TL;DR: The old test still works. Before you quit, incorporate, or buy a truck, put a real storefront on the internet and spend a little money to see if strangers will call. In 2026 that storefront does not take a month. WebZum builds a multi-page site — logo, photos, chatbot, a tappable number in the header — in about five minutes. Then you can point a small Google Ads budget at it. Google lets you set the amount; their own help says start small and gives a $10/day example that caps a month at $304. A $20–$100 burst is a test, not a launch. If the phone can start moving, then you do the LLC, the bank, the resignation. If it stays quiet, you tweak the offer or walk away. (Disclosure: WebZum is our product. Every number below is cited to the source that published it.) See a finished site free.
Verified August 26, 2026 against the linked sources.
The roofer who still works for someone else
Imagine you swing a hammer for a boss. He takes the job, you do the work, you get a slice. You already know how to roof a house. What you do not know is whether your name on a site, in your town, would get a stranger to tap Call.
Most advice says the opposite of a test. File the LLC. Open the bank account. Write a ten-page plan. Quit when the savings hit six months. That is how you start a business once you already have demand. It is a terrible first move if you do not.
The SBA’s own market-research page puts the first question as demand: “Is there a desire for your product or service?” and says research “lets you reduce risks even while your business is still just a gleam in your eye” (SBA, Market research and competitive analysis). Their startup-cost list still includes licenses, insurance, “Making a website,” and advertising. You can pay those in the wrong order.
The 2009 playbook: a page, a button, an ad
Software people learned this the expensive way.
In 2009 Eric Ries told Venture Hacks that a minimum viable product is often “an ad on Google… Or a landing page” you can build “in a day or a week.” He used a product IMVU actually built: a Kerry-vs-Bush debate avatar. Two weeks of work. They sold zero, even when they gave it away. What they should have done, he said, was start with a page that promised the product, buy the AdWords they were going to buy anyway, and watch the click-through. “0% of people would have clicked through… you don’t need to go through the effort of building out the product” (Venture Hacks, March 23, 2009).
Joel Gascoigne did the version that worked. Before he wrote Buffer, he put up two pages: one that explained the idea, one that collected an email. Then he stuck a pricing page in the middle to see if people would click a paid plan. He tweeted the link. Then he built the real app “in evenings and weekends over a period of 7 weeks.” First paying customer “within 4 days of launching” (Buffer, February 16, 2011).
Same idea, two endings. The test is cheap. Building first is not.
A local service business is not Buffer. You are not faking a product that does not exist. You are the product. The van, the skill, the Saturday you already have. The 2026 test is a storefront plus a little paid traffic — not a fake door for software.
What changed in 2026 (and what did not)
Two clocks collapsed.
The site. Wix’s own blog still times a small-business site at 3–7 days on a template and 1–2 weeks in the editor (Wix). Their 2026 State of Websites report, from 310 million users, puts average time-to-publish at 4 days, down from 8. GoDaddy still says a builder site is “a few hours to a day” and a designer “a few weeks” (GoDaddy). Durable’s own page is titled “Create a Website in 30 Seconds” — a draft, then an editor. WebZum is the fastest finished shop site we know of for this job: about five minutes, before a card. The build-time post has those receipts.
The traffic. You do not wait for Google to index you. Google’s own docs say a new site can take days to weeks just to get noticed, and ranking is slower still (how long Google takes to index). Ads skip that line. Google’s help is blunt: “When you advertise with Google Ads, you link your online ads to your website.” If you have no site they name Squarespace, or they let you dump clicks onto a Google-hosted Business Profile page (Google Ads Help). A real site you control still beats a rented Google page — and you need somewhere for the stranger to tap Call.
What did not change: most tiny firms are not already doing this. Census Bureau BTOS data from December 2025 to May 2026 shows overall business AI use hovering between 17% and 20%, and “less than 20% of firms with four or fewer employees reported using AI” (Census, May 26, 2026). The 2026 stack is available. It is not the default.
And “no market” is still how things die. CB Insights looked at 431 VC-backed shutdowns since 2023. “Ran out of capital” tops the list at 70%, but they call that the last chapter. The root they name is “poor product-market fit (43%)” — and two-thirds of those were early companies that “never found a market” (CB Insights). That is venture-backed software, not a cleaning route. The lesson still travels: do not quit for a market you have not poked.
The 2026 test: a site, then $20–$100 of ads
Do this while you still have the job.
- Put up a real site with your name, your services, your town, and a number a phone can tap. Not a “coming soon.” Not a Facebook page. The header has to start the job (click-to-call).
- Buy a small, capped burst of Google Search ads aimed at the words a homeowner types — “roofer near me,” “house cleaner in [city],” “emergency plumber.” You set the daily budget. Google: “The amount of your budget is entirely up to you” (Google Ads Help). Their example: $10/day can bill up to $304 in a month ($10 × 30.4), and a single day can spend up to 2× the daily number (budgets overview). Their own tip is “Start small” (Google Ads Help). There is no Search-minimum we can print from those pages. $20 or $100 is your test, not Google’s floor. WebZum’s optional same-day ads start at a $20 spend on our side. That is a product number, not a Google rule.
- Watch the phone for a week. Not your cousin. Not a like. A stranger who found you because they needed the job done.
What that money can buy, if you believe industry averages and not a promise: LocaliQ / WordStream’s June 2026 search benchmarks (their customer campaigns, not Google’s official prices) put all-industry cost per click at $5.42 and cost per lead at $66.69. Home & Home Improvement — the bucket a roofer lives in — is $8.33 per click and $90.92 per lead. Personal services: $7.17 / $54.60. Beauty: $4.62 / $39.25 (LocaliQ, updated June 1, 2026). A $100 test can buy a handful of clicks. On home-improvement averages it is roughly one lead’s worth of spend. It often buys none. Can, not will.
What the test can show — and what it cannot
| This test can show | This test cannot replace | |
|---|---|---|
| Demand | Whether a stranger will tap Call | A license, bond, or insurance |
| Offer | Whether “roofer in [city]” gets clicks | The skill, the truck, or the quote |
| Timing | Same-day ads vs waiting to get indexed | Organic rank |
| Money at risk | A daily budget you cap | Payroll, LLC fees, or quitting |
| If the phone is quiet | Tweak the offer or walk away | Proof the trade is dead |
| If the phone moves | Then incorporate, bank, resign | Instant full-time income |
| How you run it | A WebZum site in about five minutes, then ads you cap | A career change on a feeling |
Squarespace’s March 2026 home-services guide is the honest vendor line: “Trades like plumbing, HVAC, and electrical work may require additional licensing. Check your local regulations before advertising your services.” And: “Insurance isn’t optional in this industry. It’s essential from day one” — general liability, commercial auto, workers’ comp if you hire (Squarespace). A website plus ads is a demand test. It is not a contractor’s license. Roofing and contractor rules vary by state. Check yours. We are not your lawyer.
Same for the day job. Read the contract. Non-competes and moonlighting clauses are real. Do not use the boss’s truck, the boss’s phone, or the boss’s customer list. This post is not employment advice.
Do the paperwork after the phone moves
LegalZoom, which sells LLCs, says the quiet part: “You don’t need an LLC to run a small business.” If the thing “has only made a small amount of money or is in a low-risk industry, you might consider using the money it would cost to set up an LLC on revenue-generating activities instead” (LegalZoom, updated December 19, 2024). They also flag construction, real estate, and healthcare as higher-risk. A roofer should not hear “skip the LLC forever.” Hear: test first, entity when the calls are real. Opposite of our April stack, which is incorporate-then-market. That post is for after this one works.
If the phone moves: then the LLC, the bank, the insurance, the resignation. If it does not: change the offer, change the town, or keep the job. Your loss is a Saturday and a small ad bill — not a career.
SBA Advocacy’s 2024 FAQ (citing BLS Business Employment Dynamics) is the long clock: from 1994–2021, 67.9% of new employer establishments survived two years, 49.2% five years (SBA Office of Advocacy FAQ). Most small businesses have no employees at all (81.9%). A side test is not an employer firm. Do not read those survival rates as your personal odds. Read them as: a lot of things that get incorporated never had a market.
What WebZum does with this
You type the business name. The system researches the shop, writes the pages, makes the images, builds a logo, puts one tappable contact in the header, and adds a chatbot that knows the services. About five minutes. You see it before you pay. Starter is $19/month. Pro is $34/month (custom domain, registered and paid by us, branding off). The price post has those receipts.
Then, if you want the test, you can launch a same-day Google ad from the same conversation — copy written from the site, keywords picked for the town. Spend starts at $20 on our side. You manage it by talking: “target this zip more,” “change the wording.” Your phone can move. We will not say it will.
That is the 2026 update to Ries’s landing page. The page is a real local site. The ad is still an ad. The point is still: do not quit for a maybe.
FAQ
How do I test a business idea before quitting my job? Put up a real website with a number people can tap. Spend a small, capped amount on Google Ads — $20 to $100 is a test, not a launch. Watch the phone. If strangers call, then do the LLC and the resignation. If they do not, you lost a little money, not the job. See a site free.
How do I start a side business while working full time? Nights and a Saturday. The site is about five minutes on WebZum. The ads run while you are on someone else’s clock. Do the licensed, insured work on your own time. Read the employment contract first.
Should I quit my job to start a business? Not on a feeling. Quit when the test has already produced calls you can handle after hours — or when you have a different kind of runway. This post is the test, not the resignation letter.
How do I start a service business on the side? Same test. Cleaner, roofer, plumber, salon, landscaper: the “product” is you. The site is the storefront. The ad is how a stranger finds you this week. Licenses and insurance still apply before you advertise in a lot of trades (Squarespace).
How do I start a business without quitting my job? You start it as a test, not as a resignation. Website first. Tiny ad budget. Paperwork if the phone moves. That is the opposite of “go all in on Monday.”
Can I test a business idea with Google Ads? How much does it cost? Yes. You set the daily number. Google’s $10/day example caps a month at $304. Industry averages (LocaliQ 2026) put a lead around $40–$90 depending on the trade. A $100 burst can produce a call. It is not a promise.
See the storefront before you quit
Type your business name. See the finished site free. If the number is wrong, tell it. You do not owe a card to look — and you do not owe your boss a two-week notice to find out whether anyone would have called.