Playbook · Chapter 4 · The audience

Lever three: segmenting the audience.

The same email sent to two qualified segments can behave like two entirely different campaigns.

The short answer

Split your cold email list into segments before you send, by sub-vertical, company size, role, seniority, observable triggers and geography. Send each segment as its own campaign so you can see which ones convert. Usually two or three segments carry most of the pipeline, and that finding is worth more than any copy change.

Teams treat their ICP as one undifferentiated pool. It never is. Within any market there are segments that convert at three times the rate of others, and until you segment before sending, that variation is invisible. It hides inside an average.

Segment before you send, not after

If you send one blended campaign and get a 6% reply rate, you have learned almost nothing. It might be uniform. It might be 15% from one third and 1% from the rest. Those are completely different businesses and they demand completely different decisions.

You can sometimes slice results after the fact, but it is unreliable. The segments end up uneven in size, the copy was not written for any of them, and the numbers in each slice are too small to trust. Decide the segments up front and give each one its own campaign.

The dimensions that actually matter

Split the market by the dimensions that plausibly change the buying trigger:

How big should a segment be?

Big enough to report honestly. A segment needs enough contacts to give each copy variant a few hundred sends, otherwise you are reading noise. If a segment is too small for that, merge it with its nearest neighbour for testing, then split it again once you know what is working.

Relevance decides who gets on the list at all

Segmentation is not just for analysis. It is how you make each email relevant without writing every one by hand. When a whole segment shares a problem and a trigger, one well written email can feel specific to everyone in it. Be relevant at the segment level first. Individual personalisation is expensive and only earns its place once the data says it helps.

The same logic sets the bar for who enters the list. Every contact should have a reason to be emailed: they fit the segment, and something about their situation makes the problem live. A list built on that rule beats a bigger list built on job titles alone.

Then follow the money

Once each segment has enough volume to report honestly, you can see where the yield actually is. Usually two or three segments carry most of the pipeline. That finding is worth more than any copy improvement, because it tells you where to point coverage and where to stop spending.

The strongest combinations of segment and copy become what we call pillars: proven lists that get emailed again every 30 to 60 days. A mature programme runs several of them at once. More on that in chapter five.

Do not discard the weak segments permanently

A segment that fails on one angle has not necessarily failed. It may need a different framing, since its problem is different. Retest weak segments against the angles that won elsewhere before writing them off.

Common questions

Where do I get the data to segment by?

Firmographic data (industry, size, location) comes from any decent B2B data provider. Triggers such as hiring and funding usually need a second source or some light research. The quality of the segment depends on the quality of that data, so verify it.

How many segments should I start with?

Three to six. Enough to see real variation, few enough that each one gets meaningful volume.

Is my market big enough to segment?

If your whole market is under about 10,000 contacts, heavy segmentation gets hard and you may be better served by researched one-to-one outreach. See when cold email doesn't work.