How Do Postcodes Work? Turning an Address Into a Sorting Instruction

Sorting mail by reading full addresses is slow and requires clerks who know the geography, and the volume of post grew faster than the supply of such people. The first systems were coarse: London was divided into lettered districts in 1857, and several countries introduced district numbers for large cities in the following decades. Germany introduced a national numeric system in 1941, the United States introduced the five-digit zone improvement plan code in 1963, and the United Kingdom rolled out its alphanumeric postcode nationally between 1959 and 1974. The driver in each case was mechanisation: a machine can read a short code far more reliably than a handwritten address, and once the code is read the item can be routed automatically. The finer the code, the further into the delivery chain automation can reach, which is why later systems are more precise than earlier ones and why several countries have extended their codes since introduction.

The structures differ and the logic is similar, moving from coarse to fine as the code is read:

• •The British format has an outward code identifying the area and district, such as the letters for a postal town followed by a district number, and an inward code of a digit plus two letters identifying a sector and then a unit, with the unit typically covering around fifteen addresses, which makes it among the most precise national systems
• •The American five-digit code identifies a delivery area, with the optional four-digit extension narrowing it to a block or building, and the leading digit indicating a broad region of the country
• •Most European systems use four or five digits allocated broadly west to east or by region, with finer detail carried by the street address
• •The Netherlands uses four digits plus two letters, which identifies a street side within a block precisely enough that a house number alone completes the address
• •Several countries, including Ireland until 2015, managed without national postcodes entirely, relying on townland names and local knowledge
• •Some systems encode the sorting office rather than the destination, so codes for one town can belong to a neighbouring town's postal area, which is why postcode boundaries so often cross county lines

Postcodes turned out to be so convenient as a short geographic label that they escaped the postal system entirely. They are used to set insurance premiums, to allocate school places and health services, to target advertising, to validate credit card transactions, to calculate delivery charges and, through geodemographic classification, to assign whole marketing profiles to households. Each of these uses inherits assumptions that the postal system never made. Postcodes are not areas but collections of delivery points, so a unit may consist of addresses along a street with other addresses interleaved, and drawing a polygon around them is an approximation. They change, with new ones created for new developments and old ones retired, which breaks time series. They vary enormously in the number of people they contain, so treating them as comparable units biases any per-postcode statistic. And the phrase postcode lottery exists because using them to allocate services makes an administrative artefact into a determinant of what people receive.

Addressing systems are being rebuilt where postcodes do not work well. Large parts of the world have no formal street addressing at all, which obstructs deliveries, emergency response, banking and voter registration, and several approaches have emerged. Coordinate-based encodings compress a latitude and longitude into a short string, with open standards including plus codes generating a code from any point on earth without needing a database. Proprietary systems assign memorable word combinations to fixed squares, which are easy to say aloud and depend on a company's database remaining available. Several countries have built national address databases as public infrastructure, with India, Rwanda and others rolling out digital addressing schemes. The unresolved questions are governance rather than technology: whether an addressing system is a public utility or a private product, who may use it freely, and what happens to the millions of addresses encoded in a scheme if the organisation maintaining it changes its terms.

A postcode encodes a sorting and delivery instruction rather than a place, which is why its boundaries follow postal convenience and frequently cross administrative lines. Systems moved from coarse city districts to national codes as mechanised sorting arrived, with the British alphanumeric format narrowing to about fifteen addresses and the Dutch format identifying one side of a street. Because postcodes are sets of delivery points rather than areas, vary greatly in population and change over time, using them as neighbourhood units introduces real error.