Building a multilingual multisite that serves millions

The National Hypertension Control Initiative’s site had to do several jobs at once: publish health education in multiple languages, deliver structured training to professionals, and help patients find care near them, all reliably enough to carry millions of visitors under a high-visibility national brand. That is a lot of surface area for one platform, and the architecture is what kept it from becoming a lot of chaos.

One site per language, one theme for all of them

The platform was a WordPress multisite: one site per language, one shared custom theme. That split is the whole trick. Editorial teams work in their own language without stepping on each other, but a fix to a Gutenberg block or a build tweak ships once and lands everywhere.

The alternative, a single site bolting translations onto every post, turns every content change into a coordination problem. Multisite made the languages independent and the code shared, which is exactly the ratio you want.

Give editors blocks, not HTML

Content changed constantly, so editors could not depend on a developer for layout. The theme was built on ACF-driven Gutenberg blocks: a fixed library of components an editor composes freely, with an SCSS/Gulp pipeline behind them. Editors got range; the design system stayed intact.

The feature that justified the platform

The piece I am proudest of was a health-center directory. A visitor enters a zip code and radius; the search geocodes it and returns participating centers as both a list and pins on a Google Maps view:

$centers = query_centers_within( $zip, $radius_miles );   // geocode + distance filter
wp_send_json( [
    'list'    => array_map( 'format_center_card', $centers ),
    'markers' => array_map( 'to_map_marker', $centers ),
] );

That single search was the bridge between the program’s content and actual treatment, the step most awareness campaigns never build.

Reliability was the real feature

The site ran in multiple languages, without drama. At this scale the win is not a clever feature; it is that a national brand’s platform simply kept working. Boring, at scale, is the hardest thing to build.