Organization leaderboard: go-live (record oversized orgs, rebrand, recent-lookup chips, show toggle)

src/components/MetricBar.tsx

@@ -41,7 +41,7 @@ export function MetricBar() {
 
 	let modes: LeaderMode[] | null = null;
 	if (routeId === "/") {
-		// The company board (?kind=org) ranks by commits only — no metric to pick yet.
+		// The organization board (?kind=org) ranks by commits only — no metric to pick yet.
 		modes = boardKind === "org" ? null : LEADER_MODES;
 	} else if (routeId === "/$user") {
 		if (lookup?.kind === "org") {

src/components/OrgView.tsx

@@ -6,7 +6,7 @@ import type { OrgMemberEntry, OrgResult } from "#/lib/org";
 import type { OrgSummary } from "#/lib/org-cache";
 
 /**
- * Org ("company") views rendered by the /$user route when a login resolves to an organization
+ * Organization views rendered by the /$user route when a login resolves to an organization
  * (GitHub logins share one namespace, so /paritytech IS the org page). Header + lifetime totals
  * of the members' contributions *to this org*, plus the building/error states. No chart yet:
  * orgs have no monthly data until the background worker lands (issue #84 follow-up).
@@ -153,6 +153,27 @@ export function OrgResultView({
 		);
 	}
 
+	// Valid but too large to build on demand — recorded and queued for the background worker.
+	// A gentle notice, not a failure: no Retry (retrying can't help), just "check back later".
+	if (result.indexing) {
+		return (
+			<main className="mx-auto max-w-md px-6 py-24 text-center">
+				<h1 className="text-xl font-semibold">
+					We’re still indexing {result.login}
+				</h1>
+				<p className="mt-3 text-sm text-muted-foreground">{result.indexing}</p>
+				<div className="mt-6">
+					<Link
+						to="/"
+						className="text-sm text-muted-foreground hover:underline"
+					>
+						← Try another lookup
+					</Link>
+				</div>
+			</main>
+		);
+	}
+
 	// Hard failure, or polling gave up on a stalled build.
 	const message = result.error
 		? result.error
@@ -247,7 +268,7 @@ function LoadedOrg({
 				repositories, as attributed by GitHub. Private members and private
 				contributions aren’t included.{" "}
 				<Link
-					to="/company/$slug"
+					to="/organizations/$slug"
 					params={{ slug: "stats" }}
 					className="whitespace-nowrap underline decoration-dotted underline-offset-2 hover:text-foreground"
 				>
@@ -285,7 +306,7 @@ function MemberBoard({
 					Public members ranked by their lifetime commits to {org.login}’s
 					repositories.{" "}
 					<Link
-						to="/company/$slug"
+						to="/organizations/$slug"
 						params={{ slug: "members" }}
 						className="whitespace-nowrap underline decoration-dotted underline-offset-2 hover:text-foreground"
 					>

src/content/company/members.mdx → src/content/organizations/members.mdx

@@ -6,7 +6,7 @@ updatedAt: "2026-07-08"
 order: 2
 ---
 
-The member leaderboard on a company page lists the organization's **public
+The member leaderboard on an organization page lists the organization's **public
 members** — and that's almost always the answer to "why am I not on it":
 **GitHub makes organization membership private by default.** Unless you've
 explicitly published yours, nobody outside the organization (including us) can
@@ -21,24 +21,24 @@ see that you're a member at all.
 
 GitHub only lets *you* do this for yourself — an org admin can't publish your
 membership for you. Once it's public, you'll appear the next time the
-company's data is refreshed, and your contributions start counting toward the
-[company totals](/company/stats) too.
+organization's data is refreshed, and your contributions start counting toward
+the [organization totals](/organizations/stats) too.
 
 ## Other reasons someone is missing
 
 - **They're not a member.** Contributing to an org's repositories — even a
   lot — doesn't make someone a member. Outside contributors don't appear on
   the member list and don't count toward the totals.
-- **The list hasn't refreshed yet.** Membership is captured when the company
-  is first looked up and re-synced periodically, so a freshly-publicized
-  membership can take a while to show up.
+- **The list hasn't refreshed yet.** Membership is captured when the
+  organization is first looked up and re-synced periodically, so a
+  freshly-publicized membership can take a while to show up.
 - **The account is suspended on commit-history.** Profiles under review are
   hidden from every leaderboard, member lists included.
 
 ## Why someone shows zero
 
 Being on the list with a **0** is normal and not a bug. The numbers are
 contributions **to this organization's public repositories** only — see
-[how company stats are calculated](/company/stats). A member shows zero when
-their work in the org happens in private repositories, under an email not
-linked to their GitHub account, or simply in other places entirely.
+[how organization stats are calculated](/organizations/stats). A member shows
+zero when their work in the org happens in private repositories, under an email
+not linked to their GitHub account, or simply in other places entirely.

src/content/company/stats.mdx → src/content/organizations/stats.mdx

@@ -1,15 +1,15 @@
 ---
-title: "How company stats are calculated"
-description: "Where a company's commits, pull requests, reviews and issues come from: each public member's contributions to that organization's repositories, summed — and what that number deliberately leaves out."
+title: "How organization stats are calculated"
+description: "Where an organization's commits, pull requests, reviews and issues come from: each public member's contributions to that organization's repositories, summed — and what that number deliberately leaves out."
 publishedAt: "2026-07-08"
 updatedAt: "2026-07-08"
 order: 1
 ---
 
-A company page answers a narrower question than it might look like: **how much
-have this organization's public members publicly contributed to the
+An organization page answers a narrower question than it might look like: **how
+much have this organization's public members publicly contributed to the
 organization's own repositories?** Every number on the page — the headline
-stats, the member leaderboard, the company's position on the company
+stats, the member leaderboard, the organization's position on the organization
 leaderboard — is built from that one measure.
 
 ## The calculation
@@ -24,8 +24,8 @@ Within that scope, GitHub's usual contribution rules apply, the same ones your
 personal profile uses: commits count on the default branch, under an email
 linked to the account, in non-fork repositories.
 
-The company totals are simply those per-member numbers **summed**. The member
-leaderboard shows the same rows the sum is made of, so the two always
+The organization totals are simply those per-member numbers **summed**. The
+member leaderboard shows the same rows the sum is made of, so the two always
 reconcile.
 
 ## What the number is *not*
@@ -35,32 +35,33 @@ things are excluded, and they can make the totals smaller than you'd expect:
 
 - **Private members.** GitHub keeps organization membership private by
   default, and we can only see members who made theirs public. An active
-  company whose people all keep membership private shows a total of zero —
-  see [why am I not on this member list?](/company/members).
+  organization whose people all keep membership private shows a total of zero —
+  see [why am I not on this member list?](/organizations/members).
 - **Outside contributors.** Someone who sends pull requests to the org's
   repos without being a member isn't counted. The measure is about the
-  company's people, not its repos' traffic.
+  organization's people, not its repos' traffic.
 - **Private repositories.** We only see public activity. Work in the org's
   private repos is invisible to us, even for public members.
 
-So treat the company numbers as the org's **public footprint**: what its
+So treat the organization numbers as the org's **public footprint**: what its
 publicly-listed people have visibly built there.
 
 ## How this differs from a personal profile
 
 Your personal page counts your public contributions **across all of GitHub**.
-A company page counts only what landed **in that org's repositories**. That's
-why your number on a member leaderboard is always less than or equal to the
-number on your own profile — often much less, if most of your work happens
+An organization page counts only what landed **in that org's repositories**.
+That's why your number on a member leaderboard is always less than or equal to
+the number on your own profile — often much less, if most of your work happens
 elsewhere.
 
 ## Freshness and quirks
 
 - Totals cover every **completed month** up to now; the current in-progress
   month rolls in once it ends (the same convention as personal profiles).
-- A company's numbers are collected when it's first looked up and refreshed
-  periodically after that, so very recent activity can take a while to appear.
-- Company pages don't have a commit chart yet — we currently store lifetime
+- An organization's numbers are collected when it's first looked up and
+  refreshed periodically after that, so very recent activity can take a while
+  to appear.
+- Organization pages don't have a commit chart yet — we currently store lifetime
   totals per member, and the per-month history that powers the chart is
   collected in a slower background pass.
 - In rare cases GitHub's API cannot serve a specific slice of a member's

src/lib/cache.ts

@@ -435,7 +435,7 @@ async function persistEntity(
 	}
 }
 
-async function recordLookup(database: DB, id: string, now: Date) {
+export async function recordLookup(database: DB, id: string, now: Date) {
 	try {
 		await database.insert(lookups).values({ entityId: id, searchedAt: now });
 	} catch {

src/lib/commit-history.ts

@@ -115,6 +115,10 @@ export interface RecentEntry {
 	login: string;
 	name: string | null;
 	avatarUrl: string | null;
+	/** Drives the chip's avatar shape (org = square, user = circle) and the verified badge. */
+	kind: "user" | "org";
+	/** Org-only verified badge; null for users. */
+	isVerified: boolean | null;
 }
 export interface StartPageData {
 	recent: RecentEntry[];
@@ -220,20 +224,29 @@ async function queryRecent(limit: number): Promise<RecentEntry[]> {
 			login: entities.login,
 			name: entities.name,
 			avatarUrl: entities.avatarUrl,
+			kind: entities.kind,
+			isVerified: entities.isVerified,
 			last: sql<string>`max(${lookups.searchedAt})`,
 		})
 		.from(lookups)
 		.innerJoin(entities, eq(entities.id, lookups.entityId))
-		// kind filter is defensive — org lookups aren't recorded (v1), but the strip links every
-		// entry to /$user, which for an org login would kick off a bogus user build.
-		.where(and(isNull(entities.suspendedAt), eq(entities.kind, "user")))
+		// Users and orgs both belong in the strip (each links to /$user, which resolves either);
+		// repos are the only other kind and don't get a page here, so they're excluded.
+		.where(
+			and(
+				isNull(entities.suspendedAt),
+				inArray(entities.kind, ["user", "org"]),
+			),
+		)
 		.groupBy(entities.id)
 		.orderBy(desc(sql`max(${lookups.searchedAt})`))
 		.limit(limit);
 	return rows.map((r) => ({
 		login: r.login,
 		name: r.name,
 		avatarUrl: r.avatarUrl,
+		kind: r.kind === "org" ? "org" : "user",
+		isVerified: r.isVerified,
 	}));
 }
 

src/lib/content.ts

@@ -1,9 +1,9 @@
 /**
  * The MDX content collections:
  * - src/content/metrics/<slug>.mdx → /metrics/<slug> (individual metrics, /metrics/explained hub)
- * - src/content/company/<slug>.mdx → /company/<slug> (company/org context — deliberately its own
- *   collection, NOT mixed into the individuals' hub: the definitions differ, e.g. org-scoped vs
- *   global contributions)
+ * - src/content/organizations/<slug>.mdx → /organizations/<slug> (organization context —
+ *   deliberately its own collection, NOT mixed into the individuals' hub: the definitions differ,
+ *   e.g. org-scoped vs global contributions)
  *
  * Two globs per collection with different costs:
  * - an eager, frontmatter-only glob (tiny — just the exported metadata objects), used
@@ -69,36 +69,34 @@ export function loadArticle(
 	return load?.();
 }
 
-// ── Company collection (/company/<slug>) ─────────────────────────────────────
+// ── Organization collection (/organizations/<slug>) ──────────────────────────
 
-const companyFrontmatters = import.meta.glob<ArticleFrontmatter>(
-	"../content/company/*.mdx",
+const orgFrontmatters = import.meta.glob<ArticleFrontmatter>(
+	"../content/organizations/*.mdx",
 	{ eager: true, import: "frontmatter" },
 );
 
-const companyComponents = import.meta.glob<{ default: MDXContent }>(
-	"../content/company/*.mdx",
+const orgComponents = import.meta.glob<{ default: MDXContent }>(
+	"../content/organizations/*.mdx",
 );
 
-/** All company articles in curated reading order. */
-export const companyArticles: ArticleMeta[] = Object.entries(
-	companyFrontmatters,
-)
+/** All organization articles in curated reading order. */
+export const orgArticles: ArticleMeta[] = Object.entries(orgFrontmatters)
 	.map(([path, fm]) => ({ slug: slugOf(path), ...fm }))
 	.sort(
 		(a, b) =>
 			(a.order ?? Number.MAX_SAFE_INTEGER) -
 				(b.order ?? Number.MAX_SAFE_INTEGER) || a.title.localeCompare(b.title),
 	);
 
-export function getCompanyArticleMeta(slug: string): ArticleMeta | undefined {
-	return companyArticles.find((a) => a.slug === slug);
+export function getOrgArticleMeta(slug: string): ArticleMeta | undefined {
+	return orgArticles.find((a) => a.slug === slug);
 }
 
-/** Lazily import a company article's compiled MDX module (shape fits React.lazy). */
-export function loadCompanyArticle(
+/** Lazily import an organization article's compiled MDX module (shape fits React.lazy). */
+export function loadOrgArticle(
 	slug: string,
 ): Promise<{ default: MDXContent }> | undefined {
-	const load = companyComponents[`../content/company/${slug}.mdx`];
+	const load = orgComponents[`../content/organizations/${slug}.mdx`];
 	return load?.();
 }

src/lib/org-cache.ts

@@ -1,4 +1,5 @@
 import { and, eq, isNull, sql } from "drizzle-orm";
+import { recordLookup } from "#/lib/cache";
 import { type DB, db } from "#/lib/db";
 import { entities, orgMembers } from "#/lib/db/schema";
 import {
@@ -12,12 +13,13 @@ import {
 } from "#/lib/github";
 
 /**
- * Incremental org ("company") cache — the org sibling of cache.ts.
+ * Incremental organization cache — the org sibling of cache.ts.
  *
  * An org's numbers are the sum of its members' contributions *to that org* (org-scoped
  * `contributionsCollection(organizationID: …)`, a different number from each member's global
  * totals). Those per-member sums live in `org_members`; the roll-up lands on the org's
- * `entities` row so the company leaderboard ranks orgs exactly like the user board ranks users.
+ * `entities` row so the organization leaderboard ranks orgs exactly like the user board ranks
+ * users.
  *
  * Builds are **resumable** at member granularity: enumeration inserts every member as a pending
  * `org_members` row (`lastFetched` null), each fetched member is stamped the moment its totals
@@ -43,9 +45,10 @@ const BUILD_BUDGET_MS = 6_000;
 const MEMBER_CONCURRENCY = 3;
 
 // Orgs with more visible members than this get a clean refusal instead of an on-demand build:
-// at ~2 requests per member a mega-org lookup would eat the shared token's hourly budget and
-// poison it for every visitor. Lifting this needs the background worker.
-const MAX_ORG_MEMBERS = 400;
+// at ~2 requests per member even a mid-size org lookup can eat into the shared token's hourly
+// budget and poison it for every visitor. We keep this deliberately low and record the refused
+// orgs (see getFromDb) so the background worker can backfill them later, off the request path.
+const MAX_ORG_MEMBERS = 25;
 
 export interface OrgSummary {
 	login: string;
@@ -141,11 +144,19 @@ async function getFromDb(
 
 	if (!row) {
 		// First sighting: the profile fetch validates the login and yields the node id (keys every
-		// org-scoped contribution query) + createdAt (caps member windows). Refuse oversized orgs
-		// BEFORE writing anything — a half-enumerated mega-org would burn quota on every poll.
+		// org-scoped contribution query) + createdAt (caps member windows). Record the profile
+		// (a single cheap upsert into `entities`, builtAt still null) BEFORE refusing an oversized
+		// org — this leaves a tracked row the background worker can later backfill, instead of
+		// discarding the lookup. The request path still won't build it live.
 		const profile = await fetchOrgProfile(login, token);
-		assertBuildable(profile);
 		row = await upsertOrgProfile(database, id, profile, now);
+		// Record the lookup before the size check so even an oversized org we refuse to build
+		// still surfaces in "recently looked up" — the row exists and the strip links to /$user.
+		await recordLookup(database, id, now);
+		assertBuildable(profile);
+	} else {
+		// Known org, revisited: bump its recency so it re-sorts to the front of the strip.
+		await recordLookup(database, id, now);
 	}
 
 	const complete = row.builtAt != null;
@@ -183,6 +194,13 @@ async function getFromDb(
 		throw new GitHubError(`Could not resolve "${login}" on GitHub.`, 502);
 	}
 
+	// Enrolled but oversized: the row is recorded for the worker, but the request path must refuse
+	// it — cheaply, from the stored memberCount, so a repeat lookup never pays to re-enumerate its
+	// members before bailing. (assertBuildable already caught it at first sighting.)
+	if ((row.memberCount ?? 0) > MAX_ORG_MEMBERS) {
+		throw new GitHubError(tooLargeMessage(login, row.memberCount ?? 0), 422);
+	}
+
 	// Enumerate members once per build: every member becomes a pending org_members row, which
 	// doubles as the resume marker. Re-enumeration (membership drift) is the worker's job.
 	if ((await memberRowCount(database, id)) === 0) {
@@ -278,7 +296,7 @@ function assertBuildable(profile: OrgProfile) {
 }
 
 function tooLargeMessage(login: string, count: number) {
-	return `"${login}" has ${count.toLocaleString()} public members — too many for an on-demand build yet.`;
+	return `${login} has ${count.toLocaleString()} public members — a bit much for us to crunch on the spot. We’ve added it to the queue and are indexing it in the background; check back a little later. Thanks for your interest!`;
 }
 
 async function memberRowCount(database: DB, id: string): Promise<number> {