feat(growth): restore daily reporting and campaign attribution (#7832)

* feat(growth): restore core reporting and attribution

* test(growth): bound cyclops receiver contract

* fix(posthog): combine growth rollout scopes

---------

Co-authored-by: f-trycua <195596869+f-trycua@users.noreply.github.com>
CloudCyclopsCs-RevId: e8886a4b693af319d7c6f933f68df5ba48522270

Author: f-trycua

libs/fleet/backend/handlers/analytics_attribution.go

@@ -19,6 +19,7 @@ const (
 var fleetAttributionProperties = map[string]string{
 	"campaign_id":  productanalytics.FirstTouchCampaignIDProperty,
 	"content_id":   productanalytics.FirstTouchContentIDProperty,
+	"utm_content":  productanalytics.FirstTouchContentIDProperty,
 	"utm_source":   productanalytics.FirstTouchUTMSourceProperty,
 	"utm_medium":   productanalytics.FirstTouchUTMMediumProperty,
 	"utm_campaign": productanalytics.FirstTouchUTMCampaignProperty,
@@ -102,6 +103,9 @@ func validateFleetAttribution(record fleetAttributionRecord, now time.Time) (map
 		if !ok || !validFleetAttributionValue(value) {
 			return nil, false
 		}
+		if existing, exists := setOnce[property]; exists && existing != value {
+			return nil, false
+		}
 		setOnce[property] = value
 	}
 	return setOnce, true

libs/fleet/backend/handlers/analytics_attribution_test.go

@@ -33,7 +33,8 @@ func TestRecordFleetAttributionBindsExternalFirstTouch(t *testing.T) {
 		Values: map[string]string{
 			"utm_source":   "x",
 			"utm_medium":   "organic-social",
-			"utm_campaign": "openclaw-2-launch",
+			"utm_campaign": "cursor-cloud-fleets",
+			"utm_content":  "thread-post-5",
 		},
 	}, &auth.User{ID: "subject-1", Email: "person@example.test", EmailVerified: true, AZP: "cyclops-cs-spa", PrincipalType: auth.PrincipalTypeUser})
 	response := httptest.NewRecorder()
@@ -49,11 +50,32 @@ func TestRecordFleetAttributionBindsExternalFirstTouch(t *testing.T) {
 	}
 	if event.SetOnce[productanalytics.FirstTouchUTMSourceProperty] != "x" ||
 		event.SetOnce[productanalytics.FirstTouchUTMMediumProperty] != "organic-social" ||
-		event.SetOnce[productanalytics.FirstTouchUTMCampaignProperty] != "openclaw-2-launch" {
+		event.SetOnce[productanalytics.FirstTouchUTMCampaignProperty] != "cursor-cloud-fleets" ||
+		event.SetOnce[productanalytics.FirstTouchContentIDProperty] != "thread-post-5" {
 		t.Fatalf("set once = %#v", event.SetOnce)
 	}
 }
 
+func TestValidateFleetAttributionContentAliases(t *testing.T) {
+	now := time.Now()
+	for _, values := range []map[string]string{
+		{"content_id": "thread-post-5"},
+		{"utm_content": "thread-post-5"},
+		{"content_id": "thread-post-5", "utm_content": "thread-post-5"},
+	} {
+		setOnce, valid := validateFleetAttribution(fleetAttributionRecord{Version: 1, CapturedAt: now.UnixMilli(), Values: values}, now)
+		if !valid || setOnce[productanalytics.FirstTouchContentIDProperty] != "thread-post-5" {
+			t.Fatalf("values/setOnce/valid = %#v/%#v/%v", values, setOnce, valid)
+		}
+	}
+
+	if _, valid := validateFleetAttribution(fleetAttributionRecord{
+		Version: 1, CapturedAt: now.UnixMilli(), Values: map[string]string{"content_id": "legacy", "utm_content": "standard"},
+	}, now); valid {
+		t.Fatal("conflicting content aliases must be rejected")
+	}
+}
+
 func TestRecordFleetAttributionBindsMissingEmailIdentity(t *testing.T) {
 	capture := &analyticsCapture{}
 	h := Handlers{Analytics: capture, AuthCfg: configAuthForAnalytics()}

libs/fleet/docs/fleet-attribution-phase1.md

@@ -4,11 +4,11 @@ This public contract defines bounded first-touch campaign binding and workload q
 
 ## Browser capture
 
-Before Keycloak bootstrap, the SPA captures first-touch values from the current URL: `campaign_id`, `content_id`, `utm_source`, `utm_medium`, and `utm_campaign`. Values use a conservative ASCII set (`A-Z`, `a-z`, `0-9`, `.`, `_`, `~`, `-`), are non-empty, and are at most 128 characters. The serialized version-1 record is limited to 2048 UTF-8 bytes and expires after 7 days. Unknown query keys are ignored; repeated allowed keys, malformed values, and oversized records are rejected. The first valid record is write-once in browser `sessionStorage` and is removed after a successful bind.
+Before Keycloak bootstrap, the SPA captures first-touch values from the current URL: `campaign_id`, `content_id`, `utm_source`, `utm_medium`, `utm_campaign`, and the standard `utm_content` alias. `utm_content` is stored and bound under the existing canonical `content_id` field, preserving compatibility with older `content_id` links and the `fleet_first_touch_content_id` person property. If both aliases are present they must have the same value. Values use a conservative ASCII set (`A-Z`, `a-z`, `0-9`, `.`, `_`, `~`, `-`), are non-empty, and are at most 128 characters. The serialized version-1 record is limited to 2048 UTF-8 bytes and expires after 7 days. Unknown query keys are ignored; repeated allowed keys, conflicting aliases, malformed values, and oversized records are rejected. The first valid record is write-once in browser `sessionStorage` and is removed after a successful bind.
 
-The five stable fields are acquisition-channel neutral. They can represent email, SEO, partners, future channels, and organic or paid social traffic without provider-specific behavior. For example, both `utm_source=x&utm_medium=organic-social` and `utm_source=x&utm_medium=paid-social` are ordinary valid values. Provider SDKs, pixels, credentials, arbitrary provider payloads, and provider-specific success semantics are outside this contract. Pixels are optional downstream adapters and are not part of Phase 1.
+The five canonical fields are acquisition-channel neutral. They can represent email, SEO, partners, future channels, and organic or paid social traffic without provider-specific behavior. For example, both `utm_source=x&utm_medium=organic-social` and `utm_source=x&utm_medium=paid-social` are ordinary valid values. Provider SDKs, pixels, credentials, arbitrary provider payloads, and provider-specific success semantics are outside this contract. Pixels are optional downstream adapters and are not part of Phase 1.
 
-Use `content_id` for placement detail, not `utm_content` (an ignored unknown key). For the GitHub README CTA, use this capture-compatible URL:
+New links may use standard `utm_content`; existing `content_id` links remain supported. For the GitHub README CTA, this existing capture-compatible URL remains valid:
 
 ```text
 https://run.cua.ai/?utm_source=github&utm_medium=referral&utm_campaign=fleet_activation&content_id=repo_readme

libs/fleet/src/auth/fleet-attribution.ts

@@ -4,6 +4,7 @@ export const ATTRIBUTION_MAX_VALUE_LENGTH = 128
 export const ATTRIBUTION_MAX_RECORD_BYTES = 2048
 
 const allowedKeys = ["campaign_id", "content_id", "utm_source", "utm_medium", "utm_campaign"] as const
+const queryKeys = [...allowedKeys, "utm_content"] as const
 type AttributionKey = (typeof allowedKeys)[number]
 export type AttributionRecord = { version: 1; capturedAt: number; values: Partial<Record<AttributionKey, string>> }
 export interface StorageLike { getItem(key: string): string | null; setItem(key: string, value: string): void; removeItem?(key: string): void }
@@ -20,10 +21,14 @@ export function captureFleetAttribution(href: string, options: Options = {}): vo
   let params: URLSearchParams
   try { params = new URL(href).searchParams } catch { return }
   const values: Partial<Record<AttributionKey, string>> = {}
-  for (const key of allowedKeys) {
+  for (const key of queryKeys) {
     const all = params.getAll(key)
     if (all.length > 1 || (all.length === 1 && !validValue(all[0]!))) return
-    if (all.length === 1) values[key] = all[0]!
+    if (all.length === 1) {
+      const canonicalKey: AttributionKey = key === "utm_content" ? "content_id" : key
+      if (values[canonicalKey] !== undefined && values[canonicalKey] !== all[0]) return
+      values[canonicalKey] = all[0]!
+    }
   }
   if (Object.keys(values).length === 0) return
   const record: AttributionRecord = { version: 1, capturedAt: (options.now ?? Date.now)(), values }

libs/fleet/tests/fleet-attribution-doc.test.mjs

@@ -18,6 +18,9 @@ test("public contract is channel-neutral and says unknown query keys are ignored
   assert.match(contract, /utm_source=x.*utm_medium=organic-social/s)
   assert.match(contract, /utm_source=x.*utm_medium=paid-social/s)
   assert.match(contract, /unknown query keys are ignored/i)
+  assert.match(contract, /standard `utm_content` alias/i)
+  assert.match(contract, /stored and bound under the existing canonical `content_id` field/i)
+  assert.match(contract, /both aliases are present they must have the same value/i)
   assert.match(contract, /pixels.*optional downstream adapters/i)
   assert.match(contract, /keyed-HMAC pseudonym/i)
   assert.match(contract, /internal and unknown identities are discarded/i)

libs/fleet/tests/fleet-attribution.test.ts

@@ -76,10 +76,7 @@ test("GitHub README CTA preserves all four fields through a simulated Keycloak b
     },
   }
 
-  for (const href of [
-    cta,
-    `${cta}&utm_content=ignored&email=person%40example.test&identity=raw-subject&url=https%3A%2F%2Fexample.test`,
-  ]) {
+  for (const href of [cta, `${cta}&email=person%40example.test&identity=raw-subject&url=https%3A%2F%2Fexample.test`]) {
     const storage = memoryStorage()
     captureFleetAttribution(href, { storage, now: () => firstTouch })
     assert.deepEqual(JSON.parse(storage.getItem(ATTRIBUTION_STORAGE_KEY) ?? "null"), expected)
@@ -102,12 +99,25 @@ test("GitHub README CTA preserves all four fields through a simulated Keycloak b
       getAccessToken: async () => "access-token",
     })
 
-    // Exact records exclude the raw URL, email, identity, and unknown utm_content.
+    // Exact records exclude the raw URL, email, identity, and unknown keys.
     assert.deepEqual(bound, [expected])
     assert.equal(storage.getItem(ATTRIBUTION_STORAGE_KEY), null)
   }
 })
 
+test("standard utm_content is stored under canonical content_id without loss", () => {
+  for (const href of [
+    "https://run.cua.ai/?utm_source=x&utm_medium=organic-social&utm_campaign=cursor-cloud-fleets&utm_content=thread-post-5",
+    "https://run.cua.ai/?utm_content=thread-post-5&content_id=thread-post-5",
+  ]) {
+    const storage = memoryStorage()
+    captureFleetAttribution(href, { storage, now: () => 50 })
+    const record = JSON.parse(storage.getItem(ATTRIBUTION_STORAGE_KEY) ?? "null")
+    assert.equal(record.values.content_id, "thread-post-5")
+    assert.equal(record.values.utm_content, undefined)
+  }
+})
+
 test("invalid, malformed, empty, oversized, and expired records never reach a binder", async () => {
   const now = 1_700_000_000_000
   const invalidStoredRecords = [
@@ -133,6 +143,8 @@ test("capture rejects repeated, empty, invalid, and oversized values without tru
   const invalidQueries = [
     "?campaign_id=",
     "?campaign_id=one&campaign_id=two",
+    "?utm_content=one&utm_content=two",
+    "?content_id=legacy&utm_content=standard",
     "?campaign_id=has%20space",
     `?campaign_id=${"x".repeat(ATTRIBUTION_MAX_VALUE_LENGTH + 1)}`,
   ]