Original Summary

I run a small AI tool that redesigns existing PowerPoints to look professionally made. After six months of building, testing, and trying pretty much everything in the playbook to get customers, we're finally seeing some traction. The numbers are still small (okay, very small). But looking back, it's interesting how wrong we were about almost everything, and how the data was telling us what to do long before we listened. Figured I'd share the journey, numbers included, because I would've loved to read something like this a year ago. First mistake: building for the customer we wanted Our original idea was personalized pitch decks for sales teams. Upload your branding, point the tool at a prospect's website, get a tailored presentation. Our homepage, onboarding, emails, pricing, everything was written for sales reps. We spent around $2,000 on Meta and Google ads. About 5% of clicks became free signups, which seemed reasonable. But as far as we can tell, not a single one became a paying customer. In our first five months, we got four mid-tier subscriptions. All four canceled before their second payment. Not exactly the recurring revenue machine we'd imagined... Second mistake: pricing for the wrong behavior Looking at our early customers, we noticed everyone who paid had spent 1.5 to 2 hours inside the product first. People who made one deck and left never paid. And why would they? The free plan let them finish their presentation and download a PDF or PowerPoint with a small badge. They'd already gotten what they came for. We even had a $12 "remove the badge" option buried in Stripe. Nobody used it. In one 30-day period, we recorded exactly ONE checkout click. We'd built subscription pricing for people who needed presentations regularly, when most visitors just needed to finish one. So we turned off ads and made a bet on organic At some point, spending more money to acquire users who weren't paying stopped making sense. We decided we'd either figure out how to grow organically or move on. We'd read plenty about optimizing for LLMs and getting products recommended by AI assistants, so we went all in. Structured data and schema markup, clear page headings, FAQ-style answers, dedicated landing pages for individual features and use cases, internal linking, getting pages indexed in Bing, and making our content easy for assistants to find and understand. Basically everything you're supposed to do for AEO/GEO, or whatever we're calling it this week. We weren't particularly confident any of it would work. But it was something we could do without pouring more money into ads. And then something unexpected happened. ChatGPT started sending us customers We had a few pages about redesigning an existing PowerPoint, a feature we'd barely emphasized. Google search volume for phrases like "redesign my PowerPoint" was practically nonexistent, so we'd never thought much of them. But peo

  • 情报分类：服务器与云资源
  • 分类依据：内容涉及服务器、云资源或网络线路
  • 信息来源：Reddit · SaaS
  • 发布时间：2026/10/15 05:46:06