Verifying an address matches its stated postal code before checkout
A mismatched postal code and city on an order form looks like a small typo until it turns into a delivery sent to the wrong part of the country entirely.
Building a store finder that ranks by distance
Most store finders are lists dressed up as maps. A shopper picks a state, then a city, then scrolls an alphabetical list that has nothing to do with which location is actually closest. A housewares chain with several dozen stores decided that was backward and rebuilt the tool around distance instead.
The rebuild started with input. Instead of asking a shopper to pick from dropdowns, the new finder let them start typing an address or a neighborhood and offered suggestions as they went, using /v1/autocomplete. That endpoint takes a partial address and returns candidate matches shaped for exactly this kind of live search box, so a shopper typing "Oak" gets a short list of real street and place names to pick from rather than typing a full address and hoping it matches something in the store's database.
Once a shopper picked a suggestion or allowed the page to use their approximate location, the chain needed one more thing: a starting point in coordinates. For shoppers who typed an address, that came from geocoding the chosen suggestion through /v1/forward, which returns a location match with latitude and longitude fields for a given address. For shoppers who did not type anything at all, the site fell back to /v1/ip, which reads the visitor's IP address and returns coordinates along with city and region, giving the finder a reasonable starting point without asking for anything.
The store list itself was geocoded once, in a single batch, since a bulk request counts each store address as one billed item. With every store's coordinates on file, ranking became arithmetic: compute distance from the shopper's point to every store, sort, show the closest five. No dropdown menus, no alphabetical scroll, no guessing which nearby town a store might be listed under.
The chain kept the tool honest about what it could promise. Because forward geocoding returns a match for what was typed rather than a guarantee that the address exists exactly as entered, the finder always paired the ranked list with a confirmation step, letting a shopper adjust their starting point if the suggested location was not quite right. That small check mattered more than any part of the ranking logic, since a wrong starting point makes even a perfect distance calculation useless.
Traffic on the finder was modest, a few thousand lookups a month across autocomplete and geocoding combined, which sat well within the free daily allowance that comes with a My Geocode key. The chain never needed to think about the pricing model day to day, since ordinary retail traffic on a store finder rarely approaches the point where prepaid credit or an Unlimited key becomes the better option, but knowing that headroom exists at €0.0001 a request meant a marketing push or a store opening announcement would not break anything.
A store finder is a small feature, but it is often a shopper's last step before deciding to drive somewhere, and getting the ranking right changes which store gets the visit. Details on both endpoints, including field shapes and rate limits, are at /docs/address-autocomplete/, /docs/forward-geocoding/ and /docs/rate-limits/.