diff --git a/Sources/PDFArchiverWebsite/Localization/SiteStrings+English.swift b/Sources/PDFArchiverWebsite/Localization/SiteStrings+English.swift
index 420e462b..32faa06a 100644
--- a/Sources/PDFArchiverWebsite/Localization/SiteStrings+English.swift
+++ b/Sources/PDFArchiverWebsite/Localization/SiteStrings+English.swift
@@ -26,7 +26,7 @@ extension SiteStrings {
         steps: [
             Step(
                 tag: "1",
-                title: "Scan it.",
+                title: "Scan\u{00A0}it.",
                 lead: "Point your iPhone at the page.",
                 body: """
                 The scan lands in your inbox, cropped and straight. Text recognition runs on your \
@@ -35,7 +35,7 @@ extension SiteStrings {
             ),
             Step(
                 tag: "2",
-                title: "Tag it.",
+                title: "Tag\u{00A0}it.",
                 lead: "Date, description, tags.",
                 body: """
                 Apple Intelligence suggests all three, on your device. On the Mac, ↹ moves \
@@ -45,7 +45,7 @@ extension SiteStrings {
             ),
             Step(
                 tag: "3",
-                title: "Find it.",
+                title: "Find\u{00A0}it.",
                 lead: "Search a tag, or the text inside.",
                 body: """
                 It works in PDF Archiver, in Spotlight, in the Finder, and in whatever file \
@@ -69,7 +69,7 @@ extension SiteStrings {
         spotlightImagePath: "/assets/img/screenshots/archive-iphone-en.png",
         spotlightImageAlt: "The PDF Archiver archive on an iPhone, listing documents with their dates and tags",
 
-        keepTitle: "Keep it.",
+        keepTitle: "Keep\u{00A0}it.",
         keepLead: """
         Your archive is a folder of PDFs, sorted by year. This is the filename from above, at rest.
         """,
diff --git a/Sources/PDFArchiverWebsite/Localization/SiteStrings+German.swift b/Sources/PDFArchiverWebsite/Localization/SiteStrings+German.swift
index 12756190..7f2b68ee 100644
--- a/Sources/PDFArchiverWebsite/Localization/SiteStrings+German.swift
+++ b/Sources/PDFArchiverWebsite/Localization/SiteStrings+German.swift
@@ -26,7 +26,7 @@ extension SiteStrings {
         steps: [
             Step(
                 tag: "1",
-                title: "Scan it.",
+                title: "Scan\u{00A0}it.",
                 lead: "Halte dein iPhone über die Seite.",
                 body: """
                 Der Scan landet zugeschnitten und gerade in deiner Inbox. Die Texterkennung läuft \
@@ -35,7 +35,7 @@ extension SiteStrings {
             ),
             Step(
                 tag: "2",
-                title: "Tag it.",
+                title: "Tag\u{00A0}it.",
                 lead: "Datum, Beschreibung, Tags.",
                 body: """
                 Apple Intelligence schlägt dir alle drei vor, auf dem Gerät. Am Mac springst du \
@@ -45,7 +45,7 @@ extension SiteStrings {
             ),
             Step(
                 tag: "3",
-                title: "Find it.",
+                title: "Find\u{00A0}it.",
                 lead: "Such nach einem Tag oder dem Text im Dokument.",
                 body: """
                 Das funktioniert in PDF Archiver, in Spotlight, im Finder und in jedem \
@@ -69,7 +69,7 @@ extension SiteStrings {
         spotlightImagePath: "/assets/img/screenshots/archive-iphone-de.png",
         spotlightImageAlt: "Das Archiv von PDF Archiver auf einem iPhone, mit Dokumenten samt Datum und Tags",
 
-        keepTitle: "Keep it.",
+        keepTitle: "Keep\u{00A0}it.",
         keepLead: """
         Dein Archiv ist ein Ordner voller PDFs, nach Jahren sortiert. Das ist der Dateiname von \
         oben, zur Ruhe gekommen.
diff --git a/Sources/PDFArchiverWebsite/Localization/SiteStrings.swift b/Sources/PDFArchiverWebsite/Localization/SiteStrings.swift
index ea6e7298..a3b10004 100644
--- a/Sources/PDFArchiverWebsite/Localization/SiteStrings.swift
+++ b/Sources/PDFArchiverWebsite/Localization/SiteStrings.swift
@@ -2,6 +2,9 @@
 ///
 /// Page titles and body copy of the content pages are not in here — they live in the front matter
 /// and body of the Markdown files.
+///
+/// Every beat of the claim — `Scan it.`, `Tag it.`, `Find it.`, `Keep it.` — holds a non-breaking
+/// space before `it.`, so a narrow column can never split one across two lines.
 struct SiteStrings {
     /// One of the three steps the workflow section is built from.
     struct Step {
@@ -56,8 +59,8 @@ struct SiteStrings {
 
     let heroEyebrow: String
     /// The claim is not translated. Its last beat is set in the accent colour.
-    let heroTitleLead = "Scan it. Tag it."
-    let heroTitleAccent = "Find it."
+    let heroTitleLead = "Scan\u{00A0}it. Tag\u{00A0}it."
+    let heroTitleAccent = "Find\u{00A0}it."
     let heroLead: String
     let heroSecondaryCTA: String
     let trustLine: String