Email list A/B splitter: random test groups with a holdout
In short: Deal a list or CSV into 2 to 5 random test groups plus an optional holdout, with Gmail, Outlook and company addresses spread evenly and the same inbox kept in one group. Limit: It balances by mailbox provider only; other traits (country, signup date) are random but not forced even.
To split an email list for an A/B test, paste it or upload the CSV, choose two to five groups and an optional holdout share, and press Split. Rows are shuffled with a seed, balanced so each group gets the same mix of Gmail, Outlook.com, Yahoo and company addresses, and one inbox never lands in two groups. Download one CSV per group.
Price: this tool is free. Checking a whole email list is from $9/mo, and the first 100 checks are free with Google, no card. See pricing
Sample result
9 people split into A (5) and B (4), with Gmail, Outlook.com and company addresses spread evenly.
- Groups
- A 5 · B 4
- Seed
- spring
- Same-inbox repeats
- 1
What this means for you
Each version reaches the same mix of inboxes, so the email decides the test, not the list.
- 2 to 5 groups plus a holdout
- Balanced by mailbox provider
- One inbox, one group
- Same seed, same split
- A/B test
- Holdout
- CSV
By SimpleVerifier Team · Last updated · Runs in your browser; nothing you paste is sent to us
How do I randomly split an email list into A and B groups?
- Paste addresses or upload a CSV export. The email column is found automatically and every other column is kept.
- Choose the number of test groups (2 to 5) and, if you want a control that gets no email, a holdout share such as 10%.
- Leave provider balancing on, and type a seed if you need to repeat the split later.
- Press Split into test groups and download the CSV with a group column, or one file per group.
Example
- You enter
- A 10-row CSV of Gmail, Outlook.com, Yahoo and company addresses, including
[email protected]and[email protected]; 2 groups, no holdout, providers balanced, seedspring. - You get
- 9 people split into A (5) and B (4), with Gmail, Outlook.com, Yahoo and company addresses each divided as evenly as their counts allow.
[email protected]reaches the same inbox as[email protected], so it joins her group and is flagged as a repeat. The CSV adds group and same_inbox_repeat columns. - Limits
- Runs in your browser with no row cap; an uploaded file can be up to 25 MB.
Why should an A/B test split be random?
Because anything else lets the list's order pick the winner. An export sorted by signup date, by company or by name puts older, colder or similar contacts together, so taking the first half as A and the second as B compares two different audiences, not two emails. A random split gives every row the same chance of each group. This one uses a seeded shuffle, so the same seed and list give the same groups again, which matters when someone asks how the test was run.
Balancing by mailbox provider goes one step further. Gmail, Outlook.com and Yahoo each filter mail their own way, so a group that happens to draw more of one provider can land in the inbox more often for reasons unrelated to the subject line. The split shuffles each provider family separately and deals it across the groups, the stratified sampling used in experiment design. The table under the result shows the mix per group.
What is a holdout group in email testing?
A holdout is a share of the list that gets no email in the test. Comparing what the mailed groups do (purchases, sign-ins, replies) with the holdout shows the lift the email itself caused, not just which version did better. Ten percent is a common size for a large list; on a small one, a holdout can leave the test groups too small to read. Size them first with the A/B test calculator.
What happens to duplicate addresses?
Rows that reach the same inbox, ignoring capital letters and Gmail dots and +tags (Gmail Help), are given the group of the first one, flagged as repeats and left out of the counts, so one person can't get both versions. To remove them outright, use remove duplicate emails.
Is this better than RAND() in Excel?
| This tool | Excel RAND() and sort | |
|---|---|---|
| Repeatable | Yes, with a seed | No, RAND() changes on every edit |
| Provider mix balanced | Yes | No |
| Same inbox kept in one group | Yes | No |
| Holdout and up to 5 groups | Yes | Manual |
In Excel: add =RAND() in a helper column, sort by it, then take the top half as A. Paste the values first, or the order changes.
Who uses this tool, and for what?
- Cold emailers
- Test two openers across several sending inboxes, with each inbox getting a balanced share of both groups.
- Newsletter teams
- Hold back 10% of a promotion to measure what the email added on top of normal sales.
- Growth teams
- Split a winback list into control and treatment for an experiment the analytics team can audit later from the seed.
What can’t this tool tell you?
- It balances by mailbox provider only; other traits (country, signup date) are random but not forced even.
- A seed repeats the split only for the same rows in the same order.
- It can't tell which addresses will bounce; bad rows land in every group.
- Up to five test groups and a holdout of up to 50%.
Common questions
- How big should each A/B test group be?
- Big enough to detect the difference you care about. The A/B test calculator gives the sample size per variant from your current rate and the lift you want to detect.
- Can I split a list into more than two groups?
- Yes, up to five test groups (A to E), plus an optional holdout.
- Will I get the same groups if I run it again?
- With the same seed and the same list in the same order, yes. Leave the seed blank for a new random split; the seed used is shown with the result.
- Does Mailchimp or my ESP split the list for me?
- Many do, inside their own A/B feature. Splitting yourself helps when you send from several tools or inboxes, need a holdout, or want the groups on file.
- Is anything uploaded?
- No. The split runs in your browser.
Sources
A tidy list can still be full of dead addresses.
Extracting, deduplicating and syntax checks all work on the text of an address. None of them can tell whether the mailbox exists. Only the server that hosts it can answer that, through an SMTP check, and that is what the paid product does.
- Upload the cleaned file as it is: CSVs up to 100,000 rows
- A real SMTP mailbox check on every row, with catch-all detection
- Small list? Starter at $9/mo covers 2,500 emails a month, cancel anytime
Plans from $9 a month, cancel anytime, no contract. Sign in with Google and your first 100 email checks are free, no card. The free tools stay free either way.