How to calculate cold email ROI

Outbound pays when a useful conversation with the right person costs little next to what a customer brings you. The price of an email is only the start of the math. Here is the whole calculation, with numbers you can replace with your own.

Updated 2026-10-30 · Markdown

Four different numbers

A cheap address, a cheap email, a cheap reply and a profitable customer are four different numbers, and putting an equals sign between any two of them is how outbound math goes wrong.

Cost per useful conversation: everything the channel cost over a period, divided by the conversations with people who could buy.
Cost to acquire a customer: the same total, divided by new customers.
Customers needed to break even: the total, divided by what one customer leaves you after variable costs, rounded up.

With the Sending Engine, contacts and infrastructure are already inside the monthly budget, so don't add them again.

A worked example at $800 a month

The entry Sending Engine at $800 a month runs 8 inboxes and 120 campaign emails a day, about 2,640 emails and 880 new people a month at full capacity. For planning, the calculator assumes about 2% of campaign emails get a reply from the start of month three, about 53 replies a month at $800.

The rest depends on your business. Suppose one reply in ten is a conversation with a buyer who has the problem and a budget: about five conversations a month. Suppose one conversation in five becomes a deal: one deal a month. If a deal leaves you $10,000 of margin, the channel pays for itself more than twelve times over; at $1,000 it still pays. These are illustrative assumptions; put in your own rates.

How small a customer can be

Your price can be small. What counts is everything a customer pays you while they stay, minus your costs; the first payment is only part of it.

Take a cautious assumption: only one reply in fifty becomes a customer. At $800 a month that is about one new customer a month, so the channel pays when a customer brings you more than about $760 over their lifetime. A $50-a-month subscription kept for two years brings $1,200; at about 30% cost to serve, roughly $840 stays with you, so the channel pays with a single new customer a month.

The larger the budget, the lower the bar, because each email costs less. With the same cautious assumption, a customer needs to bring you about:

Monthly budget	Planned replies a month	Customers a month at 1 in 50	Lifetime margin per customer to break even
$800	53	about 1	about $760
$3,000	264	about 5	about $570
$10,000	2,640	about 53	about $190

If replies turn into customers more often than one in fifty, the bar drops further. The calculator on the pricing page shows the emails and new people for any budget.

Saved, or lost?

Now compare a cheap option. You pay $100 instead of $800 and get no deals instead of one with $1,000 of margin. The cheap option "saved" $700 and lost $300; with a $10,000 deal the loss is far larger. A cheap channel that brings no result is the most expensive purchase in marketing, paid twice: in money and in time you don't get back.

Cost per reply versus cost per targeted reply

A founder of an AI studio once told us he got 11 replies and five booked calls for $100 in four days: about $9 a reply. But his offer sounded like "we'll build you any AI processes" and went to everyone, so anyone curious about the word AI answered. A calendar can be filled with such calls, attended, closed at zero, then filled three times over and still be at zero.

The right unit is the cost of a targeted reply from the right person. A simple example: one job costs $100 and brings 10 replies, one of them useful; another costs $300 and brings 6, all useful. A reply in the second case costs more, $50 against $10, but a useful conversation costs less, $50 against $100. Add the salesperson's time wasted on empty calls and the gap grows.

Where the money leaks

Losses at each stage multiply. If half the people you chose can never buy and half the replies that arrive get lost, you get a quarter of the possible result for the same money: every useful conversation costs four times as much. That is why the start of the chain, the research, the emails and the search, is where money works hardest.

Step	An ordinary approach	A strong approach
Emails sent	1,000	1,000
Targeted replies	10	100
Meetings, if half the replies become meetings	5	50
Purchases, if one meeting in five buys	1	10

Ten times more targeted replies means ten times more meetings and purchases, as long as the later conversion rates hold. Address accuracy works the same way. Out of 1,000 addresses, a 20% bounce rate, which in our experience is common with popular databases, lets an email reach 800 people; at the roughly 1.7% bounce rate we measured on our own sending, it reaches 983, 23% more, before anyone has read a word.

Give the channel time

The first paid month of the Sending Engine includes setup and warmup: the first emails go out on day 22. Then the inboxes ramp up, and some people answer days or weeks later, sometimes only to a follow-up. The honest horizon for judging the channel is about three months; judging by the first week is like judging a harvest on the day you sow. The first two weeks of sending say almost nothing about demand, because over a short stretch bad luck and normal variation look the same. Money in B2B often comes late, so watch the leading numbers first: replies, warm conversations, calls.

Budget at least two months, better three. When you evaluate, credit each deal to the month its lead appeared: a deal closed in month seven belongs to the month the conversation started. And before demanding results in the first month, remember how long your own last B2B purchase took, from first conversation to payment. In our agency years we reminded clients who expected payback in a month that they had signed with us on day 86 after our first conversation.

When it's too early

We say "not yet" when the channel won't help:

The budget is below the working minimum of 8 inboxes and 120 campaign emails a day for $800 a month. Less volume won't produce a result worth your time; the $500 playbook alone can be a good start.
Your buyers are random consumers. We write to people in specific roles at work addresses. A personal product fits only when it connects to a role, such as investment property for company owners.
Your audience doesn't read business email in English. The country alone doesn't decide this; the language people do business in does.
Your buyers are companies in Germany, where we don't send because the law requires prior consent.
The offer has no edge yet. Then build something people want first.

Validating an idea is a different matter, and it is never too early for that. Customer discovery is cold sales in disguise: describe what you plan to build and see who wants to talk, or better, who will sign a letter of intent with a price and a volume. Such letters are strong evidence for investors.

Frequently asked questions

Is cold email worth it for a small B2B business?

Yes, when a customer brings you more over their lifetime than the channel costs to win them: with the cautious one-customer-in-fifty-replies assumption, about $760 at the $800 entry budget. Even a $50-a-month product clears that bar if customers stay for two years. Below an $800 monthly budget, start with research rather than sending.

How soon does cold email pay back?

The first emails go out around day 22, and replies build until the start of month three, so judge the channel over about three months. If one deal leaves you more than $800 of margin, one deal a month pays for the entry engine.

What is a good cost per lead for B2B?

Compare the cost of a useful conversation, not of a form or a reply. Published 2025 benchmarks put a LinkedIn lead form at about $202 and a Google Ads lead at about $524 across B2B advertisers; at $800 a month our calculator plans replies at about $15 each. See how the channels compare.

Does outbound work for a $50-a-month product?

Yes, if customers stay. Two years at $50 a month is $1,200; after about 30% cost to serve, roughly $840 stays with you, which is more than the $800 entry engine costs in a month.