Which postcode API should I choose: PostcodeAPI.nu, PostNL or something else?
For most SME applications, PostcodeAPI.nu is a good starting point: simple onboarding, a free tier for low volumes and a clean REST API. Postcode.nl is more commercially oriented and has a more extensive dataset (including ownership information and historical address versions), which is interesting if you need more than just address autocomplete. The PostNL Postcode API is PostNL's official offering, handy if you already ship with PostNL and want to match address data with how PostNL uses it itself. For enterprise volumes, or when you need direct BAG compliance, we would rather integrate directly with the Basisregistratie Adressen en Gebouwen via PDOK. We advise per project, as it depends heavily on volume, use case and the other systems the data needs to go to.
Can a postcode API also be used offline?
Yes. For applications where working offline matters, such as field-service apps, inspections or deliveries in rural areas, we build a hybrid. A local subset of the BAG (the Dutch official address registry) is downloaded to the device periodically. While there is a connection, the app uses the online API; if it drops, the app quietly switches to the local dataset. We determine the sync frequency and the size of the subset based on your working area and device storage. For a fleet of tablets across the Netherlands, a full-NL BAG subset is perfectly workable; for mobile-first apps with storage constraints, we split it up by region. When the device is back online, changes (newly created addresses, location adjustments) are automatically sent back to the server, with a clean audit trail of what was entered when, in which offline session.
What determines whether a postcode API becomes expensive?
Mainly the number of API calls. Many teams send a lookup call on every keystroke, which adds up quickly for a busy web shop. With debouncing (waiting until the user pauses typing), caching at postcode level and a second layer at house-number level, we can often cut the number of calls substantially. Another frequently overlooked factor is bot and crawler traffic that hits lookup endpoints without ever resulting in an order. For that, we set up per-IP rate limiting and CAPTCHA fallbacks on heavily used endpoints. At very high volumes, your own BAG export becomes interesting: it requires an upfront investment, but after that there are no variable costs and you are not tied to a single provider's pricing.
How do we handle international addresses?
Dutch postcode APIs only cover the Netherlands. For cross-border customers, such as international web shops, logistics and SaaS, we combine a Dutch provider for Dutch addresses with the Google Address Validation API or Loqate for the rest of the world. We make that choice based on the countries that actually appear in your order flow and on privacy considerations (Loqate is a European supplier, while Google processes data in its own data centres). Address normalisation ensures the end result in your database always has the same fields, regardless of the source. That makes a big difference for downstream integrations with carriers, invoicing software and your CRM.
What is the difference between using the BAG directly and using a postcode provider?
The BAG (Basisregistratie Adressen en Gebouwen, the Dutch register of addresses and buildings) is the official Dutch registry, managed by the Kadaster and available via PDOK. Postcode providers also use BAG data as their basis, but put a user-friendly API on top with fast autocomplete endpoints and a commercial SLA. For most applications, a provider is quicker and simpler to integrate. Connecting directly to the BAG via PDOK is worthwhile where there are compliance requirements (real estate, municipalities, housing associations), at very high volumes where provider costs become significant, or if you already use other PDOK data, such as parcels, buildings or WOZ (property valuation) data, and it makes sense to draw from the same source.
What if the chosen API goes down or hits its rate limit?
We build for that structurally. A retry mechanism with exponential backoff absorbs short hiccups. In the event of a longer outage, the application falls back to manual entry or a second provider. For mission-critical applications, we often opt for two providers in parallel (the first hits its rate limit, the second takes over) or a hybrid with a local fallback. We actively monitor response times and error rates so you see a problem before your users notice it. Alerts go through your existing channel, whether that is Slack, Teams or your own monitoring platform.
Do you work with our in-house IT team?
Almost always. We deliver not only working code, but also documentation, an incident runbook and knowledge transfer in the final sprint. Your team can then maintain the integration itself, or we can do it under an ongoing support contract. Often it's a mix: you manage it day to day, and we're available for extensions, provider changes and BAG version upgrades. The code lives in your own repository, following the same standards (linting, tests, CI) as the rest of your codebase — no vendor lock-in and no separate stack to learn.