My first SaaS took 9 months and 4k in ad spent and never made a penny. Here's what my second made in 4 weeks, and everything I learned.
I have been building software on my own for a while, and I want to write down everything that happened between my first real product and my second one, because the gap between them taught me more than anything I have ever read or watched, and I honestly wish someone had sat me down and shown me all of this before I started throwing money at things.
The first one: nine months, £4k, zero payers
The first one was called Ancestorii, a private place for families to keep their memories, photos and stories together. I spent about six months building it with ChatGPT writing most of the code, and by launch day I genuinely believed it was perfect, I thought people would find it and love it the way I did, and what actually happened is nobody came. So I did what I think most of us do and started buying attention. Meta ads, Google search and YouTube ads, Reddit ads, UGC videos from creators, a £70 podcast sponsorship, a £120 post on a family tree magazine's Facebook page, my own podcast on Spotify, SEO pages, and cold messages to genealogy creators where not one of them replied. I even printed physical memory books before a single stranger had uploaded a single memory, which looking back was probably the most expensive mistake in the whole thing.
Some of the numbers are almost funny now. £210 on Meta in seven days brought around 15,000 visits and not one payer. When I took the paywall away the first signup came after £2.80 of spend, and then I sat at about four signups a day on £4 a day of Meta until I had 125 signups and zero people paying. Today it has 177 signups and has never had a paying customer. The worst part is I noticed in the very first week that the signups weren't turning into anything, and I kept spending anyway, on hope, because nobody told me to stop and I didn't want to be the one to say it.
The diagnosis took me far too long to accept. It was a vitamin, not a painkiller. Nice to have, nobody needs it today, nobody loses anything if they never sign up, and no amount of ads fixes that, ads just buy you the bad news faster.
The second one: £25 on ads, and people pay
Prep66 is an exam prep site for the Part 66 licence exams, 13 modules, about 25,000 practice questions with explanations, revision guides, mock exams, the lot. I launched it in the middle of August with one Reddit post in the community where those students hang out, and the first person paid full price nine days later without a discount code.
The only money I've ever put into ads was about £25 on a Reddit campaign, and it brought one signup and zero payers, so I killed it after five days and everything since has been organic. These are the last four weeks straight off the Stripe dashboard: monthly recurring revenue went from about £51 to £171, active subscribers went from 3 to 11, £230 came in gross, zero failed payments, zero refunds. It's not life changing money and I'm not pretending it is, it's less than two days of my shift pay, but the line only goes one way and it's doing that with no ad budget at all, while my first product burned £4k and never moved once.
So what was actually different? I ran both products back through the same questions afterwards and ended up with six things a product needs before a stranger will pay for it. Prep66 has all six. Ancestorii failed five of them.
The six things a product needs before anyone pays
One, the problem has to be mandatory. Nobody has to save their family stories. Someone sitting a licence exam has to pass it, there's no version of their career where they skip it.
Two, there has to be a deadline. An exam has a date, and a date makes people act this week instead of someday. Ancestorii had no clock on it at all, which is exactly why "I'll do it later" killed nearly every signup.
Three, failing has to cost something real. Fail an exam and you lose money and months, sometimes a pay rise. Not using a memory app costs you nothing today, even if it might cost you something in twenty years.
Four, the alternatives have to be expensive or bad. The other ways to prepare are long paid courses, thick textbooks, or older question banks that look like nobody has touched them in years. When the alternatives are pricey or tired, something cleaner and cheaper sells itself.
Five, it has to look credible the second someone lands. A student can tell in about ten seconds whether a prep site was made by someone who actually knows the exam, from the module names, the paper formats, the way the questions are written. Ancestorii was asking strangers to trust a no name site with their family photos, which is the exact opposite.
Six, you have to be able to find the buyers. Exam students hang out in a couple of subreddits and they search Google for the exact module they're stuck on. Who's the buyer for a family memory app? Everyone and no one, which in practice means you pay Meta to guess for you.
How I picked it and checked it before writing any code
The market research for Prep66 wasn't fancy at all. I picked a world I already knew inside out, so I knew what students complain about and what they already pay for, and then I checked four things. People already pay for this today, not "would pay", actually pay, right now, to someone. The options they're paying for are old and dated. Building it used to be really expensive, because it's a mountain of content, and with AI it isn't anymore. And it's too small for a big company to bother with but big enough to pay one person properly. If all four aren't true, I don't touch it now.
Then the validation was the launch itself. I built it lean, put it in front of one subreddit and watched whether strangers paid, and they did inside nine days. With Ancestorii I had the same kind of signal in week one, it just said no, and I ignored it for months.
What actually brought people in: SEO and GEO
Before anything else I put tracking on every signup, so every new account gets stamped with where it came from, either the UTM tags on the link or the site that sent them. That one decision changed how I saw everything, because of the first 13 tracked signups, 6 came from ChatGPT citing the site, 5 came from Google, 1 came direct and 1 came from the £25 ad. Gemini started showing up as a source not long after. I genuinely did not expect AI answers to be my biggest channel, and I'd never have known without the tracking.
What I learned doing it is that you write your titles and headings in the exact words people type, not the words you would use yourself. AI tools quote facts, not vibes, so the pages that got picked up were the boring factual ones: a glossary, a facts and figures page, a page comparing the ways you can study for the exam that doesn't trash anyone by name, and a features page with the prices pulled live so they're never out of date. ChatGPT leans on Bing, so I submitted the sitemap to Bing Webmaster Tools and set up IndexNow so Bing gets pinged every time a public page changes, and the robots file lets the AI crawlers in instead of blocking them.
By late September Google Search Console showed 194 clicks and 6,110 impressions with an average position of 8.5, and about 1,580 of those impressions came from Google's own AI answers. It's not all solved either, Gemini still barely knows we exist, and that's my whole job for October.
Reddit posts written like a normal person talking, plus SEO and GEO, did basically everything.
Where all of this went: AskSami
Once Prep66 started paying I kept thinking about how much of that Ancestorii money I would have kept if someone had looked at what I'd built in week one and told me straight that it was a vitamin and to stop buying ads. Every time I looked at my ad numbers they meant nothing to me, so I'd paste them into ChatGPT or Claude and get the same generic "have you tried different hooks" answers, because they didn't know my product, they didn't know what I'd already tried, and they forgot all of it the next chat.
So I built AskSami. Sami reads your code from your GitHub repo, read only, so it already knows what you've built without you having to explain it, and then it tells you honestly whether you've built a painkiller or a vitamin, who's already selling the same thing, what your edge is if you actually have one, and where your effort and money should go, free channels before paid ones. It ties your signups and payments back to the exact post, page or ad that brought them, which is the tracking lesson from Prep66 done for you, and it remembers everything you've tried so you're never starting from zero again.
You can talk to Sami at asksami.app without even signing up.
If you're sitting where I sat with Ancestorii, with signups coming in and nobody paying, I know exactly how that feels, and I hope something in here saves you a few months and a bit of money. If you want to connect, swap notes or just chat about what you're building, please do, my DMs are always open 🙂