How to set free plan limits that don't invite extra accounts
Free accounts can eat the capacity paying customers rely on. How to set free plan limits that pay less per extra account, and which basic fixes break.
Free users are slowing down your paying customers, and cutting your free plan limits feels like the only way out.
Hi, this is the third Monday in a row our 9 a.m. export has crawled. Today it took 47 minutes. It used to take 2. My team builds the client report from that file and sends it at 9:30, so we've now missed that twice. Your status page says everything's fine. We've paid you $79 a month for 14 months and I've recommended you to two other agencies. Is this the new normal? If it is, I need to start looking elsewhere this week.
Say that's in your support inbox at 9:52 on a Monday morning. It's a scenario with numbers you can check, so read it as if the name at the bottom is the customer you'd least like to lose. Your free plan limits are all working exactly as you set them.
You open the job queue. The status page is right: nothing's down. Your 20 workers are busy, and at 2 minutes a job they clear 600 jobs an hour. Your paying customers need about 420 an hour at this time on a Monday, which used to leave plenty of room. What's filling the room is the free plan. It gives every account 30 runs a week, it refills at 8 a.m. on Mondays, and 210 free accounts are each sending 3 runs an hour, the most your rate limit allows. Say those accounts belong to 14 people with 15 accounts each. Not one of them has broken a rule you wrote.
- 210 accounts × 3 runs = 630 free jobs an hour
- 630 + 420 paying jobs = 1,050 jobs arriving an hour, against 600 cleared
- 1,050 minus 600 = a line that grows by 450 jobs every hour
- At 9 a.m., 450 jobs sit ahead of that export. At 10 jobs a minute that's 45 minutes, plus 2 to run: 47
Those accounts don't stop at noon. Thirty runs at 3 an hour takes 10 hours, so they're still going at 6 p.m. Over the week that's 6,300 runs (210 × 30), or 10.5 hours of everything your workers can do, handed to 14 people who've never paid you.
The customer who wrote to you is the one who bothered. Others just leave. Say 5 customers on that plan cancel this quarter: 5 × $79 = $395 a month, or $4,740 a year. The one in your inbox has paid you $1,106 so far (14 × $79).
You could buy your way out. Clearing 1,050 jobs an hour takes 35 workers (1,050 ÷ 30), so 15 more. At $60 a worker that's $900 a month, or $10,800 a year. It takes 12 customers at $79 to cover it, and if you spend $20,000 a month, it's more than half a month of runway, spent so 14 strangers don't have to wait.
Which leaves the option you've been avoiding all morning. You wrote "30 free runs a week" on your pricing page yourself. Say half your paying customers started there, including the one you're about to apologize to. Now you're working out how far to cut it, knowing that every honest free user will pay for what 14 people did.
Why free plan limits invite extra accounts
A free plan limit is counted per account, and a new account costs nothing to open. So every limit you publish also tells people what a second account is worth:
At 30 runs a week, one extra account is worth 1,560 runs a year (30 × 52). A more generous plan raises that reward, and opening the account still takes minutes.
The obvious fix is a smaller number. Cut 30 to 10 and the reward falls to 520 runs a year. But someone who wanted 450 runs a week now needs 45 accounts in place of 15 (450 ÷ 10), and accounts are free. Your honest users lost two-thirds of the plan, and the person you were aiming at spent a little longer on it. What actually changes the reward is what you limit and when it refills, so start there.
How to set free plan limits that pay less per extra account
Sort every limit by one question. Would ten accounts get ten times this?
| Free plan limit | What ten accounts get | Multiplies? |
|---|---|---|
| Runs, credits, minutes, API calls | Ten times as many | Yes |
| Members in one workspace | Ten workspaces of one | No |
| How long history is kept | The same | No |
| Largest file or job | The same | No |
Spend your generosity on what doesn't multiply. Members, history and job size are what people feel every day, and ten solo accounts don't add up to one team workspace. Before you move limits around, decide whether they sit per user or per workspace, because a new workspace can reset them too.
Size what multiplies to the first useful result. Look at how many runs a real user needs before the product clicks, then add a little margin. If most people get there in 8 runs, 30 a week was paying for something else. The same reasoning works for AI credits.
Stop refilling everyone at once. Refill on each account's own signup weekday, or only when someone comes back. If those accounts' signup days are spread evenly, 630 jobs an hour becomes 90 (630 ÷ 7), and 420 + 90 = 510 fits inside 600. Accounts opened on the same day still refill together.
Let the allowance grow with the account. Try 5 runs in week one, 10 in weeks two and three, and 30 from week four. Fifteen accounts opened today get 75 runs in their first week (15 × 5), where they'd have had 450. Honest users who stay reach the full plan. Extra accounts get older too, so this buys you time and nothing more.
Give free traffic its own capacity and a total cap. Reserve 16 workers for paying customers. They clear 480 jobs an hour, which covers the 420, and your export is back to 2 minutes. The other 4 clear 120 free jobs an hour. A shared pool like this, or a hard cap on free spending, is the one limit that doesn't grow with the number of accounts. Honest free users carry the cost: in that first hour the free line grows by 510 jobs (630 minus 120), so a run started at 9 a.m. waits 4 hours 15 minutes (510 ÷ 120).
All six steps lower what an extra account is worth, but a new account still gets whatever a new account gets. So you'll try to stop the new account itself.
The basic checks people add, and how they break
At the email field
Email verification catches typos and dead addresses, and a unique-email rule catches the same address used twice. Neither stops a second real inbox. Gmail ignores dots in a username, and a plus sign followed by any word still reaches the same inbox, so one mailbox can be written many ways. Disposable-domain blocks catch known throwaway services, miss whatever isn't on the list yet and turn away people who use privacy forwarding, and the list needs upkeep for as long as you run it.
On the signup form
CAPTCHA stops simple scripts and does little about a person opening accounts by hand, and it gets in the way of people using screen readers or privacy tools. A phone or card requirement stops casual second accounts. Anyone with another number or card carries on, you pay for every text, and some real people leave instead.
At the network
One signup per IP address catches repeats on one connection. It also lands on offices, campuses and households, and then you're writing exceptions. VPN blocks catch known ranges, and they turn away privacy-minded users and staff on company VPNs while you maintain a second list.
Inside the product
Rate limits and usage caps hold one account to its limit. Your limit of 3 runs an hour was on that Monday, and all 210 accounts sat under it, because a rate limit counts inside an account. Caps also land on your most active honest users. Shorter trials and smaller allowances take value from everyone, and more accounts make up the difference.
By hand
Manual review catches what a person can spot. Say you have 3,000 honest free accounts on top of the 210. At 2 minutes each, 3,210 accounts is 6,420 minutes, or 107 hours. That's more than two and a half work weeks, and it's out of date as soon as the next account opens. A tired reviewer will also flag some real people.
Stack all of it and your signup asks for a code, a puzzle, a phone number and a card, while someone on your team keeps two lists and a spreadsheet of exceptions. Honest people feel every step, and next Monday's refill still goes out to every account that got through.
Deciding who gets the free plan at all
Your limits decide how much a free account gets. Nothing above decides whether this claim should go through, and that's the job AbuseZero is built for. Before a free plan or a refill goes out, your server makes one API call and gets a clear allow or deny. AbuseZero flags abuse that single checks miss and detects linked claims, with multi-layer AI risk analysis examining 140,000+ signals. There's also a proprietary algorithm, with AI that checks the claim again. The domain list that went stale? AbuseZero checks mailbox aliases, disposable domains and mail routing. The IP limit that landed on an office? AbuseZero sees beyond an IP address with VPN, hosting, network reputation, Tor and proxy intelligence. And when the emails keep changing, it recognizes repeat devices across new accounts, so creating a fresh account isn't enough to erase previous activity.
The rules stay yours. The free plan can be once per user and the weekly refill once per calendar week, and Observe mode shows you what would be denied before anyone is blocked. Your app still enforces the limits you just set, and you don't have to build your own abuse prevention system to decide who receives them. The quickstart covers your first benefit.
Frequently asked questions
What are good free plan limits?
Enough for a real user to reach a first useful result, generous on what doesn't multiply across accounts, and tight on what costs you money each time it runs.
Do usage caps stop people opening extra accounts?
No. A cap holds one account to its limit, and every new account arrives with a fresh one. Keep your caps, and see how the wider problem works.
Should I just remove my free tier?
Treat that as a last resort, because it also removes the way many people find you. Here's how to decide. AbuseZero evaluates each claim before the benefit goes out, so the size of your free plan can stay a product decision.
Protect the free plan you built
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.