Testing an idea before building it: a tool to catch scope creep before it happens
I'm a non-technical student in India, exploring my first idea: Margin.
The problem: freelancers lose money to "quick favor" requests that quietly drift outside scope. One consultant on Reddit put it perfectly — "one quick question turns into about 5 hours of unpaid work."
The idea: instead of manually logging every request (what most existing tools require), Margin would watch your conversations and flag the moment something drifts out of scope — before you agree to it.
Where I'm at: nothing built yet. Just a landing page, doing real customer discovery first — Reddit, freelancer communities, direct conversations — before writing any code.
What I've learned: there's already a real competitor (ScopeMatter) doing a manual version — you log and track requests yourself, share transparent scope views with clients. That's sharpened my actual question: do freelancers want automatic detection, or is manual-but-transparent good enough?
Would love feedback on:
— Is automatic detection solving a real problem, or would you rather stay in manual control?
— If you've dealt with scope creep, what did you actually do?
Landing page: https://tinyurl.com/try-margin
This'll be a paid product eventually if it's worth building — right now I just want honest input.
Validating before writing any code is the right order, especially with an existing manual competitor already in the space, that tells you the problem is real but also that you need a real reason to be different. The alert fatigue risk in the comments is worth taking seriously though, over-flagging minor requests would make people ignore it exactly like they ignore every other notification that cries wolf. Is the plan to detect automatically, or lean toward just prompting the awkward conversation at the right moment instead?
I'm curious what convinced you the real problem is detecting scope creep rather than helping freelancers confidently push back once it appears.
Did your customer conversations consistently point to recognition being the harder part, or to handling the conversation after it's recognized?
Honestly, not fully — good catch. Most of what I've found so far points to real pain on both sides, but I haven't specifically asked people which part is harder: noticing the drift, or having the awkward conversation once they've noticed. My guess now is it might actually be the conversation part, not detection which would mean the drafted reply piece matters more than I originally weighted it. Going to go ask people that directly. Thanks ,this actually changes what I should be testing next.
Appreciate the context.
Glad the question was useful. Would be interesting to continue the conversation as you learn from those tests.
What's the best email to reach you on?
Appreciate that — happy to keep you posted. [[email protected] ] works, or feel free to just keep commenting here too, whichever's easier for you.
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.
Strong problem. Scope creep is real, and starting with discovery instead of code is the right move.
On automatic vs manual: I would not trust auto-detection as the core product at first. Conversations are messy, sarcasm and context get misread, and a false "out of scope" flag can damage a client relationship. Manual logging feels annoying, but freelancers may accept it if the client-facing transparency is the win (what ScopeMatter seems to lean on).
Where automation might still help: a lightweight assist, not autopilot. For example, draft a "this looks outside scope" reply the freelancer can edit and send. Keep the human in control.
What people actually do today: eat the hours, raise the rate next time, or send a change-order email after they already feel burned. The moment of failure is usually before they agree, so your "flag before you say yes" framing is good.
One discovery question worth asking: would you pay monthly for this, or only if it sits where you already work (email, Slack, WhatsApp)? Integration friction may matter more than detection quality.
Good luck. Honest that nothing is built yet makes the ask more credible.
The real risk with automatic detection isn't accuracy, it's alert fatigue. If Margin flags every borderline request, freelancers tune it out within a week, same as spam filters that fire on non-issues. I'd rather see it flag rarely but be right when it does, even if that means missing some early drift. Manual tools like ScopeMatter don't have that trust problem because the freelancer is the one deciding what counts, not the algorithm.
The commenter who pushed on the awkward conversation is onto the real bottleneck. I see the same pattern reviewing diffs from a coding agent. It rarely does one obviously out of scope thing, it does five small changes that each look reasonable on their own and only add up to scope creep once you step back and look at the whole diff instead of the individual commit messages. If Margin can show the accumulated drift instead of just flagging the first message that crossed a line, that would make the case for automatic detection much stronger, because the freelancer still has to decide it is worth raising before the conversation gets easier.