TL;DR
- The Setup: Almost every company I work with has a segmentation slide, and most of them are a description of the customer list sorted three ways.
- The Problem: Industry, headcount, and job title describe who a buyer is. They do not predict what the buyer will do next, and prediction is the only reason to segment at all.
- The Cost: A segment that changes no campaign, no message, and no sales motion is a chart with a budget attached. The team spends a quarter building it and nothing downstream moves.
- The Fix: Define segments around a trigger, the thing that changed for the buyer in the last ninety days, and keep the attributes as filters rather than as the definition.
- The Payoff: Fewer segments, sharper copy, and a sales team that can explain why a specific person is in the pipeline instead of pointing at a persona name.
I have sat through more segmentation presentations than I can count, and they follow the same script. Someone opens a slide with four colored circles. Each circle has a name like “Growth Stage Operators” or “Enterprise Modernizers.” Under each name there are three attributes, a rough headcount of the addressable market, and an estimate of contract value. The room nods. The deck gets filed. Six months later nobody can tell you which circle a deal came from, and the circles have not changed a single email.
That is not a data problem. The problem is that the segmentation answered a question nobody needed, which is who is in my market, instead of the question that decides revenue, which is who is about to move and why. Those are different exercises, and only one is worth the quarter.
Every Segmentation Deck Answers the Wrong Question
The standard deck sorts by firmographics because firmographics are available in any data provider, join cleanly to a CRM, and fit in a circle. The exercise feels rigorous. You can count accounts, size the value, and put a percentage on the slide. What it cannot do is tell you which account has a reason to buy this quarter, and that is the only variable marketing can influence.
Big data made this worse before it made it better. With forty attributes you have more ways to slice, and slicing feels like insight because it produces a new view. A team can spend a month building an industry hierarchy and never ask what made last quarter’s best customer pick up the phone. The dataset grew and the decision did not sharpen.
Two companies with identical firmographics can be in completely different situations. One just hired a CRO who wants attribution inside a quarter. The other is in a hiring freeze. A segmentation that puts them in the same circle has told you nothing you can act on, and it guarantees the copy written for that circle lands for one and misses the other.
The Attribute Is Not the Reason
Attributes correlate with purchase. They do not cause it. That distinction is where most of the wasted spend lives, because a correlation is comfortable. It lets you build a list, and a list feels like a plan.
Consider the difference between two descriptions of the same buyer. “Director of demand generation at a five hundred person SaaS company” is a filter. It tells you who to send the email to. “Director of demand generation whose new CRO asked for pipeline attribution and who has already tried two agencies” is a segment. It tells you what to write, what to leave out, and what the first call should be about.
The second description is harder to compile. It lives in call notes, in the reason a deal slipped, in the sentence a champion used when asking you to help justify the spend. That material exists in every company. It just does not arrive in a data provider, which is why it gets skipped for the columns that do.
This is the same argument I make about content, which is that the useful signal usually arrives before the clean dataset does. If you want the longer version, I wrote about building campaigns around signal first instead of list first. The mechanics transfer directly. Segmentation is the first place the signal should be read, not the last place it gets filtered out.
A Segment Is a Decision, Not a Slide
Here is the test I use now. Name three things that change if the segment is real. The message. The sequence. The qualifier sales uses on the first call. If you cannot name all three, you have a filter with a name, and the name is doing the work the decision should do.
Run that test on your current segmentation and be honest. For most teams the message does not change between circles, because the same three proof points get used everywhere. The sequence does not change, because the same seven emails ship. The qualifier does not change, because sales asks the same five questions. The segmentation exists on the slide and nowhere in the operation.
When a segment is real, you feel it in small decisions. The first sentence of the email is different. The case study you lead with is different. The thing you refuse to say is different, because you know it reads as noise to this group. That is what a segment buys, and it is worth more than a market sizing exercise nobody reads after the offsite.
The harder test is to ask the person who owns the number, not the person who built the deck. If the segment does not change what your sales team says on a call, it is not a segment yet, and the money spent researching it is a research expense rather than a revenue decision. Turning research into a decision is a systems problem more than an analytics problem, which is why I keep coming back to the stages of building an operating system around this work instead of treating each analysis as a one off.
What I Ask Instead
I replaced the attribute inventory with four questions, asked about the last ten deals won and the last ten lost. The answers are specific enough to write from, and they cluster in a way a firmographic sort never does.
What changed for them in the last ninety days? A new leader, a missed number, a reorg, a competitor move, a board meeting on the calendar. The change is the reason the conversation is happening now rather than last year, and it belongs in the first line of the message.
What are they accountable for this quarter? Not their job description. The number they have to defend in the next review. Two buyers with the same title can be accountable for completely different things, and the one who is accountable for pipeline will read a different pitch than the one accountable for retention.
What have they already tried? This question does more work than any other. A buyer who tried an agency and got burned needs a different opening than a buyer who tried hiring and could not fill the role. It also tells you which objection is already loaded before you speak.
Who else has to agree? The shape of the deal is the shape of the approval. A segment where the champion has budget and a segment where the champion has to borrow credibility from a CFO are two different campaigns, even when the company size matches.
Four questions, answered from material you already own. The output is not a prettier deck. It is two or three groups where the message, the sequence, and the qualifier all change, plus a team that can say out loud why an account belongs in one. That is a segmentation you can run a quarter on.
Your segmentation is a filter with a name on it. If the message, the sequence, and the qualifier are the same across every circle, you did not segment the market. You sorted it.
Take your last ten closed won accounts and write one sentence for each about what changed for them in the ninety days before they replied. Do not sort them yet. Just write the ten sentences. The repetition you see across them is a segment. The attributes on your current slide are probably not.
What I Actually Think
Data did not make marketing smarter. It made sorting cheaper, and cheap sorting invites the belief that the work is done. The teams that get value are not the ones with the most attributes. They narrowed to two or three groups and changed what they say, when they say it, and who they say it to.
If your segmentation cannot survive the three question test, the honest move is to delete it rather than defend it. A smaller number of groups built from why people actually bought will outperform a larger number built from what a data provider sold you. The market does not care how many circles you drew. It cares whether the first sentence was written for the person reading it.
Build a segmentation your team can act on →
