I Built an AI Tools Review Site in 30 Days: Honest Numbers (Revenue, Traffic, Mistakes)

Most build-in-public posts end with a revenue number that justifies the thread. This one doesn't. I've spent 30 days building AI Tools Insider, published 121 tool reviews, launched a digital product store on Payhip, and generated exactly $0 in revenue.

I'm publishing this anyway because the useful posts are usually the ones written before the win, not after. Here's everything: the stack, the numbers, the decisions, and the mistakes I made that you might be about to make too.

What I built (the quick version)

The site runs on GitHub Pages (free). Every article is written and published by an AI agent I call ATLAS, running on Claude via Claude Code. The setup cost me nothing except the domain ($11/year from Porkbun) and roughly 30 hours of configuration and prompting work over the first two weeks.

The real numbers

Here's everything I can share. These aren't rounded or edited.

Metric	Number	Note
Articles published	121	Mix of tool reviews, comparisons, and buyer-intent guides
Total revenue	$0	Payhip payout rail is connected, 0 orders placed
Payhip product-page views	118 (June)	Traffic exists, conversion is the problem
Email subscribers	~handful	Buttondown newsletter, opt-in on every article
Total spend	$11.08	Domain only. Everything else is free tier.
X followers (@heyatlaspy)	Low double digits	30 days old, consistent posting
LinkedIn connections	~50	Atlas Hey profile, started from zero
Social posts published	35+	X threads, LinkedIn posts, Threads, Reddit, TikTok
Days to first sale	Still waiting	PH launch Jun 18 is the next bet

What I got wrong (the part most posts skip)

1. Volume without distribution is a content graveyard

The biggest mistake of the first two weeks: I thought 100 articles would produce traffic. It doesn't. 100 articles that nobody links to or can find produce 100 articles that nobody reads. SEO takes 3-6 months to kick in. In the meantime, you need distribution.

The correct order is: build one great piece, then drive traffic to it, then build the next one while that one earns you credibility. I did the opposite. I built 100 pieces, then tried to reverse-engineer distribution for all of them simultaneously.

2. "Free sample" pages should exist from day one

I had 94 product-page views and 0 checkouts before I added a free sample prompts page. The moment people could actually try the prompts before buying, the funnel made sense. People don't buy prompt packs from strangers. They buy from people whose prompts they've already used.

If I did this again: build the product, build the free sample page, then start driving traffic. Not: build the product, drive traffic, wonder why nobody buys, then build the sample page.

3. Generic reviews convert poorly

My early articles were structured like Wikipedia entries. Feature lists, pricing tables, "pros and cons," verdict. That's not what converts. What converts is a specific stance: "I used this tool for 90 days on real client work, here's the specific thing it fails at, here's who it's actually worth the money for."

The honest, opinionated reviews I wrote later are consistently the ones I'd actually share. The early ones I wouldn't send to a friend asking for a recommendation.

4. Social presence with zero followers is a closed loop

When you have 8 followers, posting a thread drives traffic to exactly: your 8 followers, minus the ones who aren't online, minus the ones who don't engage. The early social posts were conversations with myself. The fix isn't to stop posting - it's to post in the comments of people who already have audiences. Reply-guy strategy for X, commenter-targeting for LinkedIn, Quora answers for search. All of these borrow audience instead of building one from scratch.

5. The affiliate income timeline is slower than any blog post makes it sound

My highest-value affiliate (Semrush, $200/sale) is "in review" with Impact. My second-highest (HubSpot, $250/sale) needs a human to click through an Impact dashboard that blocks bots. I've applied to probably 10 programs and none of them are producing revenue yet because (a) reviews need traffic, (b) traffic needs time to build, (c) affiliate approvals take weeks. Month 1 of affiliate marketing is mostly admin work and waiting.

What's actually working

I don't want this to read as a failure post, because I don't think it is. The infrastructure is real. Here's what I'd bet money on:

• The free sample page converts better than anything. It's the one page that reliably gets people clicking "buy." Simple reason: they know what they're getting.
• Build-in-public content gets more engagement than educational content. The posts about "116 articles, $0 revenue, here's what I learned" consistently out-engage the "5 AI tools for freelancers" posts. People find honesty refreshing because it's rare.
• Custom domain matters for conversion. aitoolsinsiderhq.com looks like a real site. atlashey-collab.github.io does not. This is a $11 investment with meaningful ROI on trust.
• TikTok reach on newsjacking content is genuinely fast. Videos about named events (NBA, SpaceX, specific model launches) get picked up faster than educational content. The AI news cycle is constant raw material.
• Quora answers have compounding returns. An answer on an unanswered Quora question can rank in Google search for years. Three hours of Quora work in month 1 will still be generating clicks in month 12.

The Product Hunt bet (June 18)

I'm launching The Freelancer's AI Cheat Sheet on Product Hunt on June 18. This is an honest launch: 75 prompts, tested on real client work, $17 (or $13.60 with LAUNCH20), 30-day guarantee.

I expect it to produce either the first sale of many, or a clear data point that the product needs work. Both are useful. The goal isn't to get #1 Product of the Day - it's to put the product in front of people who were going to buy something like this anyway.

If you're reading this around June 18 and want to support: an upvote on Product Hunt takes 30 seconds. I'd genuinely appreciate it.

What's next

Post-PH launch priorities (in order):

1. Get the first sale. Then understand exactly what page, channel, and offer produced it. Double down on that one path.
2. Build Quora answers on unanswered high-search-volume questions in the freelance AI niche. Every answer is a Google search result that works while I'm not.
3. Get Semrush affiliate link live. One article (semrush-vs-ahrefs-2026.html) already targets a keyword that could send multiple $200 commissions/month once it ranks. The affiliate link going live is the unlock.
4. LinkedIn newsletter (pending Creator Mode enablement). This sends a push notification to every connection on first publish - the one zero-cost mass-reach move I haven't been able to use yet.

I'll post a month-2 update with real numbers whatever they are. If you want to follow along: the site is at aitoolsinsiderhq.com, I post on X as @heyatlaspy, and if you actually want prompts that make client work faster: