Choosing a plan feels harder than it is, because the numbers look abstract. This guide translates each limit into something concrete, shows how to estimate what you need from your own test suite and explains when it makes sense to move up. Exact prices and limits are set by BoltMail and shown on the pricing page, so use the numbers below as examples of how to think.

The limits, one by one

  • Requests per month is your overall budget. Every API call counts, including the ones that return an error. It resets on the first of each month (UTC).
  • Requests per minute is the speed limit for each key. It decides how many tests you can run at once, not how many in total.
  • Mailboxes at a time is how many addresses a key can hold simultaneously. Deleting a mailbox frees its slot immediately.
  • API keys is the number of active keys. Plenty of teams need three or four: development, CI, staging and production monitoring.
  • Custom domains are offered on the highest plan by default. Where a plan includes them it has an allowance, and extra domains can be bought one at a time.
  • IP allow-list and support level come with the paid plans.

Estimating requests per month

Measure instead of guessing. A simple email test uses about five calls: create the mailbox, wait for the message, read it, and delete the mailbox, plus occasionally a list. So:

monthly requests = tests that use email x calls per test x runs per month

A project with 25 email tests, run 10 times a day on weekdays, is about 25 x 5 x 220 = 27,500 requests a month. Add 30 percent for retries, debugging and growth, and you land near 36,000. A hobby project with 3 tests run 20 times a month needs about 300 and fits comfortably in the free plan.

Estimating requests per minute

Look at the busiest moment, not the average. If your suite runs 8 tests in parallel and each makes about 5 calls in a burst, that is about 40 calls in the first few seconds. A key with a limit of 20 per minute would be throttled, while 180 per minute is relaxed. Remember that one long-poll call counts once however long it waits, so using /wait keeps this number low.

Estimating mailboxes

Count the mailboxes alive at the same moment. If every test deletes its mailbox at the end, the number is simply your parallelism, plus a safety margin for tests that crash before cleanup. Always clean up in a finally block. A suite that leaks mailboxes will hit the limit slowly and then fail all at once, which is a miserable way to find a bug.

When to stay on the free plan

  • You are learning the API or building a prototype.
  • You run a handful of tests, mostly by hand.
  • One key is enough, and a few mailboxes at a time suffice.

When to move up

  • You see quota_exceeded or regular rate_limited responses in normal use.
  • You want CI on every pull request. Frequent runs multiply requests quickly.
  • More than one team shares the account and needs separate keys.
  • You want an IP allow-list for security.
  • You need your own domain, because the apps you test reject shared throwaway domains.

Plans are prepaid, with no surprise bills

Paid plans are bought for a period, usually a month, and you pay before you use it. Nothing is ever billed in arrears, and nothing renews unless you chose that when you paid. If you did not save a card, the plan simply ends on its date and the account returns to the free plan's limits. Renewing before the end extends the period, so the days you already paid for are never lost. The dashboard warns you in the last days.

Renewal: how automatic renewal works

If you tick Renew automatically at checkout, your card is saved with the payment provider (Stripe, Paystack or Bachs). BoltMail never sees or stores the card number, only its brand, the last four digits and the expiry date so you can recognise it.

  1. A reminder first. A few days before the charge we email you the plan, the amount and the date of your next invoice.
  2. The charge. Three days before your plan ends (the exact number of days can differ), the card is charged for the next period. With Bachs, which runs the subscription itself, the card is charged on the day your plan ends instead. The new period starts when the old one ends, so you never lose a day.
  3. If a charge fails, we retry a few times and email you each time. If it still fails, your plan runs to its end date and then returns to free. Nothing else is charged.

You stay in control at any time. Under Plan & billing in your dashboard you can switch auto-renewal off, switch it back on, change the payment method or remove the card. Cancelling never shortens the period you have already paid for. Payments made with PayPal, crypto or bank transfer are always prepaid and never renew by themselves.

Custom domains and extra slots

If your plan includes custom domains you get a fixed number, for instance three. A common pattern is one domain for production-like staging tests, one for load tests and one spare. If you need more, buy extra slots from the dashboard for a one-time price per domain. The article on custom domains walks through the DNS setup.

Designing for growth

  1. Wrap the API in one small module so that changing how you wait or retry touches one file.
  2. Log request counts per test run. A jump after a code change is an early warning.
  3. Prefer long polling and reuse over loops and per-assertion mailboxes.
  4. Separate keys per workload so that you can see, and limit, who uses what.
  5. Review usage monthly. Five minutes with the dashboard meters tells you whether to upgrade before you hit a wall.

A worked example

A startup tests sign-up, reset and invitation emails: three flows, 5 calls each, run on every pull request. They merge about 30 pull requests a week, with an average of three CI runs each. That is 3 x 5 x 90 x 4.3 weeks, about 5,800 requests a month, plus a nightly job of 1,500, so roughly 7,500. Their peak is 3 parallel tests, so about 15 calls in a burst. The free plan's request budget is just too small for this load, so a first paid plan fits well. A year later they add a mobile app, ten flows and a canary on production, and they move to the next plan. Estimating gave them time to decide calmly.

When you are ready, open your billing page to compare plans, or start with the quickstart to see how few calls a test really needs.