Blog

How much do fake signups cost an AI startup?

8 October 2026 · 5 min read · SignupVet team

Free credits and free trials are one of the best ways to grow an AI product. They are also an open door. Every account that signs up only to use the free part costs you real money, and it quietly distorts the numbers you use to make decisions.

This article shows where that cost comes from and gives you a simple formula to estimate it for your own product, in about five minutes.

Where the cost comes from

1. Compute. With AI products, every free generation, transcript or chat message costs you model or GPU time. A fake account uses exactly the same resources as a real one, and people who farm free credits use them to the last one.
2. Email. Fake accounts receive your welcome and onboarding emails. Throwaway inboxes bounce or never open them, which can hurt your sender reputation, so your emails to real users land in spam more often.
3. Misleading numbers. Fake signups inflate your signup count and lower your activation and conversion rates. You may end up fixing an onboarding flow that is fine, or trusting a marketing channel that mostly brings bots.
4. Time. Someone has to notice the abuse, clean up accounts and answer the occasional chargeback or support request.

Compute is the easiest part to measure, so we start there.

The formula

Monthly cost = signups per month × share of fake signups × cost of the free usage of one account

You need three numbers:

• Signups per month: from your database or analytics.
• Share of fake signups: the hardest one. The next section shows how to estimate it.
• Cost of one free account: what your model or infrastructure provider charges you for the free credits one account can use. For example, if the free tier allows 50 generations and each costs you $0.01, one fully used free account costs you $0.50.

An example with made-up numbers

The numbers below are invented to show how the math works. They are not data from real companies. Use your own numbers in the calculator further down.

The effect on your numbers is just as real. Say 40 of those 2,000 signups become paying customers. Your dashboard shows a conversion rate of 2.0%. Among the 1,600 real signups it is actually 2.5%. That difference can change which experiment you run next or which channel you spend money on.

Calculate your own number

How to estimate your share of fake signups

Take last month's signups and look for these patterns. You do not need a tool for a first estimate, a spreadsheet export is enough:

• Throwaway email domains such as mailinator.com or other temporary inbox services. Public lists of these domains are available on GitHub.
• Many signups from the same IP address or network within a short time.
• Signups from cloud servers or VPNs rather than home or mobile connections.
• Numbered or random addresses like max1@, max2@, max3@ or long strings of random letters.
• Accounts that used all free credits on day one and never came back.

Count the accounts that match at least one strong signal and divide by all signups. The result will not be perfect, but it is usually good enough to tell whether you are looking at a $20 problem or a $2,000 problem.

What you can do about it

• Unlock free credits only after the email is confirmed. This alone stops many throwaway addresses.
• Limit signups per IP address or network within a time window.
• Block known throwaway email domains and keep the list up to date.
• Review instead of block when you are unsure. For example, give borderline signups fewer free credits until they verify a phone number or add a card.
• Watch for waves. Abuse often comes in bursts, so a sudden spike in signups is worth a look before it is worth celebrating.

Pick the steps that fit your product. Even one or two of them usually remove a large part of the cost. The guide on how to stop fake signups walks through each of them with code.