Last updated: August 22, 2026
A quarter of Y Combinator’s Winter 2025 batch shipped products whose codebases were 95% AI-generated.
That part’s solved.
You can grab a domain, wire up a Carrd landing page, and validate demand before writing a line of code.
But one decision decides whether you’re running a business or a liability. It has nothing to do with your tech stack.
In This Article
- What a 2026 Launch Actually Costs: The real monthly line items, with current prices — and why the cheap stack is the easy half of the problem.
- How to Interview Customers Without Getting Lied To: The question sequence that separates real pain from polite encouragement, and why skipping it kills most SaaS ideas early.
- Who Legally Sells Your Software: The billing choice that quietly turns founders into tax collectors across dozens of jurisdictions.
- Why the Merchant Fee Hits Cheap Plans Hardest: The premium works out regressive, and the arithmetic changes which platform suits your price point.
- The Seller Test: Three questions that resolve to one billing platform, with the rate each one charges as of August 2026.
What It Actually Costs to Start a SaaS in 2026
A solo SaaS launch in 2026 runs on roughly $60 to $100 a month.
That covers an app builder, a landing page, a form tool and an email list. Hosting, authentication and a database stay on free tiers.
The app builder is the only real monthly cost. Replit Core lists at $25 a month, dropping to $20 billed annually.
Bubble’s Starter plan runs $32 a month for web, or $29 on annual billing.
A domain is the smallest line on the list.
Then add billing. You pay a percentage of revenue, so that line stays at zero until somebody actually buys.
No developer. No office. No pricing strategy to solve yet.
The bottleneck everyone warns you about doesn’t live here anymore.
Why the Standard Six-Step Playbook Is in the Wrong Order
Most six-step guides still put “build your product” at step one. That sequence made sense when building was the expensive, slow part. It isn’t anymore.
Every guide that teaches the old sequence front-loads the wrong risk.
Audience belongs before product. You need people to email before you need code to ship.
Start collecting addresses now, with nothing to sell yet. MailerLite’s free plan holds that list — check the ceiling first.
On 16 June 2026 the free plan dropped from 500 subscribers to 250, and monthly sends from 12,000 to 2,500.
Older guides still quote 1,000. That number has been wrong twice over since September 2025.
The second misplaced step is billing.
Most guides treat picking a billing platform as a launch-week detail.
It’s a legal decision. It sets who owes sales tax, VAT and GST on every sale.
How These Tools Were Chosen
Nothing here was tested end-to-end as a live SaaS with paying subscribers. This is a research post, and the selection reflects that.
Every price below came off the vendor’s own pricing page in August 2026.
None was lifted from an older roundup. Where a vendor showed monthly and annual rates, both appear.
Tools qualified on one test. Does a solo founder need this before the first paying customer?
Anything that only matters at team scale was left out.
Two figures in circulation failed verification and were cut rather than repeated. Both are named later, with what replaced them.
Validate Before You Build
When the build costs under $100 a month, the only expensive mistake left is building something nobody wants.
That’s a validation problem, not a coding problem.
Before you write a line of code, talk to people who’d actually pay, then put up a page to count who means it.
Talk to People Who Might Actually Pay
CB Insights analysed 431 venture-backed companies that shut down since 2023. Poor product-market fit was cited in 43% of failures, second only to running out of cash.
Note the date on that: March 2026. The “42% no market need” figure everyone quotes comes from a 2014 post-mortem of 110 startups.
Same lesson, fresher evidence, larger sample.
Book 15-minute calls with people who have the problem. A free Cal.com booking page handles the scheduling.
During those calls:
- Ask what they currently do about the problem
- Ask what they’ve already tried and dropped
- Ask what it costs them today, in money or hours
Don’t pitch. Ask about their life.
Rob Fitzpatrick’s The Mom Test is 136 pages on exactly this, and it’s the shortest useful thing you’ll read before launch.
If five strangers describe the same pain unprompted, you have signal.
If they don’t, you’ve saved yourself months.
Put Up a Page and Count the Signups
One page, one promise, one email field. Tally.So handles the intake free, with unlimited forms.
Skip pricing details for now.
Right now you’re counting signups, not dollars.
Building It Without Code, and Where That Breaks
No-code holds up fine at low volume, then bills you per task once usage climbs. The break point arrives when growth does.
At low usage everything feels manageable. Onboarding flows, alerts and admin tasks all run cheap.
Then growth hits. Same tasks, more volume, and per-task pricing starts working against you.
- Each additional user adds incremental cost
- Each automated task multiplies that cost
- Revenue grows linearly; cost can grow faster
Make.com now bills in credits, where one module action equals one credit. The free tier covers 1,000 a month.
Six months of running my own Make.com scenarios taught me the ceiling arrives early.
Mine burned the free 1,000 credits on one tidy-up workflow, before anything touched a customer.
That’s the honest limit of no-code ops. It’s cheap to start and priced per action forever.
None of this touches the billing decision — that’s next, and it’s the one that’s expensive to reverse.
Prices verified against each vendor’s official pricing page in August 2026: Carrd Pro, Tally pricing, Cal.com pricing, Make.com pricing and Flippa pricing.
The Billing Decision No Starter Guide Includes
Whoever bills your customer is the legal seller of that software.
Choose Stripe and that’s you, owing tax wherever you cross a threshold. Choose a merchant of record and the platform takes that liability.
Every ranking guide tells you to “pick a payment processor” like it’s a checkbox next to logo design.
It’s the most expensive decision in this whole process, and it locks in the moment your first customer pays.
What Merchant of Record Means for You
A merchant of record buys your software and resells it to the customer. The invoice carries its name, not yours.
It collects payment, remits the tax, and files the returns.
Here’s what that costs, using each platform’s published rate in August 2026.
Rates read from Paddle’s pricing, Lemon Squeezy, Polar’s fee schedule and Stripe Tax pricing, August 2026. Premium column is the difference against Stripe at 3.4% + 30¢.
That convergence on 5% + 50¢ is new. Polar’s cheaper 4% + 40¢ rate closed to organisations created on or after 27 May 2026.
Now look at what that fixed 20¢ does to a small plan:
- On a $29 plan, handing off the tax work costs about 2.3% of revenue
- On a $99 plan, the identical trade costs 1.8%
The premium is regressive. Cheap plans pay the biggest surcharge for the simpler life.
One more change undercuts the usual advice:
- Stripe bought Lemon Squeezy in 2024 and now sells its own merchant-of-record product
- On 28 January 2026, co-founder JR Farr said the goal is “an easy way to migrate” users to Stripe Managed Payments
- No end date announced, and Lemon Squeezy still takes new sellers
Pick Lemon Squeezy to dodge a future migration, and you’ve picked the one option that already has one scheduled.
The Seller Test: Three Questions
Three questions resolve this. Each endpoint carries the rate that platform published in August 2026.
The $1,379 figure is Polar’s own published breakeven, not an estimate.
Below it, the free Starter plan wins. Above it, the monthly fee buys back a lower rate.
Pricing Your Plan and What Year One Really Pays
Price against what the problem costs your customer, not against a rival’s feature list.
Then be sceptical of any guide that tells you what year one pays. Those numbers do not survive a source check.
Skip the competitor spreadsheet when you sit down to price.
If your tool saves someone five hours a month, price against the value of that time — not a rival’s $29 tier.
Now the honest part of how apps make money: most founders don’t get rich.
Be careful which year-one numbers you trust. Only two of the three below survive a source check.
Every citation chain behind the top row was followed to its origin in August 2026. The figure was reshaped at each hop.
No credible dataset publishes a median year-one MRR for micro-SaaS. Treat anyone quoting one as quoting a rumour.
What is documented is slower than the pitch decks, and steadier than the failure rate suggests.
Proof of a real business with paying customers, built on evenings and weekends without outside funding.
Build One or Buy One?
Buying skips the unpaid months and starts you at someone else’s revenue. Building stays cheaper only when you already have the audience.
Buying a business means skipping:
- The build, hosting and billing setup
- Months of unpaid work before your first paying customer
- The cost of building an audience from zero
- Starting revenue at $0 instead of inheriting a customer base
Browsing costs nothing.
Flippa’s SaaS listings are free to join and browse. The $49-a-month Premium tier is skippable for a first buy.
Note the asymmetry before you choose.
Buying gives you an audited comparable. The 3.9× multiple tells you whether a listing is priced sanely.
Building gives you no benchmark at all. Nobody credibly publishes what year one earns.
Compare both numbers before choosing either path.
Frequently Asked Questions
Do I Need an LLC Before I Sell My First SaaS Subscription?
Short answer: no.
You don’t need one before your first sale. Forming one soon afterwards separates personal assets from business liability and tidies up your tax filing.
Your timeline depends on how you bill:
- Merchant of record (Paddle, Polar): the platform is the legal seller, so launching as a sole proprietor carries less exposure. Form the LLC within 60-90 days.
- Direct billing (Stripe): form the LLC first. You’re the legal seller from the first transaction.
This is general information, not legal or tax advice — check your own position with a professional.
Can I Switch Billing Platforms Later Without Losing Customers?
Yes, but it’s a manual migration, not a toggle.
Switching typically involves:
- Exporting customer and subscription data
- Re-inviting each customer to enter payment details again
- Absorbing the churn that follows
Card details rarely transfer, because your customers’ payment credentials sit with the old merchant of record, not with you.
Pick your seller of record before launch. Switching later costs customers, not just time.
What Happens if I Miss a State Sales Tax Filing?
You’ll rack up penalties and interest that compound while the return sits unfiled.
Some states charge a late fee even when you owed $0 that period. Repeated misses can trigger an audit of your filing history or a revoked seller’s permit.
This is why Stripe Tax pricing is not the whole answer.
It calculates and collects at 0.5% per transaction on the no-code path. Registering, filing and remitting stay yours.
A merchant of record shifts that duty entirely. Paddle and Polar register, file and remit under their own name.
Do I Need a Privacy Policy Before Launching a SaaS?
Yes, and before your first signup rather than after it.
You need one if you collect emails, payment details or usage data. Both GDPR and CCPA expect the policy to be publicly posted.
Before launch:
- Use a generator or template as the starting point
- List what you collect and why
- Link it in your footer and your checkout flow
Is a Business Bank Account Required Before Accepting Payments?
No, but skipping one costs you later.
Stripe, Paddle and Polar will all technically connect to a personal account. Mixing business and personal funds makes bookkeeping messy and complicates your first tax return.
Open a free or low-cost business checking account before your first sale. Most online banks approve within days.
Conclusion
You don’t need permission to start. You need a Tuesday afternoon and a landing page. The build is close to free. Tools are cheap, templates are everywhere, and no-code gets you further than it used to.
Your real investment is the two hours you spend on the Seller Test — before a single customer pays. Do that this week.
Answer the three questions, write the platform name down, and put up the waitlist page the same afternoon.
For where software sits against the other routes, the models that actually scale is the wider map. For the day-job version of the same decision, starting without quitting is the companion piece.
Ship the small version.
