Before the first email goes out, nearly everyone already has a target. It fits on one line (“accountancy practices”, “Shopify stores”, “small firms that are hiring”) and it feels obvious. Yet it was chosen without the one piece of data that matters in cold email: how those companies react when a stranger writes to them.
That target isn't absurd, and it usually holds up on paper. It just hasn't been compared with anything, and the best target is sometimes the one next to it. What follows covers where that first choice comes from, why it misleads, and how to check it with two variants running side by side.
Four ways a target gets picked without being tested
You start from your current clients
Your clients came through word of mouth, a former colleague, a referral, a trade show. They knew you, or knew someone who did, before they signed. Concluding that you should email “companies like them” assumes strangers will behave like people who already trusted you.
The accountancy practice that hired you because one of the partners plays five-a-side with your brother tells you nothing about how another practice will read three lines from a name it has never seen.
You pick the clients you'd like to have
The agency that wants household-name brands emails household-name brands. The developer who is tired of brochure sites goes after software companies. As an ambition, fair enough. As a forecast of who will reply, it's worthless: the most desirable companies probably have a supplier already, and they likely get pitches like yours every week.
You go for the biggest market
“Every SME in the UK” feels safe, because you'll never run out of prospects. But the wider the target, the vaguer the email has to be, and by the second line the reader can tell it could have gone to anyone.
Volume isn't the real constraint anyway. At 50 prospects a week, a full year is 2,600 companies (50 × 52). A narrow target of a few thousand companies keeps you busy for more than a year.
You choose sectors where contacts are easy to find
Some sectors put everything online: named partners with direct email addresses, up-to-date Companies House records, a team page listing everyone. You end up prospecting them because the list builds itself in an hour. How easy an address is to find says nothing about how keen its owner is to reply.
Needing your offer and answering a cold email are different things
When you choose a target, you ask who needs what you sell. Your results depend on another question: who will reply this week to someone they've never heard of?
A lot sits between the two:
- The need isn't felt. A practice with no online booking would use it, but nobody there thinks about it when they walk in on Monday morning.
- The person you emailed doesn't decide. You wrote to the head of marketing; the founder signs off every invoice.
- The timing is off. This year's budget is set, or they hired a supplier last month.
- The email doesn't show you've looked at them. It describes your offer and says nothing about their situation.
This is where a signal earns its place. A public observation (a home page that's slow on mobile, an open job ad, an incomplete Google profile) turns a theoretical need into today's topic, and gives the email an opening line that only this prospect could receive. Two targets with the same need on paper can respond very differently depending on whether that signal is there.
A smaller sub-target can beat the whole target
Compare “agencies” with “agencies of 10 to 49 staff that are hiring a salesperson”. The second is a fraction of the first, but every prospect in it shares the same situation, so the email can name it.
There's an arithmetic reason too. Suppose a broad target is really two halves: one would reply at 5%, the other at 1%. The overall rate comes out at 3% ((5 + 1) ÷ 2). You see a mediocre target and drop it, when half of it was good. Splitting the target is how you find that half. These rates are assumptions for the example, not observed averages.
Starting narrow has one more advantage. If the sub-target works, you can widen it. If a broad target disappoints, you can't tell which part dragged it down.
The method: two variants that differ on one thing
Write your current target out in full
Company type, country, size, technology if relevant, signal, contact role, language. For example: accountancy practices, United Kingdom, 1 to 9 staff, signal “slow home page on mobile”, contact: partner, in English. That's variant A.
Change exactly one thing for variant B
Size, country, signal or contact role. Just one. If you change both size and signal and B wins, you won't know which change did it.
Keep the same hook and the same tone
Same hook type, same email structure, same signature. Otherwise you're comparing two emails rather than two targets.
Launch both in the same week, at the same volume
A variant launched in July can't be compared with one launched in September: the time of year weighs as much as the target. Same number of prospects per week on each side.
Measure three things, in this order
Clicks on the hook link come first and in the largest numbers. Positive replies come next (a “no thanks” counts as a reply, but not as a positive one). Meetings are what pay, and they are the rarest of the three.
Only call it once the gap is clear
Not at the first two-reply lead. The next section explains how to tell.
Here is variant A for a freelance web developer emailing accountancy practices with a free audit:
Variant A: free audit, direct toneExample
Hartley & Co: what I noticed on your website
Hi Fiona, I opened hartley-accountants.example on my phone and the home page took several seconds to load. Someone looking for an accountant between two meetings may well move on to the next practice before it appears. I've written down what I would fix first, in order: your audit Owen Price, Price Web Studio
When to call it: early signal or reliable
Early on, every reply carries a lot of weight. On 20 sends, one positive reply more or less moves the rate by 5 points: 1 reply out of 20 is 5%, 2 replies is 10%. Variant B looks twice as good, and it may be nothing but luck.
To tell a target that replies at 2% from one that replies at 5%, you need roughly 588 sends per target (a standard test at a 5% risk with 80% power). The step-by-step calculation, and what it gives for other gaps, is in how many emails it takes to compare two targets. Clicks happen more often than replies, so they separate two targets sooner.
In MonSDR, the target ranking always carries one of two labels. Early signal means the gap isn't clear yet, and the tool estimates how many more sends it will take to decide. Reliable means the gap passes a two-proportion test at a 5% risk (or the two targets' Wilson intervals no longer overlap), with at least 30 sends per group.
Example · positive replies
What to do once the gap is clear
- One variant is ahead. Double the winner's volume, cut the loser, then build a new variant B from the winner by changing a different single thing. The winner becomes your baseline.
- No gap after the planned volume. Whatever you changed doesn't matter much to these prospects. Keep the variant that's easier to feed (the larger one, or the one whose addresses are verified more often) and test a different dimension.
- Both variants are poor. The problem may lie elsewhere: the hook, or the offer itself. Keep one variant and test another hook. The piece on six hooks and when to use each helps you choose.
The app turns these decisions into one-click recommendations: cut a target, double a target, try a new target, or wait while the ranking is still an early signal.
Launch one campaign per variant of your target, in the same week, and see which one replies.
See 5 prospects for my offerFree, no card needed
Target pairs to test, by profession
Each pair differs on one thing only. The company types, signals and roles below are all options you can pick in the tool.
Freelance web developer
- Signal. Accountancy practices in the UK, 1 to 9 staff, contact: partner. “Slow home page on mobile” against “no online booking, contact form only”. You learn which problem gets a partner's attention, and so how to open your proposals.
- Size. Same target with the winning signal: 1 to 9 staff against 10 to 49. In a small practice the partner may run the website personally; in a larger one, someone else may have taken it over. The test will tell you.
- Country. Local businesses with an incomplete Google profile, prospected in English: United Kingdom against Ireland.
SEO agency
- Contact role. Online stores on Shopify with 50 to 500 products, signal “pages with no title for Google and no sitemap”: founder against head of marketing. The founder decides; marketing lives with the problem every day. The test shows which of the two replies.
- Size. Same target: 50 to 500 products against 501 to 5,000.
- Signal. Consulting firms with 10 to 49 staff: “pages with no title for Google and no sitemap” against “slow home page on mobile”.
HR consultant
- Contact role. Software companies with 10 to 49 staff, signal “currently hiring”: founder against HR lead. In a team of twenty there is sometimes nobody in HR, and the founder reads the CVs in the evening.
- Signal. Same target, contact: founder. “Currently hiring” against “recent funding round”.
- Size. 10 to 49 staff against 50 to 249.
Company news signals (hiring, funding, a new managing director) can't be read from the prospect's website, so they don't produce an audit page. For an HR consultant, a simple question or a guide fits better than a free audit.







