Honest numbers: what do you really earn with apps and SaaS?

With AI you can build a working app in an evening. What the “$10k MRR” videos don’t mention: building has become cheap, selling hasn’t.

After this lesson you’ll know: what apps and SaaS earn in practice (the median, not the outlier), which of the two routes suits you, and which three things you need to sort out before you accept a single euro from a consumer.

Two routes, very different timelines

	Route A: building for clients	Route B: your own product
What	Internal tools, dashboards and custom work for business owners	Your own micro-SaaS, app, template or MCP server
First euro	Often within weeks, via your network	±5 months (for those who make it)
Biggest risk	You sell hours, so it scales poorly	Nobody pays
Potential	Limited by your time	Large, but very unevenly distributed

Route A timings are our estimate, not a measured value; route B timings come from the data below.

Route A has demonstrable demand: on Upwork, earnings for AI Integration rose 178% (2025 vs. 2024, mainly US clients). A freelance software developer in the Netherlands charges €94 per hour on average excl. VAT (Knab Uurtarievenboekje 2026, ⚠ for AI builders without experience it’s lower, there’s no source for that).

What the data says about your own products

Best public source: TrustMRR, where startups have their revenue verified via their payment provider (Stripe, Paddle, LemonSqueezy and others). As of 2 November 2026:

Revenue	Share
$0 to $1,000	67.7%
$1,000 to $10,000	16.9%
$10,000 to $100,000	10.3%
$100,000 to $1 million	4.0%
More than $1 million	1.0%

• Median total revenue (ever, not per month) of a Stripe-connected startup: $270. 22% of them haven’t earned anything yet.
• Those who make it: about 5 months to the first dollar, another 5 months to $1,000, then another 6 months to $10,000 (TrustMRR doesn’t say whether this is a median or an average).
• Over 84% are under $10,000. “$10k MRR” is an exception, not a plan.

These numbers are, if anything, too rosy

Startups sign up to TrustMRR themselves, and it’s also a marketplace for selling your startup; those earning $0 are less likely to sign up. The timings only apply to those who reached the milestone. ⚠ The page doesn’t say whether the distribution is about monthly or total revenue; if it’s total revenue, the picture for monthly revenue is even bleaker.

Other sources point the same way:

• Stripe Atlas: 20% have a first paying customer within 30 days, but these are founders who have already incorporated a (US) company: a selected group.
• Lovable: 8 in 10 builders plan to make money ⚠, but there is no distribution and no median of what they earn. The platforms earn well; for the builders, the evidence is missing.

What does correlate with revenue

According to TrustMRR (correlation, not cause):

Characteristic	Correlation with revenue
Age of the startup	0.47
Domain Rating (how many and which sites link to you)	0.43
Number of followers on X	0.25

53.5% of founders have fewer than 1,000 followers on X. So sticking with it and being findable in search engines count for more than a large audience. On Hacker News, the success stories mainly come from experienced people with domain knowledge; those who launch without an audience often hear nothing back.

Directories as your first backlinks

Don’t start with posting, but with collecting links: submit your product to relevant directories in your niche (tool lists, alternatives sites, integration marketplaces). A maker on Hacker News: “submitting it to relevant directories… increase its DR” (that maker now sells a directory service themselves). Evidence: the correlation is measured (TrustMRR, n = 6,850), but may also work the other way round: successful products naturally get more links. The effect of directories themselves is anecdotal. Limit: bought links and link exchanges at scale are link spam according to Google and are ignored or penalised.

Honest comparison pages that AI assistants cite

“What’s the best alternative to X?” is something people increasingly ask ChatGPT and Google, and comparisons are among the most cited formats.

1. Add sources and numbers. In the GEO study (KDD 2024), citations, quotes and statistics gave up to 40% more visibility in AI answers (in a test setup), most for pages that don’t rank at the top.
2. Don’t put yourself at number 1. In a small test (⚠ ±100 B2B queries, not repeated), Google’s AI Overviews did cite self-promotional listicles, but left out the publisher’s own brand in 69% to 83% of cases.

Security: the cost nobody mentions

According to a security researcher, in February 2026 one Lovable app had 16 vulnerabilities (6 critical) and the data of over 18,000 users was exposed, due to misconfigured access rules in the Supabase database (including row-level security). Lovable: users are themselves responsible for the flagged issues.

Security

• An AI builder delivers a working app, not a secure one. Make every database table readable only by those who are allowed, and have someone try to get at other people’s data (Security check for your app).
• No API keys in the frontend. A stolen Gemini key went from $180 a month to $82,314 in 48 hours (Feb 2026). Set a spend cap per project or key before launch.
• Scan for leaked keys with a pre-commit scanner and GitHub push protection (managing secrets).

In the EU, a leak is a reportable data breach

With a login screen you’re already processing personal data, so under the GDPR such a leak is a data breach: report it within 72 hours of discovery to the Dutch Data Protection Authority (Autoriteit Persoonsgegevens), unless it is unlikely to pose a risk to the people affected. Sometimes you also report it to your users. Also needed: a privacy statement and data processing agreements with your hosting and AI provider. Chatbot in your app? Since 2 August 2026 you tell users they’re talking to AI (AI Act, Article 50).

Selling to consumers in the EU: three things

Selling to businesses is relatively simple. For private individuals, including in other EU countries, three rules apply.

1. VAT of your customer’s country. A SaaS or AI tool is a digital service. If your revenue from such services (plus distance sales) to other EU countries was ≤ €10,000 both this year and last year, you may charge Dutch VAT (btw). Above that, you charge the VAT of the customer’s country, usually declared quarterly via the OSS scheme (Union scheme).

2. Handle payments yourself or use a merchant of record. With Stripe you are the seller and you handle VAT yourself. With a merchant of record such as Paddle, Paddle is formally the seller and remits VAT worldwide, including chargebacks. As of 2 November 2026:

	Stripe (handle it yourself)	Paddle (merchant of record)
Cost per transaction	1.5% + €0.25 (EEA cards), €0.29 per iDEAL payment	5% + $0.50
On a €5 purchase	± €0.33 (± 6.5%)	± 14%, depending on the dollar exchange rate
VAT return	You do it yourself (via OSS above €10,000)	Paddle handles it

With small amounts the fixed fee hurts; work that out before you dream up a €3-per-month subscription.

3. Right of withdrawal and the withdrawal button. Consumers have a 14-day cooling-off period. For a SaaS subscription (a digital service) it only lapses once the service has been fully delivered, and only with prior express consent; if the customer withdraws earlier, they pay pro rata. For digital content (such as a download) it lapses on delivery, provided the customer consents in advance and acknowledges losing their right. Since 19 June 2026 online sellers must show a clear withdrawal button (EU Directive 2023/2673). ⚠ Several law firms agree on that date, but we haven’t yet been able to check the exact Dutch implementation.

Cost

Starting budget: one AI builder or AI subscription for coding, usually $20 to $30 per month (as of November 2026), plus a domain name. Hidden costs: credits spent on debugging, and credits that expire if you don’t use them. Work it out with the AI cost calculator.

What courses promise

Vibe Coding Academy on Skool charges from ±$99 per month up to a VIP subscription of $3,000 per month (plans page, 2 Nov 2026), with a “21 day AI app building program”. Building in 21 days is perfectly doable; the question is what happens after that.

• “Vibe-code your SaaS to $10,000 per month.” Over 84% of verified startups are below that; those who make it usually spend well over a year on it.
• “Building is the hard part.” Most products fail on distribution and paying demand. A maker on Hacker News: “It’s hard to get people to pay anything.”
• “You can put a Lovable app live just like that.” See the leak above.

Work out route A

With jobs of around €1,000 and a small maintenance fee per client, you can see how many conversations you need per month.

Revenue goal per month (€, excl. VAT)Average project price (€)Recurring per client per month (€) maintenance/subscription, 0 if you don't offer itClients already paying recurringCalls → client (%) 20%Outreach → call (%) 5%

New clients needed per month

0

• Recurring revenue now –
• Sales calls per month –
• Outreach touches per month –
• Outreach touches per workday –

Starting freelancers almost always overestimate their conversion. Start with low percentages and adjust once you have real numbers.

Free · runs entirely in your browser · nothing is stored or sent

A stopping point set in advance

Our advice: still no paying customer after 3 months live and around 50 conversations or attempts? Pick a different problem and keep the code; you’ll need it again next time.

Checklist

0/6

• I have chosen between route A (clients) and route B (own product)
• I know that the median total revenue of a Stripe-connected startup is $270, and why that figure is, if anything, too high
• I plan time for distribution (SEO, directories), not just for building
• I don't let real users in until the security check is done
• I know whether I sell to consumers, and what that means for VAT and the right of withdrawal
• I have agreed a stopping point with myself

Your progress is only stored in this browser.

Test yourself

1. 1According to TrustMRR, what is the median total revenue of a Stripe-connected startup?

   $270$2,700$10,000 per month

2. 2What correlates more strongly with revenue than the number of followers on X?

   The number of featuresDomain Rating (links to your site) and the age of your productWhich AI builder you use

3. 3Your app leaks data of Dutch users. What do you have to do?

   Nothing, the AI builder is responsibleReport it to the Autoriteit Persoonsgegevens within 72 hoursJust fix the bug

Next step: see the full apps and SaaS roadmap and start on the basic skills.