ran 20 users on a form and a spreadsheet for six weeks - when do i actually have to build the app

Volume is almost never what kills the manual version. What killed mine was the promise. I told people 'back within an hour' and then had a Tuesday where I was in a dentist chair at 11am and three people got their file at 6pm. That is the moment you automate - not because you cannot do the work, but because you cannot do it on a schedule a stranger can plan around.

Order I would go in: automate only the path that carries the promise (intake, processing, delivery). Billing stays manual, onboarding stays manual, support obviously stays manual. My 'real app' for the first four months was one worker script and one table, and the whole admin side stayed in a spreadsheet for another year after that.

Do the per-customer minutes, not the total. 40 minutes for 20 people is 2 minutes each. The question is whether that 2 keeps falling as you get better at it or parks. Mine went 6 -> 3 -> 2.5 and then sat at 2.5 forever, which is the actual signal: you have squeezed everything the human can squeeze and the rest only comes out with code.

Also price your own hour honestly. 40 min/day is about 13 hours a month against $45 of revenue. That is fine as a learning exercise, just do not tell yourself it is a business decision when it is a schedule decision.

One thing worth building early even while the rest stays manual: the cancel and refund path. Manual billing is where disputes come from - someone emails you to cancel at 11pm, you are asleep, they get charged in the morning, and now it is a chargeback instead of a refund. A payment link plus the hosted customer portal is about twenty minutes of work and it removes the only genuinely expensive failure in a concierge setup.

Politely disagreeing with the 'stay manual as long as possible' line above, because it depends what you are selling. If the value is judgement, manual teaches you the product. If the value is speed and self-serve, manual teaches you very little and actively hides the failure mode: people will happily wait four hours for a human and will not wait four seconds for software. I ran a concierge version of a scheduling tool for two months, everyone was delighted, and the automated version got taken apart on exactly the dimension the manual one was papering over.

'Clean csv back' sounds like speed to me.

I walked this exact path four months ago so I will over-share. Form, spreadsheet, then no-code automations, then a real app. Two things nobody warned me about. First, automation runs are metered per plan and I burned a month's allowance in nine days the week a few people started submitting twice - go look at what your plan's cap actually is before you lean on it. Second, migrating off is much easier if the spreadsheet never becomes the source of truth for anything with money attached. I had payments in one tab, users in another and a fuzzy join on email address, and reconciling that later ate a whole weekend. Let Stripe be the truth about money and let the sheet be a work queue you are allowed to throw away.

Write code when the same manual fix shows up three times in one week. Not before. Everything else is you wanting to open the editor.