How to Segment by Fit and Behaviour: The Four Boxes
How to segment by fit and behaviour: two questions sort every contact into one of four boxes, and each box tells you how much of your attention it deserves.
To segment by fit and behaviour, ask two questions of every contact: is this the kind of person or company my product genuinely serves, and have they done something lately that signals interest. The two answers place each contact in one of four boxes, and each box tells you how much time and money the contact deserves.
Most contact bases are segmented, if at all, by whatever fields the form collected. That sorts people by what they look like. This post shows the sorting that predicts what they will do next.
How to segment by fit and behaviour: define the two axes first
Fit and behaviour are different kinds of fact, and mixing them is the first way the method goes wrong. Keep them apart before you combine them.
| Fit | Behaviour | |
|---|---|---|
| What it answers | Who is this person or company | What have they done |
| Examples from the guide | The right role, the right type of organisation, the right problem | Read, clicked, visited the pricing page, filled in a form |
| How fast it changes | Slowly | Quickly |
| Does its value fade | No | Yes: a click last week matters, a click last year barely does |
| Where it is recorded | Mostly in fields | As recent activity and facts such as attended an event |
Before you segment anything, write down what good fit means for your business, in facts you can actually record. Then list which behaviours signal real intent and which are noise. If you cannot record a fact, you cannot segment on it, so a fit definition that reads like a feeling needs rewriting until it reads like a field.
Why demographics come second
The fields that are easiest to collect, such as location, age and a free-text job title, tell you least about whether someone needs what you sell right now. Two people of the same age in the same city can be evaluating, never going to buy, and already a customer.
That does not make every demographic useless. A role or a type of organisation can be a fit fact if it is part of what makes someone a good match. The test is whether the fact helps predict what the person needs, not whether it was convenient to collect. Age and city, by themselves, rarely pass.
The four boxes
Put the two axes together and you get the most useful grid in the whole method. Almost every sales and marketing decision about a contact is a decision about which of these four boxes they are in.
| Box | What it means | What to do |
|---|---|---|
| High fit, recent behaviour | The right kind of contact, and active now | Talk to them now, personally if you can |
| High fit, no recent behaviour | The right kind of contact, quiet for now | Nurture patiently with useful material, since they are worth the wait |
| Low fit, recent behaviour | Not your target, but active | Serve them well and cheaply, and do not chase them with sales time |
| Low fit, no behaviour | Neither | Spend nothing |
The grid works because it protects scarce things. Personal attention is the scarcest, and it goes to the first box. Patience goes to the second. Low cost service goes to the third. The fourth gets nothing, and that is the right answer, not a failure to care.
Handling each box
High fit, recent behaviour
This is the group sales should hear about quickly. Behaviour has a shelf life: someone who visited your pricing page this morning is in a different state from someone who did it three months ago. Treating them identically wastes the first and pesters the second. The value of reaching them is in the timing, and timing is worth more than any amount of clever copy delivered late.
High fit, no recent behaviour
Do not give up on these, and do not hammer them either. Your most useful content, less often, is the right register. They fit, so the wait is justified. If they stay silent for long enough they will drift towards the dormant group, which has its own treatment in how to retire dormant contacts.
Low fit, recent behaviour
These contacts are often the noisiest. They click, they visit, they download, and none of it means they are going to buy. Serve them cheaply: automated material, a standard reply. The mistake is to count their activity as proof of demand and give them sales time.
Low fit, no behaviour
Do not send them a stream of mail. Every ignored message teaches the inbox providers that your mail is unwanted, and that judgement falls on all your mail, including what your best customers were waiting for.
Turning the grid into recorded facts
A grid in your head helps nobody else. To make it durable, record each axis as a fact the base can hold.
- Record fit facts as fields. Role, type of organisation and similar properties have one value per contact, so they belong in a field. The rule for choosing is in tags vs fields in a contact database.
- Record behaviour as facts that stay true. Attended the onboarding session, requested a quote, came from the partner referral. These are facts, not opinions.
- Express the judgement as a rule. If you want to call someone a strong prospect, make that a segment or a score computed from the facts, so it updates itself when the facts change.
The tempting shortcut is a tag called hot or promising. It is true on the day it is applied and false soon after, nothing removes it, and two people apply it to different standards. A rule over facts never goes stale in that way.
Writing the rules as sentences
Each of the four boxes should be writable as a single sentence before you build it. For example, using only the guide's own wording, a segment can read as contacts who fit the target, have taken a deliberate action, and have not been contacted in two weeks. That sentence is something both sales and marketing can agree to, test and revise.
When a segment comes back surprisingly large, the usual cause is a stray condition that says any of rather than all of. Build narrow, then widen on purpose. Each one is also worth naming for its rule, not its campaign. Advanced interest with a recent pricing visit will still make sense in six months. October push will not.
Where scoring fits
A lead score is best understood as a segment rather than a magic number: behaviour and fit converted to points, with a threshold that says worth a conversation now. Three disciplines keep it honest.
- Let behaviour decay, so the score describes the present while fit stays.
- Require a real action before promotion. Opens and page views alone are not enough: a form, an application or a request is.
- Disqualify explicitly. Competitors, test addresses and, where you sell to companies, personal free-mail addresses should lose points or be excluded.
Then watch the share above the threshold. If almost everyone qualifies, the threshold has stopped meaning anything. Scoring has a playbook of its own, so treat this section as the segmentation view of it, not the full method.
Mistakes when sorting by fit and behaviour
- Letting one axis hide the other. A very active contact who does not fit is still a poor use of sales time. A perfect fit who has gone quiet is still worth patience.
- Treating old behaviour as current. A click from long ago says little. If nothing fades, your most recent action and your oldest look the same.
- Counting opens as intent. Opens and page views alone are not a deliberate action.
- Forgetting test contacts. Team addresses and test runs open and click everything and score as your most engaged contacts. Mark them with one consistent fact and exclude them by rule.
- Segmenting on demographics because they were easy. The full list of such habits is in segmentation mistakes founders make.
A short checklist
- I have written down what good fit means for my business, in facts I can record.
- I know which behaviours signal real intent and which are noise.
- Fit and behaviour are recorded separately.
- Every contact can be placed in one of the four boxes.
- Each box has a different treatment, and the fourth gets nothing.
- Judgements are rules over facts, not tags that expire.
Frequently asked questions
Is fit the same as lead score?
No. Fit is one input to a score. The score combines fit with behaviour that fades. Keeping the two inputs visible separately is what lets you place a contact in the right box.
What if I do not know what good fit looks like yet?
Then write your best current definition, in facts you can record, and revise it as you learn. A definition you can revise beats a feeling you cannot inspect. Splitting results by segment will show you which fit facts matter, as described in how to measure results by segment.
Which behaviours count as intent?
The guide's examples are reading, clicking, visiting the pricing page and filling in a form, with the strongest being a deliberate action such as a form, an application or a request. Decide your own list and write down which are noise.
Should low fit contacts be deleted?
Not by default. The guide treats them differently, not as waste: serve them well and cheaply, and spend nothing on those with no behaviour. Removal is for records that are wrong, such as bounced addresses or people who have clearly moved on.
The full method, with the checklists, is in The Segmentation Playbook, a free 35-page playbook.