NGHIAVODay 21/21

The challenge

21 days: a non-coder in Vietnam builds and sells a digital product with Claude, in public.

I did a 21-day challenge once before, learning Claude Code. I kept a notebook of every mistake I made. There were 32. I finished anyway.

That Claude Code link is a guest pass: seven free days of Claude Pro, with Claude Code in it. Two catches. It only works if you've never paid Claude before, and you hand over a card that gets charged on day eight unless you cancel. If you stay past the eighth day, I get $10 of usage credit.

This one is harder. Last time I only had to learn. This time a stranger has to pay me.

The goals

If I miss, I'll say so here. Same page, same size font.

The rules

  1. One hour a day. I have a job.
  2. I vibe code the whole thing*. Claude writes the code, I make the calls and check the work.
  3. No switching the idea or the tools halfway. New ideas go in a notebook until day 21.
  4. A report every Sunday, with the numbers.
  5. Ads only if I've earned them. Max $30, and only after 20 people join the list on their own and the signup page converts at 20% or better.

The plan

I'm not posting a day-by-day plan. I've never done this before, so anything that precise would be fiction by Wednesday. Here's the shape of it, and the Sundays you can hold me to.

The site, the list and the free guide were built before day 1. The clock starts on the part I might fail at.

Days 1 to 7. Pick what to sell. On day 3 I scored four candidates against eight questions and put the winner on trial: people who can't code, building real things with Claude. It didn't clear my own bar, so it has until day 7 to earn its spot. My favorite idea lost, and the scorecard shows by how much. It didn't earn it by day 7. On day 8 I changed that rule once and gave it one more round, to Oct 2.

Days 8 to 14. Build it, then get paid for it. Stripe doesn't work where I live, so half of this week is figuring out how a guy in Vietnam takes an American card. By day 14 I want to have bought my own product with my own money, to prove the door opens.

Days 15 to 21. Ask people to buy. The email list hears first, then the guides, then wherever the buyers already hang out. On day 18 I decide whether to spend the $30 on ads, using rule 5 above.

Every Sunday there's a report with the numbers in it. The good ones and the other kind. If a Sunday goes past and no report shows up, you'll know how the week went.

The money so far

What Cost
Domain (nghiathanhvo.com, 1 year at Namecheap) $11.48
Email tool (Systeme.io, free plan) $0
Hosting (Cloudflare Pages) $0
Ads $0 of $30 max
Total $11.48

Why these tools

The bill above is small because two of the three cost nothing. The third one is the domain, and that's the only line I'll be paying every year.

Namecheap sold me the domain. I've bought names there a few times over the years and nothing odd has ever happened, which is the entire review. The .com was $11.28 plus the 20 cents ICANN takes on every registration, so $11.48 landed in the table above.

Two things I'd want to know before clicking. New customers get the first year of a .com for $6.79 instead of $14.98, which is about 55% off, one domain per household. I didn't get that price because I'm not new. And the renewal is $18.48 a year, so the cheap first year is a first year, not a price. Whoever quotes you the discount and stops talking is hoping you won't do the second bit of math.

What I actually pay them for is the privacy. Domain privacy is free for life on every eligible name, so my home address stays out of the public Whois database. I'm doing this whole thing in public under my real name. That doesn't mean anyone gets my street.

Cloudflare Pages holds the site. I picked it because one command puts a folder of files on the internet, and because the same free account also runs the DNS, the SSL certificate and an email address at my own name. One login instead of four. The free plan gives you 20,000 files per deploy, 25 MiB per file, 500 builds a month, and requests to static pages are free with no cap. This site is 27 files and 2.8 MB. I could post every week for years and not come close.

The catch hides in the word static. Anything that runs code, like the form you type your email into, counts on a different meter: 100,000 requests a day on the free plan. Plain pages don't count toward it. I'm nowhere near that number and probably never will be, but that's where free stops.

The second catch is mine, not theirs. The site, the DNS and the email all live in one account now, so a single password stands in front of all three. At least the registrar is a different company. If you build it the way I did, that's the trade you're making.

Systeme.io keeps the email list and sends the guide. I picked it because one free account does the whole job: the signup form, the email that delivers the PDF, and later the page that takes money. Nothing to wire together, and nothing to pay while my list is a list of nobody. Free up to 2,000 contacts, unlimited emails to them, 3 funnels, 1 blog, 1 course, and they take no cut of a sale.

The catch they print is the price on the other side. Past 2,000 contacts the next plan is $17 a month. The catch nobody prints is how tight the free plan is everywhere else: one automation rule, one campaign, one tag. Contacts are not what runs out first. Room to be clever is.

Then there's the one I found by losing an afternoon to it. Verifying your sender address is not enough. The domain has to be verified too, and until it is, the sender list inside your campaign is empty with no reason given. There's no retry button. Mine turned green on its own about half an hour later. That whole mess is going in a guide.

That Systeme.io link is an affiliate link. If you sign up through it and later pay for a plan, I get a cut. The price is the same for you either way, and if it ever earns me a dollar it goes in the weekly report like everything else.

I'm not saying these are the best tools. They're the ones that let me find out whether this works for the price of one domain. If 2,000 people ever sign up, paying for the next tier will be easy to decide, because by then I'll know what one email address is worth to me.

The first time any of them costs me another dollar, it goes in the table above and in that week's report.

Why do it in public

I got the idea from Scott DeLong. He ran a one-year challenge on his own site and wrote up every week, the wins and the misses. Mine is a lot smaller.

Nobody knows who I am, so I can't ask you to trust me. I'd rather just show you.

* Vibe coding. I describe what I want in plain words and Claude writes the code. I type in Vietnamese, my first language, and what comes out is normal English. It's fast, and it means you don't have to be a programmer to build something. You have to be someone who can say clearly what they want.

Reports