In this guide
A familiar Friday
You push a fix at 6 p.m. A customer emails at 9: the site won't load. Nothing is wrong with your code. Your free plan ran out of allowance for the month, and the host paused every project on your account until the next billing cycle.
Free plans rarely fail because of traffic. They fail because of a rule you didn't know you were close to. This guide maps those rules for the tools most small products start on.
What to remember
- Free plans stop you in four different ways: usage caps, inactivity pauses, daily limits and terms.
- For a small paid product, the first to bite are usually email per day, database pausing and commercial-use rules.
- A sensible first paid setup costs about $45–$65 a month before card fees, and stays near that until several hundred customers.
Four ways a free plan stops you
Before looking at individual tools, it helps to know what kind of limit you're dealing with. Each kind fails differently, and each needs a different response.
1. Usage caps that pause
You get a monthly allowance. When it's gone, the service stops until the next cycle. The upside is that you can't be surprised by a bill. The downside is that your product goes offline at the worst possible moment, usually a busy one.
2. Inactivity pauses
Some free databases stop when nobody has used them for a while. This one catches quiet business tools and products that sit unused between building and launching. It doesn't care how much you use the service, only whether you used it recently.
3. Daily limits
A monthly allowance can look generous while a daily limit is what actually stops you. Email is the classic example: thousands of emails a month, but only a hundred a day. Launch day is exactly when you need more than a normal day.
4. Terms
Some free plans are technically capable of running your product but don't allow it. A clause saying the plan is for personal, non-commercial use applies the moment you charge money, even with ten visitors a day.
Hosting
| Service | Free plan | How it stops you | First paid step |
|---|---|---|---|
| Netlify | 300 credits a month, 1 concurrent build | Usage cap: sites pause until the next billing cycle, and every project on the account pauses with them[1] | Personal $9 (1,000 credits) or Pro $20 (from 3,000 credits, unlimited members) |
| Vercel | Hobby: 100 GB fast data transfer, 1M function invocations, 1 developer | Terms: Hobby is for personal, non-commercial use only | Pro $20 per developer seat, with $20 of usage credit[2] |
| Cloudflare Pages | 500 builds a month; server code on Workers: 100,000 requests a day | Daily limit on server-side requests | Workers Paid from $5 a month, 10 million requests included |
How Netlify's credits add up
Netlify counts everything in credits: each production deploy costs 15, bandwidth costs 20 per GB, compute costs 10 per GB-hour, and web requests cost 2 per 10,000.[1] Here's a realistic month for a small product on the free plan:
That's a quiet month, and it already sits just under the limit. Two extra hotfix deploys would push it over. The good news is that the free plan's limits are hard and can't create costs, so you'll never get a surprise invoice. Personal at $9 gives you 1,000 credits, more than three times the room.
Vercel's real boundary is the licence
Vercel's Hobby limits are generous for a small product. The catch is the terms, which say Hobby may only be used for personal or non-commercial purposes.[4] A product that takes payments is commercial. So the question isn't whether you'll outgrow Hobby, but whether you're charging yet. Pro is $20 per developer seat and includes $20 of usage credit.
Cloudflare is cheapest, with short limits
Cloudflare Pages serves static files on the free plan with 500 builds a month and up to 20,000 files per site.[5] Server-side code runs as Workers, which allow 100,000 requests a day and 10 ms of CPU time per request on the free plan. The paid plan starts at $5 a month.[6] It's the cheapest of the three at small scale; the trade-off is a runtime with tighter limits than a typical server.
For a full side-by-side, see Netlify vs Vercel vs Cloudflare Pages.
Database
| Service | Free plan | How it stops you | First paid step |
|---|---|---|---|
| Supabase | 500 MB database, 1 GB file storage, 5 GB egress, 50,000 monthly active users, 2 active projects | Inactivity: projects pause after one week without activity | Pro from $25: 8 GB disk, 100 GB storage, 250 GB egress, no pausing, $10 compute credit |
| Neon | 1 GB per project, 100 compute-hours per project a month | Compute sleeps after 5 minutes idle and wakes on the next query; suspends if compute-hours or transfer run out | Launch: pay as you go, about $15 a month for a lightly used database |
The two most popular hosted Postgres options handle idle time in opposite ways. Supabase's free plan pauses a project after a week with no activity;[7] a paused project needs restoring before it works again. Neon's free plan puts compute to sleep after five minutes, but wakes it automatically on the next query, usually in a few hundred milliseconds.[9] Your customer sees a short delay instead of an error.
Backups are the hidden difference
This rarely appears in comparison tables. Supabase backs up Pro, Team and Enterprise projects daily, with Pro keeping the last seven days.[8] Its documentation doesn't list automatic backups for free projects and recommends exporting your data regularly with its command-line tool.[8] If customer data matters, that alone is a reason to move to a paid plan before your first customer, not after.
Storage and egress
Text-only products stay tiny for a long time: thousands of customers can fit in a few hundred megabytes. Products that store photos, PDFs or receipts are different. A 1 GB file limit can be passed with a few dozen active customers, and transfer (egress) becomes the bigger cost once clients download those files. Four products, four bills works through a file-heavy example.
| Service | Free plan | How it stops you | First paid step |
|---|---|---|---|
| Resend | 3,000 emails a month | Daily limit: 100 emails a day | Pro $20: 50,000 a month, no daily cap; Scale $90: 100,000 a month; extra emails $0.90 per 1,000[10] |
Three thousand a month sounds like plenty until you divide it differently. The daily cap of 100 is the real limit for a product with any kind of launch. Sign-in links, confirmation emails, password resets, receipts and notifications all count, and they cluster on busy days.
The emails that matter most are also the ones that hurt most to lose. A missing newsletter is fine. A missing sign-in link means someone can't get into your product. We work through the math in Free tiers that break on launch day, and the separate problem of emails landing in spam in Why your emails land in spam.
Logins
| Service | Free plan | How it stops you | First paid step |
|---|---|---|---|
| Clerk | 50,000 monthly retained users per app | Rarely; well beyond most small products | Pro $25 a month billed monthly or $20 billed yearly; extra users from 2 cents each |
Clerk counts retained users: people who come back at least 24 hours after signing up.[11] Someone who signs up and never returns doesn't count. For a small paid product with a few hundred customers, logins are almost never the reason to upgrade. People move to paid plans for features like branding or advanced controls, not for user counts.
Payments
Payment services don't have free plans in the usual sense. There's no monthly fee; instead they take a share of every charge. That makes payments the one cost that grows with every customer, and the one most people underestimate.
| Service | Fee | Who handles sales tax |
|---|---|---|
| Stripe | 2.9% + 30¢ per US card charge; +1.5% international cards; +1% currency conversion; +0.7% for subscription billing[12] | You, optionally with Stripe Tax (0.5% per transaction on the basic plan) |
| Paddle | 5% + 50¢ per checkout transaction | Paddle, as merchant of record[14] |
| Lemon Squeezy | 5% + 50¢ per transaction, small extra fees outside the US | Lemon Squeezy, as merchant of record[15] |
Two details that rarely make it into comparisons. First, refunds: Stripe doesn't return its processing fees from the original charge when you refund a customer, so a refunded $20 payment still costs you the fee.[13] Second, disputes: each one costs $15 on Stripe, plus another $15 if you respond manually, refunded only if you win.[12]
On a $20 subscription, that's roughly $1.02 per payment with Stripe including subscription billing, or $1.50 with a merchant of record. What card fees really cost breaks this down at different prices.
Three sensible first paid setups
Once someone pays you, the goal is simple: nothing that can pause, cap or break should sit between a paying customer and your product. Here are three ways to get there.
| Setup | Hosting | Database | Logins | Monthly | |
|---|---|---|---|---|---|
| Lowest cost | Cloudflare, $5 | Neon pay-as-you-go, ~$15 | Resend Pro, $20 | Free | ~$40 |
| Solo, balanced | Netlify Personal, $9 | Supabase Pro, $25 | Resend Pro, $20 | Free | ~$54 |
| Team of two | Netlify Pro, $20 | Supabase Pro, $25 | Resend Pro, $20 | Free | ~$65 |
All three stay close to these numbers from your first customer to several hundred. What grows is card fees: at 100 customers paying $20, they add about $100 a month, more than any of these setups costs in total. See costs at 0, 100 and 1,000 customers.
Which limit will you hit first?
| Your product | Likely first limit | When |
|---|---|---|
| Sends reminders or notifications | Email per day | Launch day |
| Stores photos, PDFs or files | File storage, then egress | First few dozen customers |
| Quiet business tool, used weekly | Database inactivity pause | Before launch, during testing |
| Deployed often during launch week | Hosting usage allowance | Launch week |
| Anything you charge for | Hosting terms | Your first payment |
How to check your own limits in ten minutes
Do this for each tool you use
- Open the usage page in the dashboard and note where you are this month as a percentage of the allowance.
- Search the pricing page and terms for "commercial", "personal", "inactive" and "pause".
- Find the daily limits, not just the monthly ones. They're often in a footnote or the docs.
- Turn on alerts at 50% and 80% of the allowance, sent to an inbox you read.
- Write down what happens at the limit: pause, error, or bill. That tells you how urgent each one is.
Five mistakes we see often
- Reading only the monthly number. The daily email cap or daily request cap is usually what breaks first.
- Keeping customer data on a plan without backups. A paid database plan is cheap insurance.
- Assuming "free forever" means "free for business". Check the terms the day you add a price.
- Putting all projects on one free account. On some hosts, one project going over pauses all of them.
- Upgrading on launch day. Do it a week before, when nobody is waiting.
Questions
Can I run a paid product entirely on free plans?
Technically, sometimes. Practically, it's risky: pauses, daily caps and backup gaps all land on paying customers. Moving the database and email to paid plans is usually the first, most valuable step, and hosting follows the day you charge if your host's terms require it.
Which should I upgrade first?
Whatever would hurt customers most if it failed. For most products, that's email (sign-in links) and the database (pausing and backups). Hosting comes next, or first if your plan excludes commercial use.
Are usage-based plans cheaper than fixed plans?
At low usage, often yes. But they can surprise you if something goes wrong, like a bug that makes thousands of requests. If you choose usage-based billing, set a spending limit on day one.
Do these prices include tax?
No. Prices are as listed on providers' pages and may be subject to tax depending on where you are.
How often do these limits change?
Several providers changed their plans in the past year. We re-check every number each quarter; the month is shown at the top of this guide.
What builders are saying
Real public threads on this topic, so you can hear it from people who hit these limits themselves. Summaries are ours; follow the link for the full discussion.
Sources
Checked 14 November 2026. Numbers in brackets in the text point to these sources.
