12 questions readers keep asking (Spring 2026)

Twelve questions that keep showing up in mail, DMs, and Threads — answered as honestly as I can.

Since launching the first tool on bal.pe.kr in 2024, a recurring set of questions has arrived through email, DMs, and Threads comments. Rather than answering them ad-hoc, here are twelve of the most common ones in a single place. The answers are short, honest, and lean on real numbers from the 119-site fleet rather than generic advice. Some of the context is Korean-specific (Naver search behaviour, KRW infrastructure pricing, Korean cultural patterns around side projects), but I've translated and expanded each answer so a non-Korean reader can use it directly.

Q1. Did you really build all 119 sites alone?

Yes. Code, design, copy, and infrastructure are all handled by one person. The caveat is that more than 70% of the tools sit on top of shared components, with site-specific differences injected via props. That is why an average new site reaches a live URL within four to six hours. Roughly twenty of the tools are genuine 0-to-1 builds; the rest are variations on those twenty.

To be precise about "alone": every code and infrastructure decision is made by one operator. Some early visual mocks are generated with design assistants like Stitch or v0, but the actual structure and policy of each tool is a human decision.

Q2. How much ad revenue do you actually earn?

Honestly, not much yet. Combining all 119 sites, it took about six months to clear the first AdSense payout threshold of USD 100. The cumulative year-one total sits somewhere between USD 200 and USD 400. The top five sites generate almost all of it; the rest are effectively zero. Ad revenue is a side note, not the motivation. The real reason for building is described in Q9.

A general warning to fellow indie builders: prioritising ad revenue too early distorts operational decisions. The tool with the best AdSense RPM is rarely the tool with the best user value. Our internal priority order is user value → search traffic → ad placement, in that sequence.

Q3. Was this built by AI?

Code is written with heavy assistance from tools like Claude Code, but every decision about what to build, why it should exist, and what message it should send is 100% the operator's. Earlier in the year a few posts drew accusations of "AI smell" because the sentence structures were too uniform. The fix was to deliberately vary tone and structure across posts.

Every long post now contains at least one detail that only a human could write — a family schedule, a 3 a.m. CloudFront recovery, a specific GA4 number from a specific date. When AI writes the average well, the only remaining value of human writing is in the unaveraged details.

Q4. Which site is the most popular?

It changes every quarter. As of May 2026, the top tools are yebang (Korean vaccination reminder), firekr (Korean FIRE calculator), kmbti, and the saju family. The ranking shifts depending on whether you define "popular" as search inflow or viral reshare.

By cumulative search traffic, saju and kmbti lead. By short-duration viral spikes from Threads / KakaoTalk shares, yebang and firekr lead. Recognising these as two different metrics is one of the small operational realisations that arrived only after a year.

Q5. What was the very first tool you built?

I recommend starting with something you personally use every week. My first was pogalwage, a comprehensive-wage-validity calculator that I built because coworkers kept asking the same question over lunch. Tools you do not use yourself end up empty in the details, and users do not return.

The first tool is mainly about motivation, not market size. Tools you use weekly accumulate small irritations naturally, and those irritations become v2 and v3. Tools you do not use stop receiving attention right after launch.

Q6. When do you find the time?

One hour before work, one or two hours after the baby goes to sleep, two or three hours on a weekend morning. Around ten hours per week, total. That is not a heroic budget. It is exactly why infrastructure standardisation and the discipline of not building things matter so much.

On days with major incidents at the day job, no side code gets written. Accept this and rest that day. Trying to be consistent every single day breaks within a quarter.

Q7. Aren't domain costs expensive?

No, because all 119 sites are subdomains of a single registered domain, bal.pe.kr. The total cost is one annual renewal + free ACM + a small Route 53 line item. If each site had its own domain, the bill would be roughly USD 800–1000 per year. Consolidating to one domain from day one was the single largest cost-saving decision of the project.

There is a trade-off worth noting. Any domain-level event — an SEO penalty, an AdSense policy review, a DMARC mismatch — affects all 119 sites at once. For a solo operator the savings still outweigh the risk, but a company running a similar fleet might choose differently.

Q8. Do you target Korea or international users?

Both, with 104 Korean-market tools and 34 international-market tools, with some overlap. International sites work surprisingly well if the English metadata and llms.txt are clean — the traffic is slower to build but more durable. The viral one-line-recommendation pattern that exists in Korea is mostly absent abroad, so global tools accumulate steadily rather than in spikes.

The decision rule for whether to build an international version is straightforward: can I personally verify the English context for this topic, and is the tool independent of Korean regulations? Korea-specific tools (tax, labour law, vaccine schedules) do not translate, no matter how well the English copy is written.

Q9. What's the real motivation?

Three honest answers.

First, the need for a "small unit I can finish this week" outside the day job. At work, decision cycles take a quarter because of alignment overhead. A micro-tool finishes a full cycle in four to six hours. That difference matters more for mental health than I expected.

Second, Korea has a surprising number of empty slots for small Korean-context tools — calculators tied to specific Korean regulations, schedules tied to Korean institutions, name-style services tied to Korean culture. Global tools cannot substitute. Someone needs to build them and the marginal builder might as well be me.

Third, a multi-year bet that running this fleet would build instincts I cannot otherwise practice — capacity planning, infrastructure consolidation, public writing, AEO experimentation. Those instincts already feed back into the day job.

Q10. The decision you regret most?

Creating a per-site CloudFront OAC and Function on every early site. The AWS bill grew sixfold before I noticed, six months in. The cure (shared bal-pe-kr-rewrite-2 Function + shared static-shorts-oac) is now standard, but the lesson cost real money. Infrastructure choices made early are compounded by every site you launch afterwards.

The second regret: missing English metadata on early sites, which forfeited six months of international organic traffic. Today a lint rule fails the build if any locale's title or description is empty, so the regret is now structurally impossible to repeat.

Q11. Where do new tool ideas come from?

About 80% come from genuine personal or family annoyances. After our first child arrived on 27 May 2025, a wave of parenting-context tools followed naturally — weaning-stage guides, infant developmental milestones, vaccine schedule helpers. The remaining 20% come from search-trend data and recurring questions on Korean community sites.

The filter between idea and build is a single question: "would I use this myself weekly?" If the honest answer is no, the tool is not built, no matter how strong the search demand looks. Tools that the operator does not personally use end up shallow in the details, and shallow tools rarely retain users on a second visit.

Q12. One piece of advice for someone starting micro-SaaS?

Finish and ship the first tool. There are far more people who started five tools and shipped zero than there are people who started one and shipped one. Shipping is not the end of the work — it is the line that separates real builders from aspiring ones.

Buy a domain. Put a one-page tool on the internet. Until that page exists, no one can cheer you on. Loneliness is part of the first tool's launch, and only the people who pass through it once go on to build the next ninety-nine.

Three good questions held over to the next round

A few questions arrived that deserve longer, slower answers. I'm parking them for a future Q&A.

"If you scale down launches next year, what fills the time you free up?"
"How did you explain the project to your family when no site had any traffic yet?"
"Is there a monetisation model that works for Korean users beyond AdSense?"