Behavioral segmentation: B2B examples built from your own sales data

Behavioral segmentation groups customers by what they do rather than who they are: how often they order, how recently, how many product categories they buy and whether their spend is rising or falling. In B2B it is built from the invoice ledger with written rules, such as "no order in 90 days = lapsing", and each segment gets one commercial action.

What behavioral segmentation is

Market segmentation as a planning idea goes back to Wendell Smith's 1956 paper in the Journal of Marketing, which set it against product differentiation: accept that demand is varied and serve each part of it, rather than push one offer at everyone. Behavioral segmentation draws the lines from purchase behavior. In a B2B sales ledger, four behaviors carry almost all of the signal:

Recency: days since the customer's last order

Frequency: number of distinct orders in the window

Breadth: number of product categories bought in the window

Trend: spend in this window against the same window a year earlier

Demographic, firmographic and geographic segmentation describe the customer: industry, size, location. Behavioral segmentation describes the relationship. The two do different jobs. B2B customer segmentation from the ledger, not from personas explains why the segments you benchmark against should be defined from what a customer is; behavioral segments are for deciding what to do next with each customer, and they work best inside those firmographic groups.

The four types of market segmentation, briefly

Firmographic fields in the US usually come from standard codes: the Census Bureau describes NAICS as "the standard used by Federal statistical agencies in classifying business establishments", and the SBA's size standards band firms by employees or annual receipts per NAICS industry. Behavioral is the only one of the four that the invoice ledger supports with no extra data, which is why it is the fastest to build and the easiest to keep current.

The rows you need

One row per invoice line:

Plus a customer master that rolls subsidiaries and branch accounts up to the group, so one contractor with three billing accounts is one customer, not three.

Rules, applied in order

1. Lapsing if days since last order > 90.
2. Occasional if orders in the last 12 months < 12.
3. Broad if categories bought >= 4 of 6.
4. Otherwise Narrow.

The first rule that fires decides the segment, so every customer lands in exactly one. In Excel, with orders in B, days since last order in C and categories in D:

=IF(C2>90,"Lapsing",IF(B2<12,"Occasional",IF(D2>=4,"Broad","Narrow")))

The three measures come from the line-level table (here an Excel table named Lines). Recency is the as-of date minus the last invoice date; frequency is a count of distinct invoice numbers; breadth in Excel 365 is:

=COUNTA(UNIQUE(FILTER(Lines[Category],Lines[Customer]=A2)))

Worked example: eight trade customers

A building-supplies distributor, twelve months to September 30, 2026, six product categories.

Grouped by segment:

Brightwater Homes buys 4 categories, which would make it Broad, but it placed only 8 orders, so rule 2 fires first and it is Occasional. Rule order is part of the definition: change the order and the same data gives different segments.

The check: segments add back to the ledger

Customers: 2 + 3 + 1 + 2 = 8, the number of rolled-up customers with any sale in the window. Revenue: 762,000 + 511,000 + 96,000 + 95,000 = 1,464,000, the ledger total for the same twelve months. In Excel, a COUNTIFS on the segment column must sum to the row count, and a SUMIFS of revenue by segment must equal =SUM(E2:E9). If either check fails, a customer is in two segments or none, and a rule has a gap.

One action per segment

• Broad (2 customers, 52.0% of revenue): protect. These are the accounts to cover first and price carefully. Harlow and Oakline are where a lost relationship costs most.
• Narrow (3 customers, 34.9%): cross-sell the missing categories. Northgate orders 38 times a year but buys only 2 of 6 categories. The gap is a products-per-customer problem, and the dollar size of it is whitespace.
• Occasional (1 customer, 6.6%): raise frequency. Brightwater already buys broadly; the question is whether it buys the rest elsewhere between its 8 orders.
• Lapsing (2 customers, 6.5%): call this month. Mercer Millwork ordered 22 times last year and has been silent for 118 days. It goes on the reactivation list.

Segments are not tiers: a tier says how much effort an account deserves, a segment says what kind of effort. Segments and tiers covers the difference.

Where it goes wrong

• Unrolled customer names. One group billed through three accounts shows up as three small customers in three segments, each looking occasional or narrow.
• Rules that overlap or leave gaps. A customer that matches two rules, or none, breaks the check. Apply the rules in a fixed order with a final catch-all.
• Consumer thresholds. Thirty days without an order means lapsed for a retailer's shopper, not for a contractor who orders every quarter. Set the recency rule from each customer group's normal order gap.
• Credit memos counted as orders. A return booked as its own document inflates frequency. Count only invoices with positive value.
• Segments with no action. A segment chart with no named owner and no list becomes a slide, not work. RFM analysis for B2B customers makes the same point about scores nobody acts on.

Behavioral segments from your own ledger

Upload the invoice ledger and Covirage's tools compute recency, frequency, breadth and trend per rolled-up customer, apply your written rules in order and check that the segments sum back to the ledger's customers and revenue. The external AI model explains the result and never does the arithmetic. See share of wallet to turn the narrow and lapsing segments into a ranked cross-sell and reactivation list, and key account management for what to measure on the broad ones. For ranking customers by revenue first, see Pareto analysis; for measures by team, KPI examples.