Side Hustle Lab: July 20-27, 2026; the $6 spreadsheet that kept paying

The clearest low-barrier result from July 20-27 was a $6 spreadsheet that kept selling after its maker stopped working on it. The biggest payout came from a much harder SaaS build: more than $10,000 in July. In between, a bookkeeping service reached about $2,800 a month, while several fresh launches still had no revenue or no customer signal.

All figures below are self-reported unless stated otherwise and were not independently audited. The coverage window is July 20, 2026 at 09:00 through July 27, 2026 at 09:00 in the channel's display timezone, UTC-05:00. Ranks prioritize how quickly a full-time employee could test the model, then reported payout, then evidence quality. Currency is left as reported.

A spreadsheet for sourdough bakers is the week's most useful small bet. The maker built it in one Sunday afternoon to solve a family problem: tracking starter feedings, flour types, room temperature, hydration percentages, and bake notes. It started as a practical Google Sheet, not a product strategy. 1

The first distribution step came from the maker's mother, who showed it in a local baking Facebook group. Members asked whether they could buy the file, so the maker cleaned up the formatting, added a short PDF guide, and put it on Gumroad for $6. The link then spread through other baking forums and subreddits. The author says the sheet sells 15-20 copies a month and brought in about $114 after platform fees last month. The post does not name the calendar month, so this is a recent monthly self-report, not a claim about a stable annual run rate. 1

The startup cost is not disclosed. That matters: the $6 price is what customers pay, not proof that the maker spent nothing. What is clear is the sequence. Solve a recurring problem for one person, let that person demonstrate it in a niche community, then charge for the cleaned-up version. A 9-to-5 reader who already works comfortably in Sheets or Excel could test this pattern over a weekend without pretending that every spreadsheet has a market.

Kayli Robles used prior accounting experience to start Robles Bookkeeping in Washington state. A Liberty Times report published on July 26 says she found the idea after seeing a Facebook ad for a bookkeeping-business course, then used the course to fill gaps in bookkeeping software and customer acquisition. 2

Her disclosed cash path is unusually concrete. She spent $10 on a business license and $12 on a domain. She paid $280 for insurance after securing the first paying client, making the reported spend at least $22 before that client and $302 once insurance was added. The course price is not disclosed. 2

It took about two months to find the first customer, who paid $400 a month. Two months later, Robles took a $4,000 bookkeeping cleanup project and gained a second fixed client worth another $800 a month. The following month, recurring service and project work reached about $2,000 a month. The report says the business now averages about $2,800 a month. 2

This is a low-capital model with a real skill gate. The cost is not the hard part. Accounting competence, trust, and the ability to sell a service are. The article is a secondary report that quotes Robles rather than her original post, so the figures should be treated as self-reported interview numbers, not audited revenue.

For someone with the underlying skill, the first move is narrower than "start a bookkeeping firm": learn the software required by a target customer, define one service such as monthly close or cleanup, and sell the first engagement before paying for a larger tool stack.

An Australian entrepreneur set out to test whether a service business could start for less than $50 without using personal contacts. The reported setup included a business-name registration, a website, a Facebook page, and an AI-made video. The first sales test was 30 cold calls. It produced no meetings and no leads. Revenue was not reported. 3

The post was written as a live update, not a final business autopsy. The author suspected the weekend timing, the prospect list, or the target market, and planned another 50 calls. That uncertainty is the useful part. A $50 setup does not validate a service; it only makes it cheap enough to test the first channel.

The barrier here is low in cash and high in rejection tolerance. The first move was concrete, but the offer, list quality, and call script were not detailed enough to judge. Keep the outcome at "0 meetings and 0 leads," rather than turning one bad calling day into a verdict on cold outreach.

A Reddit user reported buying a pre-built pet niche website for $199 seven months earlier. The site already had content and Amazon affiliate links. The buyer spent a few hours adding internal links and then says they made no new content, backlinks, ads, or social posts. 4

The reported total from January 1 through July 26, 2026 was $4,010.07: $3,846.09 in commissions, $57.25 in bounties, and $106.73 from Creator Connections. The post also reports 33,564 clicks and a 9.27% conversion rate. The author says the site was still producing commissions daily, but does not disclose a monthly revenue figure. 4

The $199 purchase is the useful constraint. This was not free passive income, and the reported result came from an already-built site with an existing content base. It is a mature case reported this week, not evidence that a new buyer can purchase any pre-built site and reproduce the result. The missing variables are the site's age, rankings, content quality, and the buyer's actual payback period.

For a reader considering this model, the first step is not buying a site. It is asking for traffic, earnings, and ownership evidence before paying, then treating the purchase price as at-risk capital rather than a subscription.

TinyPDF is a small tool for compressing a PDF to an exact target size while preserving page count, page size, and layout. The maker validated the problem with five designer friends, built the core logic and interface over 12 days, tested it with 10 users, and launched after two weeks. The first 10 days produced about 200 unique visitors, 45 compressed PDFs, and 12 thank-you emails, but exactly $0 in revenue. The tool is free forever and processes files locally in the browser. 5

That is a real usage signal, but not a payment signal. The maker solved a personal annoyance and got people to use the result. They did not yet show that users would pay, that a premium feature would be attractive, or that the free tool could acquire customers at a sustainable cost. Startup costs were not disclosed.

The debt payoff planner reached an earlier failure point. Its maker built it over a couple of months on nights and weekends around a day job. Users enter balances and interest rates, choose the snowball or avalanche method, and receive a payoff calendar. After launch, revenue was exactly $0, with barely any signups and no conversions. Finance subreddit posts brought some traffic, but users did not stay long enough to create accounts. 6

These are different zeros. TinyPDF had usage and gratitude but no monetization test. The debt planner had weak retention before pricing could be evaluated. In both cases, the next move is a customer conversation or a sharper offer, not automatically more features.

BillHawk launched on July 27 as a native Mac time tracker and invoicing app for freelancers. The maker set a one-time price of $29, with the price scheduled to rise to $49 after purchase number 200. On launch day, the maker had not yet disclosed revenue. 7

The reported fixed infrastructure cost is $0 a month because the site, licensing API, storage, analytics, and payments use free tiers or usage-based services. Stripe takes about 3%, and the app is sold outside the Mac App Store to avoid its 30% cut. 7

That does not make the experiment easy. The maker still needed Swift development, a notarized macOS distribution path, license activation, payment webhooks, and a way to sell to the right audience. The product's price is known; demand is not. This is a good example of why fixed infrastructure cost and startup barrier are separate fields.

A 9-to-5 reader can copy the pricing question, namely whether a one-time purchase is more credible for a small utility than another subscription. They should not copy the assumption that free cloud tiers erase the work required to ship and support a desktop product.

The week's largest reported payout came from SocialCrawl, a second SaaS from a developer whose first product failed badly. The maker says the second product crossed $10,000 for July. Costs were not disclosed. 8

The reported change was distribution before product. The maker started with a problem they personally had, validated it through conversations and community research, then built an audience where potential users already spent time. The post reports 1,700-plus followers on X, 5,000-plus on Threads, and about 30% of traffic coming from ChatGPT. The maker also says fast replies, fixes, and implementation helped keep users moving. 8

This is a strong payout signal with a weak beginner fit. The audience, product, support load, and software-building ability took time to assemble. The post also says paid ads, promoted content, and influencers performed worse than the maker's existing organic audience. That makes the lesson specific: distribution was an asset built before launch, not a free channel that appeared after the product was finished.

The first failed SaaS is part of the evidence, not just backstory. The maker attributes it to an unclear customer profile, building something that seemed cool, and having no solid distribution channel. The second product did not remove those risks; it addressed them earlier.

Start with the smallest thing that can be shown to a real niche. The sourdough sheet reached buyers because it solved a specific problem and appeared in a community where that problem was already visible. The bookkeeping case used an existing skill and a narrow service before spending on brand upgrades. 12

If you are testing a service, treat the first 30 calls as a channel experiment. Record the list, offer, timing, and response. The Australian report got no meetings or leads, but it still produced a measurable next question: change the list or the offer before making another 50 calls. 3

If you are building software, separate usage from revenue. TinyPDF had users and thank-you emails but no payment. The debt planner had weak signups and no conversions. BillHawk had a clear price but no reported sales at launch. Those are three different tests, and the next action should match the missing evidence. 567

The practical order is straightforward: sell a narrow service if you already have the skill, package a small file if you can reach a specific niche, and publish the distribution test before spending months on a product. The $10,000 SaaS result is real as a July self-report, but the $114 spreadsheet is the easier experiment to copy this week.