Not gospel, just my read. I spent 2 years inside a structure that coached early stage startups, so I sat through a lot of pitches and watched which ones actually turned into revenue. Now I'm building my own SaaS solo, no technical background, and making my own version of these mistakes. Both sides taught me the same thing: the build was never the hard part.
The fastest way to know if your scope is too wide is to try writing your pricing page. Not the landing page, the pricing page. If you can't decide what's in the free tier, what the paid tier unlocks, and why someone upgrades, your product doesn't do one job yet, it does four half jobs.
Same test on onboarding. A focused tool has one obvious first action, the thing that makes a user go "ok I get it". An all in one tool has a dashboard with six empty widgets and no idea where to send people first.
The real cost isn't the code, it's that every extra module adds a positioning you have to defend. Three modules means three ICPs, three messages, three sets of competitors, three landing pages you'll never write. You end up talking to nobody in particular.
What I'd do instead: pick the single job the tool does, ship only that, keep a text file of everything else. Half of what's in that file stops mattering once real users show up.
Most first SaaS start from a feature the founder finds clever, then go looking for someone to sell it to. Wrong order, and it costs months.
The question isn't "is this a good idea", it's "who can I reach this month". Concretely: can you name three specific places your audience already gathers, do you know the words they use for the problem (not the words you use), and do you have any existing way in, a community you're active in, a network, a niche you used to work in. If it's no to all three, you've picked the hardest possible ground before writing a line of code.
The other thing I changed my mind on: competitors are not a reason to stop. When I was coaching, "we're the first in the world to do this" was a red flag every single time. If nobody else is doing it, nobody has proven anyone will pay for it.
What I'd do instead: before building, check whether people already spend money to acquire your audience. Meta Ads Library, search a competitor's page, see if they're running ads and for how long. Google the problem and look at how cluttered the paid results are. An ad that's been live for months is someone telling you the unit economics work. That beats 20 people saying they'd totally use it.
No custom domain, gradient landing page, the same three sections everyone has now, a hero headline that could describe any product. People clock it in two seconds, and it hurts you at the worst possible moment, when a cold visitor is deciding whether to hand over an email or a card.
This matters more in B2B than people admit. Someone spending company money needs to believe you'll still exist in six months. A weekend demo look says you won't.
What I'd do instead: buy the domain, rewrite every line of copy in your own words (the generated copy is a bigger tell than the design), put a real screenshot of the product above the fold, delete the fake logo bar and the fake testimonials. One afternoon, highest return afternoon you'll spend pre launch.
If you're at zero and nobody paid after a week, that's normal. It's not a signal, it's week one. The reflex is to declare the idea dead and start something else, and six months later you have five half built products and no learning that transfers.
Before switching, walk the funnel and find where it actually breaks. There are only four places.
No traffic at all. That's a distribution problem, not a product one. You didn't test the idea, you tested nothing, so changing the idea changes nothing.
Traffic but no signups. The landing page isn't saying what the thing does or who it's for. Usually a day of rewriting.
Signups but nobody reaches a first real result. Activation. They wanted it enough to create an account and then hit friction. Most fixable, most ignored.
Activated users who don't convert. The value is real, the price or the timing is off. Now you have something worth talking to people about.
Only one of those four is "wrong idea", and it's the rarest one. Also worth saying: don't read anything into your percentages at 15 users. Ratios are noise at that volume. Five real conversations tell you more than the dashboard will.
The thread through all four: building was never the bottleneck, distribution was. Each of these mistakes is a version of skipping the distribution question and hoping the product answers it for you.
Curious where people disagree, especially on picking the ICP before the idea.