How to Segment a B2B List for Higher Reply Rates

Why a flat list caps your reply rate

I'm Hlib Storchak. I build outbound systems for B2B founders and sales teams, and most of what follows comes from running this for clients, enough to book 2000+ meetings for B2B clients along the way. When a new client's reply rate is stuck low, the first thing I check is not the copy. It is whether the list underneath the copy is one undifferentiated pile of contacts or a set of real segments.

A flat list forces a flat message. If your list mixes a 500-person enterprise account with a 12-person startup, a compliance officer with a growth marketer, and a company that already uses a tool like yours with one that has never heard of the category, the only email that fits all of them is a generic one. Generic emails get generic reply rates. Segmentation is not a nice-to-have layer on top of a good list, it is what makes a genuinely relevant message possible in the first place.

Segmentation is not personalization

These two get used interchangeably and they are not the same job. Personalization is what you do to one contact, a line that references their specific company, role, or a signal you found about them. Segmentation is what you do to the whole list before you write a single line, deciding which groups of contacts get a fundamentally different angle, proof point, or call to action. You can have excellent personalization inside a badly segmented list and still underperform, because the underlying offer and framing is wrong for that group regardless of how well the first line is written. Get the segmentation right first. Personalization then has something worth being precise about.

Step 1: split by ICP fit tier

Before industry, size, or any other variable, split the list by how closely each contact matches your actual ideal customer profile, built from your own closed-won and closed-lost deals, not a persona slide. A simple three-tier split works for most teams: Tier 1 is a near-perfect match on the criteria that have actually predicted closed deals, Tier 2 is adjacent, close on most criteria but missing one, and Tier 3 is a long shot you are testing rather than betting on. Tier 1 gets your sharpest, most specific message and the most follow-up effort. Tier 3 gets a lighter, more exploratory ask, since you are not yet sure the fit is real.

Step 2: segment by industry, and verify the benchmark first

Industry is a real segmentation lever, buyers in regulated or gatekept sectors respond differently than buyers in categories where cold outreach is already normal. But this is also the dimension where I see teams import a number from a blog post and treat it as fact, and it is worth showing why that is dangerous with an actual example.

A "legal services gets roughly 10% reply rate, software gets under 1%" table circulates across a lot of 2026 cold email roundups, usually attributed loosely to "Belkins' data." I went and read Belkins' own published 2025 response-rate study directly, 7,530,489 emails analyzed, 34,393 replies tracked, January through December 2025. It says something different. Its actual industry breakdown puts Food & Beverage at 3.47%, the standout top performer, while Construction, Financial Services, Healthcare, and Legal Services all land together in a 0.56 to 0.60% range, with Banking and Insurance sitting at the bottom. The report itself states the spread between best and worst-performing sectors is almost 10x, and the average reply rate across all 2025 campaigns in that dataset was 0.45%, nowhere near the 3.43% platform average Instantly reports from its own, much larger dataset. Same claimed source, a materially different number once you read the primary document instead of the roundup quoting it.

The lesson is not that industry does not matter, the almost-10x spread proves it does. The lesson is that you should segment by industry using your own send data as it accumulates, and treat any published industry table, mine included, as a hypothesis to check against your own numbers rather than a figure to plan a campaign around unread.

Step 3: layer in technographic signals

Once fit tier and industry are set, technographic data, what tools a company already runs, changes the actual argument you are making. A prospect using a direct competitor needs a "here is specifically what we do differently" angle. A prospect using no comparable tool at all needs an education-first angle that establishes the category is worth a look before it pitches a specific product. Treating both groups with the same "why us" email wastes the stronger opening either one deserves.

Step 4: segment by buying role and seniority

This is the mistake I see most often when I take over an account: every segment gets the same email with a different first name and title swapped in. A manager and a VP are not the same buyer reading the same email differently, they are looking for different proof. A manager wants tactical detail, how it works day to day. A VP or C-suite reader wants the business outcome and the risk of getting it wrong, and reads a tactical, feature-first email as a sign the sender does not understand who they are talking to. Segmenting by seniority means writing a genuinely different framing per level, not the same paragraph with a heavier title in the greeting line.

Step 5: segment by intent and trigger events

The last layer, and the one that ages fastest, is timing. A company that just hired a VP of Sales, just raised a round, or just posted a job for the role your product replaces is a different segment from the rest of your list even if every other attribute is identical, because the opening line can reference the actual event instead of a generic value proposition. Trigger-based segments are usually small and short-lived, which is fine. Their job is not volume, it is a sharper hook for the handful of accounts where the timing is real.

Step 6: size each segment so it is worth building

A segment that is too small to produce a real signal is not worth the time it takes to write a distinct message for it. Here is the rough maths I use, built on stated assumptions you should swap for your own: to see roughly 5 replies, enough to tell a real pattern from noise, at an assumed reply rate of X%, you need about 5 divided by X contacts in that segment. At an assumed 3% blended reply rate, that is roughly 165 contacts. At an assumed 1% reply rate for a tougher, more regulated vertical, closer to 500 contacts. If a segment you are excited about only has 40 real contacts in your whole addressable market, it is a personalization case, not a segment, handle it as a one-off rather than building sequence variants for it.

The segment-to-message map

Once the dimensions are chosen, map each one to what actually changes in the message, not just the tag in your CRM. This is the table I build with a client before a single sequence gets written.

How many segments is too many

Teams that get excited about segmentation tend to overshoot it, ending up with 15 or 20 thin slices that each need their own sequence and nobody has time to actually write well. For a first pass, I keep clients to 3 to 5 segments, usually a fit-tier split crossed with one or two of the dimensions above that clearly matter for that specific offer, not all six at once. A segment only earns its own message if the message actually needs to be different. If two segments would get functionally the same email with a synonym swapped, they are one segment, and combining them frees up the writing time to make the fewer, real segments sharper instead.

Mistakes that quietly cap a segmented campaign

The most common one: segmenting the list in the CRM but writing one sequence for all of it anyway, which gets you the reporting benefit of segmentation with none of the reply-rate benefit. Second: importing an industry or seniority benchmark from a blog post without reading the primary source, the table above shows how far a repeated number can drift from what the original study actually says. Third: building segments around data you cannot reliably get at scale, a beautifully specific technographic split is worthless if your enrichment source only covers 20% of the list with confidence. Fourth: never revisiting the split once a campaign is running, your own reply data after a few hundred sends is a better segmentation signal than anything you assumed on day one.

FAQ