Where does segmentation collide?

The division of a heterogeneous market into homogeneous subgroups of customers who respond similarly to a given offer, so that each group can be targeted with a tailored product, price, or message. Smith's 1956 Product differentiation and market segmentation as alternative marketing strategies set the term against differentiation, treating segmentation as a rational response to a demand curve made of many smaller, distinct demand schedules. The referent is a partition of buyers, drawn on demographic, psychographic, or behavioural bases, and it has nothing to do with a body plan, an image raster, or a stream of speech.[1]

Unit

A subgroup of a market, defined by shared response to an offer

Recovery means

Not applicable

Measured by

Identified by clustering survey and purchase data on demographic, psychographic, and behavioural variables, and judged by whether the segments are substantial, distinct, and reachable

The division of an animal body plan into a series of repeated units, or metameres, laid down along the head-to-tail axis during embryonic development, and the hierarchy of genes that establishes them. Nüsslein-Volhard and Wieschaus's 1980 Mutations affecting segment number and polarity in Drosophila identified the segmentation genes that partition the fly embryo, sorting them into the gap, pair-rule, and segment-polarity classes that act in sequence. The referent is a spatial pattern of body parts and the genetic program behind it, not a slice of a market, a labelled image, or a parse of speech.[2]

Unit

A repeated body segment, or metamere, along the embryonic axis

Recovery means

Not applicable

Measured by

Scored in embryos by counting and staining segments, and dissected genetically through mutations that delete or duplicate segments

The partitioning of a digital image into regions or objects that are internally uniform by some criterion, so that each pixel is assigned to a meaningful part of the scene. Haralick and Shapiro's 1985 Image segmentation techniques surveyed the classical approaches, from thresholding and clustering to region growing and edge-based splitting and merging, and set out what makes a good segmentation. The referent is a labelling of pixels, whether by region, semantic class, or object instance, and it shares nothing with a customer group, an embryo's segments, or a spoken word boundary.[3]

Unit

A region or per-pixel label covering part of an image

Recovery means

Not applicable

Measured by

Computed by thresholding, clustering, region growing, or learned models, and evaluated against a ground-truth mask by overlap scores such as intersection over union

The process of locating word or phoneme boundaries in the continuous acoustic stream of speech, which carries few reliable silences between words and so must be parsed by the listener. Saffran, Aslin, and Newport's 1996 Statistical learning by 8-month-old infants showed that infants can find word boundaries from the transitional probabilities between syllables alone, after only brief exposure to a spoken stream. The referent is the listener's parsing of sound into linguistic units, not a division of a market, a body, or an image.[4]

Unit

A word or phoneme boundary recovered from the speech stream

Recovery means

Not applicable

Measured by

Probed by tracking listening or looking times to familiar versus novel sequences, and modelled by transitional-probability and prosodic cues to boundaries

In labor economics, segmentation refers to the division of the labour market into primary and secondary segments with different wages, job stability and rules of advancement. This meaning comes from Doeringer (1971) in Internal Labor Markets and Manpower Analysis.[5]

Unit

Recovery means

Measured by