Should you remove your free tier?
Should I remove my free tier? Recount your numbers without repeat accounts first. A decision framework for keeping, shrinking or removing a free plan.
Is the free tier actually hurting anyone who pays us? Say this is waiting in your inbox on a Tuesday morning, from a customer on your $49 plan who can't get in. The scenario is made up, and the arithmetic is all here so you can swap in your own numbers.
Hi. Your app has been showing my team "Service busy, try again later" since 8:15 this morning. There are four of us and we've got a client deliverable due at 2. This is the third morning this month. I saw your post about passing 40,000 users, so congratulations, I guess, but we're one of the ones paying you $49 a month and we can't get in. If this isn't fixed by Friday, please cancel us and refund November.
Sam
You open your model provider's dashboard. Your shared quota ran out at 8:15, and last month free accounts used $14,400 of your $20,000 in model spend. You raise the limit from your phone and write back to Sam. Then you and your co-founder pull 500 free accounts at random and read through them until lunch.
By 1 p.m. you've marked 200 of them as the same people coming back for another allowance. It's a sample and it's rough, but 40% of 40,000 is 16,000 accounts. Those accounts use their allowance to the last credit, about $0.60 a month each, which comes to $9,600. The other 24,000 average $0.20, which comes to $4,800. So the 40,000 that Sam congratulated you on was really 24,000, and every number built on it moves.
| The number | What you've been using | With those accounts out |
|---|---|---|
| Free users | 40,000 | 24,000 |
| Free-to-paid rate, 600 customers | 1.5% | 2.5% |
| Monthly cost per free user | $0.36 | $0.20 |
| Free tier cost per new customer, 30 a month | $480 | $160 |
| Months for a $49 customer to repay that | 9.8 | 3.3 |
Now go back through your year with that table open. The 40,000 is on your homepage and in the deck you raised on in March, and your next investor update either repeats a number you no longer believe or explains where 16,000 users went. In June you hired an infrastructure engineer at $11,000 a month, all in, because free usage was climbing toward a wall. That's $132,000 a year, and two-thirds of the usage on that graph came from people who weren't new. In August you raised your plan from $39 to $49 to cover an expensive-looking free tier. The extra $10 from 600 customers brings in $6,000 a month, and the repeat accounts spend all of it and $3,600 more.
Divide $9,600 by 600 customers and you get $16 each. Sam pays you $49, and $16 of that buys model time for the accounts that locked Sam out this morning.
The bank account says the same thing. With $420,000 and a burn of $35,000 a month, you've got 12 months. Without the $9,600, burn is $25,400 and you've got 16.5. Four and a half months of runway went to people who were never going to pay you, and you didn't know it when you hired, priced or fundraised.
At 2:40 p.m. your co-founder sends one line: "Should we just kill it?" That's how you end up typing should I remove my free tier into a search bar. It feels like the only move big enough. But the case for pulling it rests on that $480, and the $480 is wrong. Look at the bottom row of the table again. You were about to end something that repays itself in 3.3 months because it looked like 9.8.
From our early beta
Fake users drained one team's shared model provider quota. Their paying customers couldn't use the product until the team raised its budget to cover the abuse.
When fake users come in through a free plan, that's the capacity cost of repeat claims. It lands on the people paying you.
Why free tier numbers mislead you
A free tier hands something to each account, and your reports count accounts. When one person opens 30 accounts for 30 allowances, every dashboard you own records 30 people. Those accounts are often your busiest free accounts too, because using the allowance is why they exist. So activation looks healthy while conversion looks poor, and you can spend a quarter rewriting onboarding to fix a problem it doesn't have.
Removing the free tier fixes the bill. It doesn't fix the numbers you already reported, and it doesn't touch the cause. A 14-day trial is one more allowance per account, so the same people can claim that instead. Meanwhile the honest 24,000 lose the thing that brought them to you, and you can't easily bring it back once you've announced it's gone.
Should you remove your free tier? Work it out in this order
Do this before anyone writes the announcement.
Estimate how many free accounts are repeats. A random sample of a few hundred, read by hand, is enough for a decision. Then clean the duplicates out of your metrics before you trust anything downstream.
Recalculate the cost of a real free user, and what the free tier costs per new customer. This cost walkthrough has the formula.
Find out where your customers started. Take your last 100 customers and count how many used the free tier before paying. Then ask some of them whether they'd have tried a paid plan or a trial instead.
Walk the tree below with those numbers.
Yes is a real answer here. If almost nobody starts on free, or the honest cost still doesn't pay back, remove it and don't feel bad about it. The options compare like this in the scenario:
| Option | Monthly saving | What it risks | It fits when |
|---|---|---|---|
| Remove the free tier | $14,400 | Every customer who needed free to get started | Few customers start on free |
| Swap it for a trial | Depends on trial usage | People who need longer, and trials can be claimed again | Your product shows its value in days |
| Shrink the allowance | Part of $14,400 | Honest users get less, and repeat users open more accounts | Honest users don't need what you give them |
| Keep it, limit claims per person | Up to $9,600 | Real people refused if the limit is clumsy | Customers start on free and repeat claims drive the cost |
Removing saves $4,800 a month more than keeping the tier and stopping every repeat claim, so it looks like the better deal. Now say half of those 30 monthly customers wouldn't have bought without a free tier. That's a guess, so use your own answer from step 3. You lose 15 customers a month, and 15 × $49 is $735 of monthly revenue, which stacks up every month:
In month 1 removing is ahead by $4,065. By month 7 it's behind, because 735 × 7 is 5,145. The running total flips in month 13, from $270 ahead to $4,485 behind. That ignores churn and the cost of serving those customers. It also assumes you stop every repeat claim, which nobody does. Stop half of them and the monthly lines cross in month 14 instead of month 7. If none of your customers needed free in the first place, removal wins every month. Your step 3 number decides this.
The positions are a judgment call, there to show the shape of the trade-off. Bottom left is where you want to be, and it's the corner that's hardest to reach with the fixes below.
The basic fixes, and how each one breaks
If the tree says keep it, these are what most teams try first. Each does catch something.
Email rules are cheap, so they come first. Verification proves someone can open an inbox, and a unique-email rule refuses the same address twice, so together they catch typos, dead addresses and lazy repeats. Neither notices a second real inbox. Gmail also ignores dots, and mail sent to name+anything@gmail.com lands in the same inbox. Disposable-domain blocks catch known throwaway services, turn away people who use privacy forwarding on purpose, and need someone updating the list forever.
Then come the steps at the door. CAPTCHA stops simple scripts, does little about a person signing up by hand for the fourth time, and puts a hard step in front of people using screen readers. A phone or card requirement stops casual second accounts better than anything else on this list. It also loses the student or first-time founder who isn't ready to hand one over, each text is a charge you pay, and on a free plan a required card gives up much of what free was for.
Network rules are blunter. One signup per IP address catches repeats on one connection and refuses the second person in an office, a campus or a household. VPN blocks stop known ranges, miss everything off your list and land on privacy-minded users and staff on a company VPN. Both create exceptions to manage and tickets to answer.
Limits protect you from one account running away, and you should have them. They count inside an account, though, so 40 accounts under a limit of 100 add up to 4,000. A shorter trial or smaller allowance costs every honest user the same as it costs the person running 40 accounts. Manual review is how the scenario found its 40%, and it doesn't stretch far: at one minute per account, 40,000 accounts is 667 hours, about 17 working weeks.
Stack all of it and your form asks for a verified email, a puzzle, a phone number and a card. That's a worse free tier than the one you were thinking of removing. If you're doing this yourself, keep the cheap parts: verification, known throwaway domains, a hard spending cap and a separate budget for paying customers, so a bad morning on free can't reach Sam. Then read another 500 accounts every quarter.
If the answer is keep it
This is the work AbuseZero was built to take off your plate, and you don't have to build your own abuse prevention system. Every rule above looks at one thing. 140,000+ signals go into AbuseZero's multi-layer AI risk analysis, which detects linked claims and flags abuse that single checks miss, and those signals are read together, never one at a time. The address that sailed through your unique-email rule? AbuseZero checks mailbox aliases and disposable domains, and it recognizes repeat devices across new accounts, even when email addresses keep changing. The IP rule that refused a whole office? AbuseZero sees beyond an IP address with VPN, hosting, network reputation, Tor and proxy intelligence. There's also a proprietary algorithm, with AI that checks the claim again, and creating a fresh account isn't enough to erase previous activity.
For you, that's one API call when someone claims the free plan, a clear allow or deny, and rules you control. Observe mode shows you what would be denied before anyone is blocked, so you can see it before you decide anything about your free tier. Your caps and your paid capacity are still yours to run. The quickstart protects your first benefit.
Frequently asked questions
Should I remove my free tier if it isn't converting?
Recount first. Repeat accounts inflate the bottom of that rate, and in the scenario 1.5% was really 2.5%. If the corrected number doesn't pay back and customers don't start on free, remove it.
Is a free trial better than a free tier?
A trial suits products that show their value in days. A free tier suits slower adoption. Both can be claimed again from a new account, so swapping one for the other doesn't end repeat claims.
How much free tier abuse should I tolerate?
Some. Chasing zero means refusing real people. Here's how to set the number.
What should I tell existing free users if I do remove it?
Give a clear date, a few weeks of notice and a way to export their work.
Decide with the right numbers
Start with Builder
Pay as you go. Evaluate your first claim today.
See pricing
Builder, Pro, Scale and Enterprise.
Talk to us about Enterprise
High volumes and custom terms.