Glossary

Every term we use across Learn and The Signal, defined plainly. 22 and counting.

A

Attribution: How you assign credit for a conversion to the touchpoints that led to it. Every model is a simplification: pick one, document it, and read it consistently.
Average order valuealso: AOV: The average amount spent per order: total revenue divided by number of orders. AOV thresholds are a common way to make value-based segments measurable, such as "customers in the top 20% by average order value".

Churn rate: The rate at which customers stop buying or paying over a given period. Small differences compound, so churn quietly sets the ceiling on growth.
Cohortalso: cohorts: A group fixed by a shared start event, most often first purchase in the same period. Membership never changes, which makes cohorts ideal for measuring how customer quality shifts over time, and the wrong tool for ongoing behaviour, which needs a segment.
Conversion rate: The share of people who take a defined next step, measured between two funnel stages (e.g. visit → signup). Always read it alongside the volume it is calculated from.
Customer acquisition cost (CAC): The fully-loaded cost of winning one new customer: ad spend, tools, and the people involved, divided by the number of customers acquired in the same period.
Customer lifetime value (LTV): The total gross profit you expect from a customer across the whole relationship, not a single order. It tells you how much you can afford to spend to acquire and keep them.
Customer segmentation: Dividing your existing customers into groups that share something meaningful, such as behaviour, value or circumstances, so each group can be treated differently. Distinct from market segmentation, which divides a whole market to shape strategy; customer segmentation shapes execution: who gets which message, offer and experience.

Direct to Consumer (DTC): Direct to Consumer (DTC) is a business model where a company sells its products or services straight to end customers, cutting out intermediaries like wholesalers, distributors, and third-party retailers. Traditionally, a manufacturer might sell to a distributor, who sells to a retailer, who sells to the consumer. A DTC brand skips those middle layers and owns the entire customer relationship, typically through its own e-commerce site, though it can also include brand-owned physical stores or subscription models.
Dynamic segmentalso: dynamic segments: A segment defined by a rule that is re-evaluated continuously, so membership updates as customer behaviour changes. The opposite of a static list. "No purchase in 180 days, valid consent" stays correct every day it runs, because whoever matches today is in today.

Gross margin: Revenue minus the direct cost of the goods sold. Margin-aware segmentation distinguishes customers who generate profit from those who merely generate revenue, which matters wherever discounts or returns eat into thin margins.

Holdout groupalso: holdout, control group: A slice of a segment deliberately excluded from a campaign so results can be compared against it. The difference between contacted and held-out customers is the campaign's incremental effect: what happened because of the campaign, rather than what would have happened anyway.

Ideal customer profile (ICP): A precise description of the accounts most likely to become your best, longest-lasting customers, used to focus targeting, qualification and spend.
Identity resolution: Recognising that records held in different platforms belong to the same person, usually by matching email addresses, customer IDs or hashed identifiers, then deduplicating. The foundation of a single customer view: without it, every platform holds a different partial version of the same customer.

Marketing-qualified lead (MQL): A lead that has shown enough intent, by an agreed and written definition, to be worth a salesperson's time. A shared definition matters more than the threshold.

Personaalso: personas: A fictional portrait of a typical customer, used to guide tone, creative and empathy. A persona is not a segment: you can't email a portrait. If a persona matters commercially, define the measurable segment that approximates it and check the two actually match.
Propensity scorealso: propensity: A modelled estimate of how likely a customer is to take a specific action: buy, churn, upgrade or respond. Propensity scores turn prediction into segments a campaign can act on, such as "high churn risk", before the behaviour happens.

RFMalso: RFM analysis: Recency, frequency, monetary value: a scoring approach that ranks customers by how recently they bought, how often, and how much they spend. A practical bridge from simple rule-based segments to scored segmentation.

Single customer viewalso: single audience view: One joined-up record per customer, combining what each platform knows: orders, email engagement, ad interactions, site behaviour and consent status. Segments built on a single customer view can use the whole picture rather than one platform's slice.
Static listalso: static lists: A snapshot export of customers who matched a rule on a particular day. It starts drifting immediately: members stop qualifying, new qualifiers are missing, and consents change. Fine for genuinely fixed groups, such as event attendees; wrong for anything defined by ongoing behaviour.
Suppression listalso: suppression segment, suppression segments: A maintained set of people a campaign must not reach: withdrawn consent, unsubscribes, complaints, recent purchasers of the promoted product, or contact fatigue. Suppression belongs inside a segment's definition, not as an afterthought at send time.