What I learned launching JobPilot AI in 7 days as a solo indie maker

7 days ago I launched JobPilot AI on Product Hunt as a solo maker from Ukraine. Privacy-first job search PWA — no signup, AI runs locally, free.

Here's an honest post-mortem with real numbers. Not a victory lap.

The numbers

Product Hunt: #71 of the day, 4 upvotes, 0 comments from strangers. Modest.

Traffic over 7 days:

Day 1 (launch):  ~150 visits  (PH spike)
Day 2:            ~70 visits  (PH cliff)
Day 3:            ~50 visits
Day 4:            ~45 visits  (Threads/LinkedIn posts mostly didn't help)
Day 5:            ~40 visits
Day 6:            ~50 visits
Day 7:            ~55 visits  (slowly stabilizing)

So roughly 400 unique visitors in 7 days. Not viral. Not zero either.

Channels (referrers):

Direct                  ~35%   (people typing URL or coming from PH dashboard)
LinkedIn                ~20%   (my own post + 2 reposts from friends)
Threads (Instagram)     ~15%   ← surprise winner, 0 effort
Product Hunt            ~12%
Hacker News             ~8%    (my regular URL submit, NOT Show HN)
dev.to                   ~6%
Twitter / X              ~4%

What worked

1. Threads with zero strategy

Threads referrals tied with HN. I posted one thread about the migration to Cloudflare Workers, got ~15 visits over 3 days. Compared to LinkedIn where I worked hard on copy and got ~80 visits — Threads gave me 15 for nothing.

Takeaway: experiment with channels you've never tried. The cost is 5 minutes.

2. Migrating backend to Cloudflare Workers

On day 2 I noticed Render free-tier was making users wait 30s on cold starts. I ported the whole Python backend to a Worker in 4 hours. 12× speedup.

That migration became my most viewed post (1200 reads on dev.to in 3 days). More than the launch announcement itself. Technical content > product announcements.

3. Privacy-first positioning

Every comment I got mentioned privacy. Not features, not AI — privacy. "Like that you don't collect data." That's the moat I can build on.

What didn't work

1. Show HN

Got shadow-banned. HN auto-flags second submissions from new accounts. The "regular URL" submit on day 1 worked (got 8 upvotes), but my day-2 "Show HN" was invisible to non-logged-in users.

Lesson: don't burn account credit on duplicate submits.

2. Product Hunt forums

Wrote a thoughtful discussion post. PH moderators rejected it twice as "self-promotional." Even after I removed all product links and made it a generic question, still rejected.

Lesson: PH forums are tightly moderated. Not a reliable channel for indie products.

3. Most cold outreach

Emailed 5 newsletter editors (Bytes.dev, JavaScript Weekly, Pragmatic Engineer, Marc Lou via Twitter, Cloudflare DevRel). Zero responses so far. One delivery delay (gmail retried 47h, still nothing). One bounce (CF's developers@ doesn't exist).

Lesson: cold email response rate at indie-product stage is ~1-5%. Need 20 sends minimum to hope for 1 reply.

What surprised me

1. Day-by-day engagement is invisible

I had no idea what users were doing for 6 days. Cloudflare Web Analytics shows "70 visits from Bulgaria today." That's it. I had no clue:

• Did anyone click the AI feature?
• Did anyone fill the resume?
• Did the wizard work or did they bail at step 1?

I shipped 8 features completely blind. On day 7 I finally added a privacy-friendly event tracker (Cloudflare KV + tiny /api/event endpoint). First 24h showed: mostly me. But now the infrastructure is there.

Lesson: add analytics on day 1, not day 7.

2. Feature creep is intoxicating

Over 7 days I added: live search across 11 boards, AI scoring, skills breakdown, red flags detection, reply generation in 5 styles, resume tailoring, interview coach, batch URL analysis, smart resume hints, onboarding wizard, 5 languages, PWA install, Kanban tracker, PDF export, donation link.

Each one felt necessary at 2am. Looking back: probably 60% of these are unused. I won't know until I see the analytics in 2 weeks.

Lesson: every feature has a maintenance cost. The first version should be 1 thing done well, not 8 things done OK.

3. The hardest part is patience, not coding

After launch day, the temptation to keep shipping is enormous. "Maybe one more feature will spark growth." Wrong. Growth comes from time (Google indexing, directories crawling, word-of-mouth) and outreach (cold emails, comments, partnerships), not code.

I'm a developer. Coding is comfortable. Marketing is uncomfortable. Guess which one moves the needle for indie products?

What I'm doing for the next 2 weeks

1. Feature freeze. No new code beyond bug fixes.
2. Directory submissions: Tier 1 done (4/8 free), Tier 2 starts now (BetaList, MicroLaunch, TinyLaunch already in queue).
3. One cold email per day. Targeted, personal, not blast.
4. Wait for analytics signal — at ~100 daily visits I'll see real patterns.

Numbers I'm watching

Goal by July 1:   100 DAU sustained
Cost so far:      $0  (Cloudflare free tier)
Time so far:      ~120 hours over 7 days
Tip jar:          opened today (Ko-fi)

Try it

Live: jobpilot-ai.pages.dev — search remote jobs, AI scores your match against your resume, none of your data leaves the browser.

If you're building something solo, drop a comment. Always happy to compare notes.

— Dmytro Oriekhov (solo, Ukraine)