Hitting $15k/mo by purchasing defunct domains and building for pre-existing demand

Less guesswork. More profitable sales.

If you're still judging campaigns by clicks alone, you're probably leaving revenue on the table. Smart brands optimize for sales—not vanity metrics.

Learn 7 battle-tested strategies to increase external traffic, improve Amazon performance, and scale revenue with confidence.

Most founders spend enormous energy creating demand. Erik Aronesty does the opposite: he finds demand that already exists, then quietly steps into it.

Erik is a software engineer with 30 years of experience. He built the first online prime brokerage, built and sold ZoneEdit.com and MoonCostumes, took a detour into bioinformatics research that helped treat his father's cancer, worked on NLP at Bloomberg, published patents and papers in cryptography, and now builds AI-native agentic accounting at Puzzle, where he still works full time.

Everything else he does runs on nights and weekends — and, increasingly, runs itself. He has monthly revenue from more than 30 websites. Some make only $60 a month, but together the long tail and a few larger properties add up to roughly $15k in monthly revenue. AI manages the sites, monitors the logs, and acts on feedback with minimal input from him.

The two biggest earners are OnwardTravel and DirtSignal.com. His newest, more of a labor of love, is PrismClip.com.

Here is the method underneath the portfolio, in his own framing.

The core idea is straightforward and counterintuitive at once. Rather than inventing a new product and hoping people want it, Erik finds defunct companies that still have demand — businesses whose domains are highly cited, highly ranked, and now sitting parked and unused. Then he buys the domain before it drops or gets auctioned, and rebuilds exactly the service that visitors already expect to find there.

The rule he stresses most: build what people expect, not something clever. A parked domain with existing traffic and ranking is a signal of specific demand. If you honor that demand and rebuild the expected service, you inherit the traffic. If you get creative and build something different, Google will de-rank the site — so the discipline of building precisely what people are looking for is not just strategic, it is technically enforced.

He used BizSnipe.com to run this process. It identifies defunct companies with pre-existing demand, tells you which parked domains are easiest to develop into a new service, and suggests what the service should do and how to price it. In Erik's words, it told him what to buy, and he bought it. Sniped names are expensive, but he considers them worth it — the demand is already there.

The portfolio's revenue is concentrated in two very different products, plus a long tail.

OnwardTravel is a "dummy ticket" service. It books a real Passenger Name Record on an airline and issues an itinerary and a verifiable hold — useful for travelers who need proof of onward travel for visa or immigration purposes without committing to a full-price ticket. It accounts for roughly 70% of revenue.

DirtSignal.com web-scrapes about 200 municipal code enforcement and magistrate hearing portals and sells access to the data. Its real value-add is handling Freedom of Information Act requests across many jurisdictions — a process that requires substantial handholding to even begin. It contributes about 20% of revenue. A long tail of one-shot projects makes up the remaining 10%, and PrismClip is still pre-revenue, though Erik hopes it eventually eclipses the rest.

One honest caveat he offers: most of this revenue is not recurring. Travel-related systems never are. It is a portfolio built on volume and replacement rather than predictable monthly retention, and he is clear-eyed about that tradeoff.

Erik's view on technology choices has shifted sharply, and it is worth quoting the shift directly. Six months ago, the debate was React versus Hono, or Python versus Rust. Now, he argues, that choice barely matters. Systems matter more than stack.

The clearest example: he built and deployed a Cloudflare-friendly web proxy in Rust to keep edge workers from getting blocked. He does not know Rust. But he knew performance was the real constraint — running dozens of these sites cheaply in low-memory environments demanded it — so he let AI handle the language and focused on the requirement. He open-sourced the result.

The key tools in his kit:

Cost control is a recurring theme in his choices. LakeQL and lower-cost managed Postgres were crucial for keeping expenses down and avoiding burning savings — and he warns specifically that Cloudflare's D1 fees can destroy a business if you store too much data in it. Total investment across everything was modest: roughly $2k on a sniped domain, and another $2k on advanced AI tooling, hosting, and one-off outreach.

This is the most speculative part of Erik's thinking, and he presents it as a live experiment rather than a settled conclusion. His claim: fixed product definitions are becoming obsolete. Companies used to have target markets and a defined offering — and he is no longer sure why that has to be true.

He has a new experimental site that automatically adapts what it offers based on who arrives. The business model, in his framing, is simply to do what the customer wants. If a visitor asks for something the site does not provide, an agent builds it in about 20 minutes and then provides it.

Whether or not that generalizes, it is a useful window into how he thinks: the product is not a fixed artifact but a responsive system, and the cost of building a new feature has dropped far enough that building-on-demand becomes a plausible model rather than a fantasy.

Because Erik buys pre-existing demand and builds into it, he needs very little marketing. The traffic is already there; the job is to deserve it. That is the entire point of the sniping approach.

He is blunt about the alternative. If you do not snipe a name, you have to build traffic slowly — potentially waiting up to a year for search engines to notice you without tripping spam filters or getting banned from social platforms for self-promotion. In his assessment, there is no good way to promote a genuinely new business right now, and the market is structured to keep new players out.

He is equally blunt about paid ads: a waste. The cost of acquisition is punishing, and spending leaves you with nothing to show for it. Worse, in his experience roughly 90% of the leads Google and Facebook generate are fake — produced by operations milking the algorithm. He points to one of his own sites, a bot-building tool called askthe.bot, which has 2,000 users, of whom he estimates around 1,900 are fake, created to drain his ad dollars with fake leads. From what he has seen, Google and Meta cannot stop this, and bot traffic is now extremely sophisticated.

His defense against it is elegant: charge a dollar. CAPTCHAs help only a little, but bots will not pay. They will log in, check emails, fill out forms, and enter fake addresses — but a real payment stops them, because a bot typically generates only about ten cents in ad payments and will not spend a dollar to earn it.

The only growth tactic beyond sniping that Erik has found genuinely useful is direct messages on Reddit and X. The mechanism is human, not algorithmic: once people realize you are a real person just trying things out, they respond better.

From those conversations, he has gotten bloggers to write articles about his products. It is slow and manual, but in a landscape saturated with bots and fake engagement, being visibly real turns out to be a genuine advantage — the rare signal that cannot be automated at scale.

Erik's headline advice runs against the instinct of most technical founders: do not try to be clever or unique. Find your distribution channel first. Before you build anything, line up ten newsletters that will tell people about your product and ten users who will actually use it.

Beyond that, his operating principles:

From here, Erik plans to focus on his open-source projects, both of which have already proven their worth in his own portfolio.

BoxPDF addresses the sprawling world of PDF handling — it is a memory-bounded, box-model PDF and HTML-to-PDF generator, and he is layering on forms, form handling, and signing, each capability stacking on the box model beneath it. LakeQL is the one he is most excited about: it lets him put a 10GB database on Cloudflare R2 for about a penny a month, then deploy a worker that lets people query it at roughly 10ms latency. That means 10GB of valuable data, with paid access, running for a penny.

Both were major unlocks for how cheaply he can run data-heavy products, and he would like to see other builders innovate on top of them.

You can see the full collection of his projects at q32.com, and explore the two open-source tools at boxpdf.dev and lakeql.com.

Stop guessing on external traffic. Levanta's free playbook breaks down 7 proven strategies driving millions in off-Amazon revenue. Grab the FREE playbook.