What is a postcode?

A postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. is a routing instruction, not a geographic area. It tells a postal service how to sort and deliver mail. It does not tell a geocoder where a building is, what municipalitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. it belongs to, or what polygon contains it. Confusing these two things is the most common error in address geocodinggeocodingThe process of converting an address into geographic coordinates (latitude and longitude). Mailwoman geocodes in a multi-tier cascade: exact address-point match → street interpolation → locality centroid. Each tier is progressively coarser but more widely available. — and the source of a surprising fraction of production bugs.

What postcodes encode

Every postal system uses postcodespostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. differently, but the common threadthreadA parallel workstream within a release. Threads compose; they are not sequential milestones like phases. is hierarchical routing:

CountrycountryThe top-level address component (an ISO country). Closed-vocabulary, so it is best handled by a deterministic matcher feeding a proposal rather than a retrained model head.	Format	What each level means
US (ZIP)	12345-6789	5-digit: sectional center → post office. +4: block faceblock faceOne side of a street between two intersections. Several postal systems specify delivery to block-face granularity — e.g. the USPS ZIP+4 extension. or building. Last 2 (delivery point): mailbox.
UK	SW1A 1AA	Area (SW) → district (1A) → sector (1) → unitunitA subdivision of a building — apartment, suite, floor — that refines a street address. Mailwoman's unit component; a designator plus identifier forms a subpremise. (AA). Covers ~15 addresses.
France	75008	First 2 digits: département (75 = Paris). Last 3: commune-level routing.
Canada	K1A 0B1	Forward sortation area (K1A) → local delivery unitunitA subdivision of a building — apartment, suite, floor — that refines a street address. Mailwoman's unit component; a designator plus identifier forms a subpremise. (0B1). One side of one block faceblock faceOne side of a street between two intersections. Several postal systems specify delivery to block-face granularity — e.g. the USPS ZIP+4 extension..
Ireland (Eircode)	D02 X285	Unique per delivery point. The postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. IS the address.
Japan	100-0001	3-digit regional + 4-digit local. Coarser than US +4; municipalitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. is the finer unitunitA subdivision of a building — apartment, suite, floor — that refines a street address. Mailwoman's unit component; a designator plus identifier forms a subpremise..

The hierarchy is for sorting mail, not for describing geography. A US 5-digit ZIP gets a letter to the right post office — that's it. The carrier who delivers it knows which streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. are in their route. The ZIP doesn't encode that information explicitly.

Why postcodes are not polygons

The US Postal Service does not publish ZIP code boundaries. The ZIP code is a route, not a shape. A single ZIP can contain:

• Multiple municipalities (a rural route that crosses town lines).
• Gaps (a building served by a different post office than the surrounding area).
• Overlaps (different carriers serve different sides of the same streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.).
• Non-geographic assignments (PO boxesPO boxA numbered mailbox at a post office used as a delivery address instead of a physical street location. Mailwoman tags it as the po_box component; structurally the same family as a subpremise., military APO/FPOAPO/FPO (Army Post Office / Fleet Post Office). US military postal addresses for personnel stationed overseas or aboard ship. They route through military postal hubs rather than geographic locations, so they break geographic parsing assumptions. addresses, unique large-volume mailers).

When you see a ZIP code displayed as a polygon on a map, you are almost certainly looking at a ZIP Code Tabulation Area (ZCTAZCTA (ZIP Code Tabulation Area). The US Census Bureau's polygon approximation of a ZIP code, generalized from census blocks rather than USPS delivery routes. Explicitly not USPS ground truth and often wrong in rural areas.) — a statistical approximation produced by the US Census Bureau. ZCTAsZCTA (ZIP Code Tabulation Area). The US Census Bureau's polygon approximation of a ZIP code, generalized from census blocks rather than USPS delivery routes. Explicitly not USPS ground truth and often wrong in rural areas. are generalized from census blocks, not from USPS delivery routes. USPS explicitly disclaims them. The Census Bureau's own documentation says ZCTAsZCTA (ZIP Code Tabulation Area). The US Census Bureau's polygon approximation of a ZIP code, generalized from census blocks rather than USPS delivery routes. Explicitly not USPS ground truth and often wrong in rural areas. "are not exact representationshidden stateThe model's internal vector for a token after the encoder has mixed in surrounding context. The contextualized representation the classifier head reads to assign a label. of USPS ZIP Code service areas."

This matters because the errors are systematic:

• Rural areas are the worst. A rural ZIP may cover hundreds of square miles. The ZCTAZCTA (ZIP Code Tabulation Area). The US Census Bureau's polygon approximation of a ZIP code, generalized from census blocks rather than USPS delivery routes. Explicitly not USPS ground truth and often wrong in rural areas. centroid can be 10+ miles from any actual delivery point. GeocodinggeocodingThe process of converting an address into geographic coordinates (latitude and longitude). Mailwoman geocodes in a multi-tier cascade: exact address-point match → street interpolation → locality centroid. Each tier is progressively coarser but more widely available. a rural address to its ZIP centroid routinely produces errors of 5-15 miles.
• New development is invisible. When a new subdivision is built and assigned a new ZIP+4, the ZCTAZCTA (ZIP Code Tabulation Area). The US Census Bureau's polygon approximation of a ZIP code, generalized from census blocks rather than USPS delivery routes. Explicitly not USPS ground truth and often wrong in rural areas. won't reflect it until the next decennial census — up to 10 years.
• PO boxesPO boxA numbered mailbox at a post office used as a delivery address instead of a physical street location. Mailwoman tags it as the po_box component; structurally the same family as a subpremise. cluster at post offices. A ZIP code used primarily for PO boxesPO boxA numbered mailbox at a post office used as a delivery address instead of a physical street location. Mailwoman tags it as the po_box component; structurally the same family as a subpremise. (common in small towns) will produce a centroid at the post office building, not at the residences the PO boxPO boxA numbered mailbox at a post office used as a delivery address instead of a physical street location. Mailwoman tags it as the po_box component; structurally the same family as a subpremise. holders live in.

What to do with a postcode in a geocoder

A geocoder should treat a postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. as supplementary evidence, not as a primary locator:

1. Use the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. to narrow the resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon.'s search space (limit gazetteergazetteerA geographical index that maps place names and postcodes to real-world coordinates. Mailwoman uses a custom-built Who's On First (WOF) SQLite database as its gazetteer — the 'atlas' half of the grammar/atlas architecture. lookups to places within the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one.'s rough regionregionThe first-level administrative subdivision of a country — a US state, a French region, a province. The component between country and locality.).
2. Use the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. to validate the parseaddress parsingThe process of decomposing a free-text postal address string into structured components — house number, street name, locality, region, postcode, and country — so a geocoder can resolve them to coordinates. (if the parsed localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. and the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one.'s known municipalitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. disagree, flag the ambiguity).
3. Do NOT return coordinates from a postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. alone unless the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. is a delivery-point code (Eircode, UK postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. with full unitunitA subdivision of a building — apartment, suite, floor — that refines a street address. Mailwoman's unit component; a designator plus identifier forms a subpremise., US ZIP+4+2). Even then, flag it as "postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one.-level resolution," not "address-level."

A user who types 90210 into a geocoder is not asking "where is ZIP code 90210?" They are asking "where is the place I associate with 90210?" The correct answer is "Beverly Hills, CA" with a note that the ZIP covers multiple municipalities and the precise location cannot be determined from the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. alone. The incorrect answer is a pin dropped at the ZCTAZCTA (ZIP Code Tabulation Area). The US Census Bureau's polygon approximation of a ZIP code, generalized from census blocks rather than USPS delivery routes. Explicitly not USPS ground truth and often wrong in rural areas. centroid with the labelcomponent tagOne of the 33 labels in Mailwoman's address schema — street, locality, region, postcode, house_number, unit, po_box, country, venue, intersection, and others. Each parsed span carries exactly one component tag. "Beverly Hills."

Why the postal system does this

PostcodespostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. are designed for sorting machines and carrier routescarrier routeThe delivery path one postal carrier covers on a shift. A ZIP code is fundamentally a set of carrier routes, not a geographic polygon, which is why ZIP boundaries are fuzzy and shift over time., not for maps. Every design choice follows from this:

• They change. When a carrier retires and routes are redrawn, postcodespostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. can split or merge. The USPS changes approximately 5,000 ZIP codes per year (additions, deletions, boundary adjustments).
• They prioritize delivery efficiency over geographic coherencecoherenceThe property of a parse whose resolved places form a consistent geographic hierarchy — the resolved locality really does sit inside the resolved region.. A postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. boundary that follows a creek saves the carrier a bridge crossing. It makes no geographic sense on a map, but it saves 20 minutes of driving per day.
• They serve multiple delivery modes. The same ZIP code can include streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. addresses, PO boxesPO boxA numbered mailbox at a post office used as a delivery address instead of a physical street location. Mailwoman tags it as the po_box component; structurally the same family as a subpremise., and general-delivery addresses, all routed to different places at the same post office.

A postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. is built to move mail, not to locate things. The geocoder's job is to treat it as a spatial hint for the resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon., not as a spatial fact on its own.

See also

• What is a ZIP Code and how is it structured? — the US system in detail
• The database fallacy — why there is no master list of postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one.-to-place mappings
• How mail delivery actually works — the system postcodespostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. were designed for