The Half-Life of an Address

Address and POI data decay is the steady loss of accuracy in location records as real buildings, businesses, and streets change faster than the databases describing them get updated. A storefront closes. A street gets renamed after a local hero. A new housing block gets a number nobody has entered anywhere yet. None of this is rare. It's the default condition of any geographic dataset, from the day it's published onward.

Physicists have a word for this kind of predictable decline: half-life, the time it takes for half of something to break down. Address and POI records behave the same way, just on a civic timescale instead of an atomic one. Every record has a rough half-life, a point past which the odds it's still accurate drop below what's usable for navigation, delivery, or compliance. The number varies by place and by data type, but the shape of the curve never does.

This piece leads with the European Union, because the EU has done more than most regions to formalize the fight against decay, through national address registries, the INSPIRE directive, and municipal addressing law. France, Germany, Switzerland, Belgium, and the Netherlands each handle it differently, and those differences are instructive. Where the pattern holds outside Europe too, such as the United States postal system, we'll say so.

What Decay Actually Means

In geospatial terms, a point of interest, or POI, refers to a discrete, named place on a map — a restaurant, a pharmacy, a charging station, a bank branch. Unlike a street address, which describes a fixed geographic position, a POI describes something operating at that position, and operations change constantly.

Decay, then, isn't a single failure mode. It's three separate clocks ticking at different speeds, all pointed at the same dataset.

Three Clocks That Never Stop

The first clock is human mobility. People relocate, companies move offices, households split and merge. Every move leaves a trail of stale records behind it in every system that held the old address — customer files, delivery platforms, emergency-service databases.

The second clock is the built environment. Streets get renamed. Blocks get renumbered when a development fills in a gap. New construction creates addresses that don't exist anywhere yet, sometimes for months. Utility and telecom infrastructure mapping runs into this constantly: a fiber crew works from a cadastral layer that's accurate on the parcel but blind to the three new buildings on it.

The third clock is commercial churn, and it's the fastest of the three. Consider a corner bakery in Lyon that closes on a Tuesday. By Friday, a phone-repair kiosk has taken the lease. The street address hasn't changed at all — the municipality's registry is still perfectly correct. But every POI layer built on top of that address, from a navigation app to a delivery platform's merchant list, is now wrong until somebody notices and corrects it. Address decay and POI decay are related but not identical problems, and conflating them is how freshness programs end up solving the wrong half of the issue.

Europe Built Registries for Exactly This Reason

Several European countries maintain authoritative, legally grounded address registries specifically to slow the first two clocks. They differ enough in design that a side-by-side view is more useful than a country-by-country narrative.

Country	Registry	Who maintains it	Update model
France	Base Adresse Nationale (BAN)	Communes feed local address bases; IGN and DINUM aggregate nationally	Continuous, with the national file refreshed daily
Netherlands	Basisregistratie Adressen en Gebouwen (BAG)	Municipalities, as legal "source holders"; Kadaster runs the national facility	Continuous; mandatory for public bodies since 2011
Switzerland	Gebäude- und Wohnungsregister (GWR/RegBL)	Municipalities and cantons supply data; federal publication via swisstopo and the statistics office	Municipal input is ongoing; the federal published snapshot refreshes on a slower, periodic cycle
Germany	ALKIS cadastral systems	Surveying authorities in each of the 16 Länder; no single open national base	Fragmented — cadence and access terms vary by state
Belgium	Regional address registers (Flanders, Wallonia, Brussels)	Regional and municipal authorities	Continuous locally; not unified into one national feed

The pattern is clear. France and the Netherlands treat addressing as a single national responsibility with daily or near-daily refresh (data.gouv.fr, 2026; Digitale Overheid, Netherlands). France's BAN describes itself as a government-wide reference database, built from the Bases Adresses Locales that each commune is now obligated to maintain and publish (data.gouv.fr, 2026). Germany and Belgium instead delegate addressing to subnational authorities, which keeps data accurate locally but makes any national-level freshness claim harder to verify.

Switzerland sits in between. Municipalities feed the federal building and dwelling register continuously, but the register was originally built for statistical census purposes rather than public address lookup, and full open access to street-level address data only arrived after a 2017 revision of the governing ordinance (OpenStreetMap Switzerland/GWR wiki). That history still shapes how current the published data feels in practice.

Across the whole EU, the INSPIRE directive sets a baseline for what an "address" even means in an interoperable sense, defining the theme as the location of properties based on identifiers such as road name, house number, and postal code (INSPIRE Addresses Data Specification, 2024). INSPIRE deliberately stops short of dictating how member states maintain their registers internally — it harmonizes the output format, not the update discipline behind it.

Why a Central Registry Still Isn't Enough

A national address registry answers "does this address exist, and where." It does not answer "what's operating there right now." That second question is the POI layer, and no European address registry is designed to track it.

This gap gets filled by a mix of commercial map providers, crowdsourced platforms, and field verification — each with a different lag. OpenStreetMap is instructive here because its coverage is measurable and public. A countrywide evaluation of Swiss address data found that community-contributed addresses in OpenStreetMap covered only 38.4% of the officially registered total as of 2018, even though the authoritative federal register existed the whole time (OpenStreetMap GWR coverage evaluation, 2018). The registry wasn't the bottleneck. Getting that registry's contents reflected, maintained, and cross-checked against ground truth in a usable map was. For a deeper look at how POI records are actually built and checked in the first place, see our explainer on POI data collection and verification.

The same gap shows up in logistics and navigation platforms outside Europe. A courier routed to a "last known good" address finds a locked gate where a loading dock used to be. A rideshare app sends a driver to a building entrance that moved when a block was renumbered. None of these are registry failures. They're freshness failures one layer above the registry, in the operational data that depends on it.

How Freshness Gets Maintained in Practice

Slowing decay takes a combination of automated signals and physical verification, and the mix shifts depending on how fast a given category churns.

Automated change detection does the first pass. Comparing satellite or street-level imagery over time flags new rooftops, demolished buildings, and altered street layouts well before any human re-visits the site. Transaction and signal data — a change of registered business owner, a utility hookup, a new VAT registration — can flag commercial churn almost as fast as it happens, though these signals tell you something changed, not what it changed into.

Field verification closes that remaining gap. Someone, or increasingly a structured field team working from a defined route and checklist, physically confirms what's actually at a location: is the business open, is the signage current, does the address match what's posted. This is slower and more expensive than automated detection, but it's the only method that catches the cases automation misses entirely — a business that's open but unlisted, a sign that's wrong, a delivery entrance that moved without any permit filing to flag it.

Teampl has run business-status verification programs photographing hundreds of points of interest per day for clients who needed ground truth on which locations were actually open — the kind of question a registry, by design, can't answer on its own. Photographic evidence attached to each record matters here as much as the record itself, because it lets a client audit the verification rather than just trust it. Our piece on turning field photos into structured proof goes into how that evidence gets captured and standardized at scale.

Postal systems outside Europe offer a useful cross-check on how seriously an industry can take freshness once the cost of staleness is made explicit. The US Postal Service's Move Update standard requires that 95% of addresses in a mailing be updated within 95 days, enforced through licensed access to a change-of-address database covering roughly 160 million records (Mailing.com, 2026). That's a hard, auditable freshness threshold tied directly to postage pricing — the kind of enforcement mechanism most POI datasets simply don't have.

Measuring Decay Before You Can Fix It

You can't manage a freshness problem you haven't measured. Cross-industry data quality research outside the geospatial field gives a sense of scale: B2B contact databases are estimated to lose accuracy at rates between roughly 22.5% and 70% annually depending on the field type, with physical address changes alone affecting a meaningful share of records within twelve months (ZoomInfo Pipeline, 2026). Those figures come from CRM and marketing data, not address registries, but the underlying mechanism — real-world change outpacing database maintenance — is identical.

The practical takeaway isn't the exact percentage. It's that decay compounds continuously rather than in discrete jumps, and a dataset cleaned once a year is already meaningfully wrong by month three. A re-verification cadence has to be matched to the churn rate of the category being tracked. A fuel station's address rarely changes; its pricing and opening hours can change weekly. A shopping mall's floor plan shifts on a tenant cycle measured in months. Treating all of that with one blanket "annual refresh" schedule is how freshness programs quietly fail without anyone noticing until a customer complaint does the noticing for them.

The Limits of Any Freshness Program

No verification cadence eliminates decay entirely, and it's worth being honest about where the limits sit. Rural and low-density areas churn slowly, so intensive re-surveying there often isn't worth the cost. Seasonal businesses — a beach kiosk, a Christmas market stall — decay and reappear on a predictable cycle that a generic freshness model handles badly. Home-based businesses with no storefront are frequently invisible to both registries and street-level verification alike. Privacy regulation adds another layer specific to the EU. Field verification that involves photographing a storefront is generally fine; verification that captures identifiable individuals or private residential detail runs into GDPR constraints that don't apply in the same way in every market outside Europe. Any freshness program operating across EU borders has to build that distinction into its field methodology from the start, not retrofit it later.

So What Actually Keeps a Dataset Fresh

Is a registry enough on its own? Not for anything that depends on knowing what's happening at an address, rather than just that the address exists. Is crowdsourcing enough? Only where contributor density is high, and even strong examples show real coverage gaps years after launch. Is automation enough? It catches the obvious changes and misses the quiet ones — the shop that's open but wrong on the sign, the entrance that moved without a permit. What's left, every time, is some form of physical re-checking, scheduled to match how fast a given place actually changes. That's not a failure of technology. It's just what keeping pace with a half-life requires.

Teampl runs field data collection programs — egocentric video, in-store audits, mystery visits, GIS surveys — designed around a specific research question, not a generic panel. If this topic touches a program you're planning, tell us what you're trying to learn.

Start a Project

Every address and POI dataset starts decaying the moment it's published. The only real question is whether your maintenance cadence outruns its half-life, or falls behind it.

Frequently Asked Questions