Segmentation is a set of exclusions
Anyone can produce a list of attractive market segments. It takes an afternoon and a spreadsheet, and the output is almost always the same: several segments that look promising, none formally ruled out, and a recommendation to pursue them all with tailored messaging.
That is not segmentation. That is a wish list with a methodology attached.
Real segmentation is a set of exclusion decisions. It is the moment a leadership team agrees that a category of buyer, however willing to pay, is not who this business is built for. Everything useful about segmentation, sharper messaging, more efficient spend, better margin, follows from the exclusions rather than the inclusions.
This is why segmentation projects stall. The analysis is not hard. The decision is.
Firmographic, behavioral and needs-based
Three approaches, each answering a different question.
Firmographic segments by observable organizational attributes: size, sub-industry, geography, structure, technology in use. Easy to identify and target, which makes it operationally useful. It tells you who they are and nothing about whether they will buy.
Behavioral segments by what organizations do: hiring patterns, technology adoption, funding events, expansion, leadership change. Better at predicting readiness, harder to maintain, and increasingly available through intent and signal data.
Needs-based segments by the problem being solved and how acutely. The most predictive of the three and the least directly targetable, because need is not visible in a database.
Most B2B firms should combine firmographic with behavioral. Firmographic defines the addressable pool, behavioral identifies who within it is currently in motion.
Building the tiers
Once segments are defined, tier accounts within them by fit and value.
Tier one accounts justify individual attention: named research, tailored messaging, direct executive contact, and coordinated sales and marketing effort. This tier should be small enough that you could name every account in it.
Tier two receives segment-level personalization. Industry-specific content, role-based messaging, and coordinated campaigns aimed at a defined group rather than a named company.
Tier three receives programmatic treatment. Scalable campaigns, self-serve resources, and a sales motion triggered by inbound signal rather than proactive pursuit.
The purpose of tiering is allocation. It decides where disproportionate effort goes and, equally, where you deliberately do less.
The cost of poor fit
The argument for exclusion becomes concrete when you look at margin.
Poor-fit engagements consistently run materially below ideal-fit margin. They consume more delivery time relative to fee, generate more scope disputes, require more senior escalation, and produce weaker references at the end. The revenue looks the same on the top line and behaves very differently underneath it.
Most firms have never measured this, which is why the exclusion conversation stays theoretical. Score fit at intake, join it to engagement margin, and let six months of data make the argument. Evidence moves this discussion in a way that opinion does not.
Making it operational
Segmentation that lives in a strategy document changes nothing.
Make it operational by scoring fit at opportunity creation in the CRM, using criteria specific enough that two people would score the same account identically. Report win rate and margin by fit tier. Give sales explicit permission to decline, which requires leadership to actually mean it the first time a poor-fit deal with attractive revenue appears.
That first test is the one that determines whether the segmentation model survives.
Reviewing the model
Revisit annually, or when you enter a new market, launch a materially different product, or notice that your best recent clients do not match your defined segments.
That last signal is the important one. When the model and the reality diverge, the reality is usually right, and the model has quietly become a description of who you used to serve.
Talk to us about ICP definition, segmentation and account tiering.
Common questions
How many segments should we have?
Three to five active. More than that and messaging fragments beyond what a small team can maintain.
What if leadership will not exclude anyone?
Then you do not have a segmentation model, you have a wish list. Start by scoring fit and showing the margin difference — evidence moves this conversation better than argument.
Should we turn away revenue?
Sometimes, yes. If a segment consistently delivers lower margin, longer cycles and weaker references, the revenue is costing you more than it earns.
How does this differ from personas?
Segments are organizations. Personas are people inside them. You need both, and they answer different questions.
