to a co-founder who happens to manage your X account.
A few things that made it different from the scheduling tools that already existed:
It was algorithm-aware. X open-sourced their recommendation algorithm, and XPilot was built with deep awareness of how it actually works. Not guesswork about “best times to post.” It understood what signals the algorithm rewards and optimized content accordingly.
It was chat-first. Most social media tools feel like enterprise software shrunk down for indie hackers. XPilot felt like a conversation.
What Building It Felt Like
Building MarketingNow felt like pushing a boulder uphill while blindfolded. Every decision required debate. Every feature felt like a guess. There was a constant, nagging feeling that I was building the wrong thing, and I kept ignoring it because I had already invested so much time.
Building XPilot felt completely different. The problem was real because I had it myself. I was the first user, and I knew exactly what I needed. When I had to make a product decision, I could just ask myself: “Would this actually help me grow on X?” If the answer was yes, I built it. If it was not, I moved on. No second-guessing. No imaginary user personas. No feature debates.
The speed difference was dramatic. Decisions that took weeks with MarketingNow took hours with XPilot. Not because I was rushing, but because clarity makes everything faster. When you know exactly what problem you are solving and exactly who you are solving it for, the path forward stops being ambiguous. Once I had the product running, I built an automation stack to handle support, onboarding, and ops so I could stay focused on building.
The Lesson I Am Taking With Me
Do not build for imaginary users. It sounds obvious when you say it out loud, but it is incredibly easy to fall into. You sketch out a product, you imagine the kind of person who would use it, you build features for that person, and then you discover that person does not actually exist. Or they exist but they do not care about the problem as much as you assumed.
Do not mistake polite interest for product-market fit. People will tell you your idea is great because that is the socially easy thing to do. The only validation that counts is someone pulling out their credit card. Everything else is noise.
Build for yourself first, then find the people who share your exact problem. If you are not your own first user, you are guessing. And guessing is expensive.
One sharp tool beats a Swiss Army knife nobody asked for. Scope is not ambition. Scope is risk. The tighter your focus, the faster you learn whether you are right or wrong. And if you are wrong, at least you find out before you have burned six months on a product that does everything and solves nothing.
Update
XPilot has since been discontinued. The lessons from building it, especially around solving your own problem and keeping scope tight, still hold. If you are interested in how I approached the operational side of running it solo, check out my post on the solopreneur automation stack.