| name | cmo-cross-channel-dashboard |
| description | Generate a CMO Cross-Channel Performance Dashboard — funnel-stage executive view of marketing spend, pull-through, and per-channel efficiency across Meta, Google, LinkedIn, TikTok, Reddit, TTD, and the Discovery API long tail. Single-screen, three-tab layout (Overview / Channels / Customer Journey) with reactive filter bar (period × channel × conversion event × attribution model). Warehouse-first (ClickHouse `_all_data` views) with Discovery API fallback for channels not in the warehouse; all data from real sources. Built via /business-intelligence-editor as ONE custom-component widget that contains the entire dashboard. |
CMO Cross-Channel Dashboard
Purpose
Produce an executive-grade cross-channel performance dashboard for CMOs / VPs of Marketing / Heads of Growth running multi-platform paid media. The deliverable answers three questions in a single glance:
- Where is my money going? — daily Spend / Revenue / Conversions trend with the best-ROAS day called out, plus a per-channel mix donut.
- What is my funnel doing? — Awareness / Consideration / Purchase pull-through derived from campaign-objective auto-classification (§ Funnel-stage taxonomy below), with per-stage KPIs.
- Which channel is best at which job? — sortable 14-column channel matrix (Spend, Δ-vs-prev, Impressions, Clicks, CTR, CPC, CVR, Conversions, CPA, Avg-value, Revenue, ROAS, sparkline) + a Spend × ROAS × Conversions efficiency scatter (Scale / Keep / Test / Fix quadrants) + first-touch vs last-touch role bars + top customer-journey paths.
The dashboard is executive (read 1–3× per week, share to leadership / board), not operational (read 5–20× per day) — for the operational console, fall back to daily-performance-report. For tactical drill-down briefs with 3 prioritized actions, fall back to UC-DPA-1's tactical drill skill.
Derived from Marketing OS use-case UC-CMO-1 (CMO Cross-Channel Dashboard) v1.1. Canonical visual contract lives in the Improvado Marketing OS vault and is mirrored as the HTML in Appendix A below — that HTML is the single source of truth for the widget structure.
Invocation Context
Two trigger paths feed this skill:
Path A — Onboarding dispatch (new users)
Company personalization: if the dispatch message carries an alg-prebrief fenced block with a company_research: line, tailor exactly ONE headline/intro sentence in the closing chat hand-off to this company — always hedged ("looks like…"), never as a fact. If absent, proceed generically.
If the conversation contains an onboarding_summary message with Interview answers:, scan those answers for bias signals:
- Role
cmo-director / c-level / vp-marketing → executive voice; topbar last_sync_label reads "Live · <N> min ago"; default tab is Overview; the Channels and Customer Journey tabs are hidden by default behind a "Show details" disclosure (set editState.persona = 'executive')head-of-growth / growth → exec voice but Customer Journey tab visible by default (since growth lives on first-touch / multi-touch attribution)marketer / performance-marketer → all 3 tabs visible from first paint; default tab is Channels (sorted by spend); KPI strip emphasizes ROAS, CPA, CTRanalyst-bi → all 3 tabs + "Open in BI editor for live data" CTA in the topbar; channel matrix pre-sorted by spend; raw numbers favored over visualisation chrome
- Reconciled metrics — if
ROAS listed primary → keep ROAS as the leftmost KPI tile; if CPA listed primary → swap KPI tile order to put CPA leftmost; if CPL (B2B) listed primary → relabel "Conversions" tile as "Leads" and pre-select the Leads conversion-event filter
- AI wish — if mentions
funnel, awareness, consideration, purchase → ensure the period defaults to QTD (90d) instead of 30d (richer funnel signal); if mentions where is my budget / mix → land on the Channels tab regardless of role bias
- Industry — eCom / DTC → leftmost KPI = ROAS (already default). B2B SaaS → leftmost KPI = CPL. Healthcare → leftmost KPI = CPL + appointment-conversion. Retail multi-store → leftmost KPI = store-traffic lift (if available, otherwise revenue). See
agency.kpi_pin_industry mapping in Step 3.
Path B — Free-form chat (returning users, repeat use)
User asks naturally — "CMO dashboard", "show me cross-channel performance", "where is my budget going across channels", "marketing overview". No onboarding_summary; use defaults:
- Period: last 30d vs prior 30d
- Attribution: data-driven
- Channel filter: All (every connected ad platform)
- Conversion event: All events
- Default tab: Overview
When to Trigger
User asks for any of:
- "CMO dashboard", "executive marketing dashboard", "VP marketing dashboard"
- "cross-channel dashboard", "all my channels", "marketing overview across channels"
- "where is my budget going", "funnel view across channels"
- "awareness vs consideration vs purchase", "across the funnel"
- "spend × ROAS by channel", "channel efficiency map"
- "first touch vs last touch", "customer journey across channels"
If the user instead asks for a daily operational console ("today's report", "what's running today", "pacing dashboard") — fall back to daily-performance-report.
If the user wants a tactical leadership-ready brief with 3 issues + drill actions → invoke the UC-DPA-1 daily performance analysis skill.
If the user wants a single-channel chart ("show Meta spend chart") → use /business-intelligence-editor directly with a single-source widget; do NOT invoke this skill.
If the user wants a rule-quality scorecard with per-rule $-at-risk → invoke full-marketing-audit.
Output Mode
DEFAULT → Mode B: Dashboard via /business-intelligence-editor (ONE custom-component widget)
IF user explicitly says "document" / "markdown" / "deck" → Mode A: createDocument fallback
IF user asks for one chart → Mode C: visualizationTool (single chart in chat)
Mode B: Dashboard (DEFAULT)
Use /business-intelligence-editor skill with ONE custom-component widget. The entire dashboard (topbar, tabs, KPI strip, daily trend, alerts feed, donut, scatter, channel matrix, journey paths, role bars) lives inside that single widget — its componentCode is the canonical HTML from Appendix A with Discovery API data substituted at the marked INJECT points.
If /business-intelligence-editor is unavailable at runtime, degrade to Mode A with a one-line note ("Saved as a document — dashboard skill not available right now.").
Mode A: Document
Use createDocument. Markdown report — KPI strip as a table, channel matrix as a table, journey paths as bullet lists, alerts as a triaged list. Not as visual but readable.
Mode C: Quick Visualization
Use visualizationTool for single chart in chat. Use this when the user clearly wants a one-off — never as the default for "CMO dashboard".
Auto-Refresh Strategy
Dashboard is a point-in-time snapshot — the period window is fixed inside it (filter chrome from BIE is hidden; the canonical owns its own period picker). Each scheduled run pulls a fresh window. After creation, offer:
"Want me to refresh this every Monday at 8 AM with last-30d numbers?"
If yes → scheduleChatTool with cadence weekly-monday-08:00-user-tz. Each run re-executes the skill: re-fetches Discovery API, re-substitutes Appendix A, overwrites the saved dashboard at the same dashboardUrl.
MCP Tools Used
getCurrentWorkspaceContextTool — workspace name + idgetConnectionsTool — enumerate active ad-platform connectionsclickhouseTool — discover _all_data views via mrt_database_tables and read warehouse channel data (Step 1, primary source per Rule 0)discoveryListAccountsTool — accounts per connection (for selection rule)discoveryRequestTool — per-channel insights fetch for Discovery-fallback channels (Step 1, parallel)getCrmObjectsTool (HubSpot/SFDC, optional) — for journey paths when GA4 path-data is unavailableSkill('business-intelligence-editor') — save the assembled custom-component widget
Rule 0 — Warehouse-first, Discovery API fallback (HARD)
Every data point comes from a real source — never fabricated, never fixture/sample data, never hand-typed into the canonical HTML. Prefer the ClickHouse warehouse when it has the data; use the Discovery API as the fallback. Resolve the source per channel in this order:
- Warehouse
_all_data view — if mrt_database_tables lists a fresh {report_type}_{agency_id}_{datasource}_all_data view for the channel (freshness within the requested period), read spend / impressions / clicks / conversions / revenue + daily series from it. Faster and more reliable than the live API.
- Discovery API — only if no view exists, the view is stale, or a needed field is missing, fall back to the documented per-platform Discovery call (STEP 1).
Rules:
- ❌ Never fabricate or hand-type values; never ship sample/fixture data in the widget.
- ❌ Never use a schema prefix (
im_… or internal_analytics.) and never SHOW TABLES — discover names via mrt_database_tables (no prefix). Ignore UNKNOWN_TABLE "Maybe you meant im_…" suggestions (other tenants). See general/querying.md.
- ⚠️ Mark warehouse data older than the requested period, or live data older than 1h, with a "stale" indicator on
last_sync_label.
Plan/goal lookup (biz_marketing_targets, see STEP 2E) is workspace configuration, not marketing data. If missing, infer from prior-period actuals × growth target (see Step 2F) or set goals to zero (canonical renders flat goal bars).
Output Contract (HARD)
Visible to the user:
- ONE BIE custom-component widget covering the full dashboard (12 cols × ~28 rows in the BIE grid)
- Three tabs (Overview / Channels / Customer Journey) — see § Persona-tuned default depth above for which are visible by default
- Five KPI tiles (Spend / Revenue / ROAS / Conversions / CPA) with Δ-vs-prev, plan-progress bars
- Daily trend (Spend + Revenue dual-axis) with best-ROAS day callout
- Alerts feed ("Needs your attention") with severity-colored cards
- Sortable channel matrix (Channels tab)
- Donut (channel mix) + Scatter (Spend × ROAS × Conversions, Scale/Keep/Test/Fix quadrants) (Channels tab)
- Customer-journey paths (top by share) + journey stat grid + first/last touch role bars (Journey tab)
Not in scope:
- Slack-ready text artifact (use UC-DPA-1 for that)
- Per-creative deep-dives (use
weekly-creative-performance for that)
- Deterministic audit scorecard (use
full-marketing-audit for that)
Visual contract:
- Canonical Inter typography + indigo
#4b5cf2 accent + the existing platform brand palette per channel (Meta blue, Google blue, LinkedIn dark blue, TikTok black, …)
- Persona-tuned tab visibility per § Invocation Context
- No client logo in the topbar by default — the agency name in
#brand-name IS the brand stamp; the brand-mark monogram (#brand-mark) renders the first letter only
Quick Reference
Canonical INJECT points (Step 3 has the full mapping):
- Inline
{{INJECT:foo}} placeholders for header / topbar text (10 markers)
- Block markers
// === INJECT: <name> === ... // === END INJECT: <name> === for JS data arrays/objects (8 blocks: channels, trend_daily, ch_data, goals, alerts, journeys, journey_stats, roles)
Tab visibility flag — write editState.persona = 'executive' in the saved config to hide Channels and Journey tabs (Path A § cmo-director branch).
dashboardUrl — "clients/template/dashboards/CrossChannelEditableDashboard.tsx" (canonical host-TSX path; emit directly per the precedent set by full-marketing-audit/dashboard-template.json:104 and weekly-creative-performance/dashboard-template.json:112 v7.4.1+. Auto-refresh re-saves overwrite at the same dashboardUrl + settings_id composite key.).
STEP −1: Pre-flight checks (MANDATORY — defends Bugs #1, #2, #5–#8)
Checks before any data fetch — each defends a class of runtime bugs documented in UC-CMO-1 — Bugs and Fixes 2026-05-08.md:
mrt_database_tables query (no prefix) — list the _all_data views + freshness for every connected channel (and confirm the plan/goal table for STEP 2E). This is what decides, per channel, warehouse vs Discovery (Rule 0). Never use SHOW TABLES (returns empty via the proxy); never trust cross-tenant UNKNOWN_TABLE suggestions; never use a schema prefix. Memorized table names are stale — always discover.discoveryWebSearchTool — only for channels with no fresh warehouse view — resolve the connector's URL allowlist before composing any discoveryRequestTool body. Pasting URLs from public API docs returns HTTP 400 (Bug #1). Skip this for channels served from the warehouse.validate-widget CLI dry-run — npx tsx .claude/skills/business-intelligence-editor/skill-cli.ts validate-widget --config '<stub_with_minimal_props>' on a stub config to confirm the BIE schema version is compatible with this skill's STEP 5 example (Bugs #5–#8).
If any pre-flight check fails, surface the failure to the user and STOP — do not attempt to "guess past" a missing table or a schema mismatch.
STEP 0: Workspace + Connection Detection
0A. Resolve workspace context
ws = getCurrentWorkspaceContextTool(impersonation_context_id)
→ workspace_name, workspace_id
0B. Enumerate active ad-platform connections
all_conns = getConnectionsTool(impersonation_context_id)
active = [c for c in all_conns if c.is_active and c.data_source in MARKETING_PLATFORMS]
by_ds = group_by(active, key=c.data_source)
0C. Connection count gate
IF len(by_ds) < 2:
Surface: "Cross-channel dashboard requires ≥2 active ad-platform connections.
You have {N}. Connect another platform, or use a single-channel chart instead."
STOP — fall back to Mode C single-channel chart (or daily-performance-report if user says 'daily').
0D. Account selection per channel (§ 0.1 rule)
FOR each channel in by_ds:
accounts = discoveryListAccountsTool(connectionId=channel.id)
IF len(accounts) > 5:
Ask user which account(s) to include — never pick blindly.
IF len(accounts) ≥ 2 and ≤5:
Use highest-spend account by default; surface choice in subtitle.
IF len(accounts) == 1:
Auto-select.
0E. Build agency identity payload
agency_name = workspace_name
agency_meta = "agency_id <id> · <N> sub-brands" # N = sum(accounts) across channels
last_sync_label = "Live" # updated post-fetch in Step 1 to "Live · <N> min ago"
avatar_initials = first letters of impersonator's first+last name
owner_name = biz_active_customers.account_owner_name (if external client; else "")
owner_role = biz_active_customers.account_owner_role OR "Improvado AE"
period_label = "<from_label> – <to_label>, <year>" # e.g. "Apr 4 – May 4, 2026"
period_compare = "<from_label> – <to_label> vs <prev_from> – <prev_to>, <year>"
trend_desc = "<N> days (<from_iso> → <to_iso>) · <attribution> attribution"
foot_source = "Source: Improvado Discovery API · live"
mark_letter = uppercased first letter of agency_name
STEP 1: Fetch Per-Channel Data (warehouse-first, Discovery fallback)
STEP 1.−1: Resolve the data source per channel (Rule 0 — MANDATORY)
Using the mrt_database_tables listing from STEP −1, classify each connected channel:
- Warehouse channel — a
{report_type}_{agency_id}_{datasource}_all_data view exists and its
freshness is within the requested period. Read the period aggregates, the same-period-prior
aggregates, the daily series, and the per-event split directly from that view with clickhouseTool
(no schema prefix, LIMIT 250-style guards, alias tables). Skip the Discovery calls for this channel.
Map the view's columns into the same ch_data / channels / trend_daily structures the Discovery
path produces (STEP 2). Resolve real column names from mrt_database_columns — do not assume.
- Discovery channel — no view, stale view, or a needed field (e.g. video quartiles, attribution
split) is missing from the warehouse. Use the per-platform Discovery call below.
Only channels classified as Discovery proceed to STEP 1.0.
STEP 1.0: Resolve allowed endpoints (Discovery channels only — Bug #1 fix)
Before any discoveryRequestTool call, resolve the per-connector URL allowlist. The Discovery proxy enforces a connector-specific URL whitelist; pasting a vendor URL from public docs (or memory) hits paths that aren't proxied and returns HTTP 400 regardless of token validity.
For each connected channel, call discoveryWebSearchTool (or the connector-specific endpoint listing if exposed) and pick the URL from the returned set. Do NOT construct URLs from public API docs. The per-platform request shapes documented below describe the body — the URL must come from the allowlist response.
Run in parallel for every active channel. Each channel returns:
- Period aggregates (spend, impressions, reach/freq if available, clicks, conversions, conversion-event split, conversion-value)
- Same period prior (for Δ-vs-prev computation)
- Daily series (for the trend chart) at
time_increment=1 (or the platform's daily-granularity equivalent)
- Per-event split (lead / purchase / add-to-cart / page-view) — needed for KPI engine
ch_data (Step 2A)
Meta — /v23.0/act_<account_id>/insights via discoveryRequestTool
fields: spend, impressions, reach, frequency, clicks, ctr, cpm, cpc,
actions, action_values,
video_p25_watched_actions, video_p50_watched_actions,
video_p75_watched_actions, video_p100_watched_actions,
video_thruplay_watched_actions,
conversions, purchase_roas
level: campaign
breakdowns: []
time_range: { since: <date_start>, until: <date_end> }
time_increment: 1 # daily for trend chart
Event split derivation:
ch_data[meta].leads = sum of actions[type=lead] + actions[type=onsite_conversion.lead_grouped]ch_data[meta].purch = sum of actions[type=purchase] + actions[type=offsite_conversion.fb_pixel_purchase]ch_data[meta].ac = sum of actions[type=add_to_cart]ch_data[meta].pv = sum of actions[type=landing_page_view] OR actions[type=page_view]channels[meta].revenue = sum of action_values[action_type=purchase]
Google Ads — googleAds:searchStream (GAQL)
SELECT
segments.date,
campaign.advertising_channel_type,
metrics.cost_micros,
metrics.impressions,
metrics.clicks,
metrics.ctr,
metrics.average_cpc,
metrics.conversions,
metrics.conversions_value,
metrics.video_views,
metrics.video_view_rate,
metrics.video_quartile_p25_rate,
metrics.video_quartile_p50_rate,
metrics.video_quartile_p75_rate,
metrics.video_quartile_p100_rate,
metrics.all_conversions,
metrics.cost_per_conversion,
metrics.value_per_conversion
FROM customer
WHERE segments.date BETWEEN '<date_start>' AND '<date_end>'
For event split, run a second GAQL grouped by conversion_action.category:
SELECT
conversion_action.category,
metrics.conversions
FROM conversion_action
WHERE segments.date BETWEEN '<date_start>' AND '<date_end>'
Then map: LEAD → leads, PURCHASE → purch, ADD_TO_CART → ac, PAGE_VIEW → pv.
LinkedIn — /rest/adAnalytics?q=analytics
pivot: CAMPAIGN
timeGranularity: DAILY
fields: spend, impressions, clicks, costInUsd,
oneClickLeads,
videoViews, videoCompletions,
videoFirstQuartileCompletions, videoMidpointCompletions, videoThirdQuartileCompletions,
externalWebsiteConversions, externalWebsitePostClickConversions
dateRange: { start: ..., end: ... }
Event split: oneClickLeads → leads, externalWebsiteConversions → purch (when goal=WEBSITE_CONVERSION).
TikTok — /v1.3/report/integrated/get/
data_level: AUCTION_CAMPAIGN
report_type: BASIC
dimensions: [campaign_id, stat_time_day]
metrics: [spend, impressions, reach, clicks, ctr,
video_play_actions, video_watched_2s, video_watched_6s,
video_views_p25, video_views_p50, video_views_p75, video_views_p100,
conversion, cost_per_conversion, complete_payment_roas]
start_date: <date_start>
end_date: <date_end>
Event split via optimization_event parameter on a follow-up call: lead → leads, complete_payment → purch, add_to_cart → ac, view_content → pv.
channels[tiktok].revenue = complete_payment_roas × spend (TikTok exposes ROAS, not raw value).
Reddit / TTD / others (long tail)
Per the matching data-sources/<DS>.md § 2 live-API endpoint. If a platform has no Discovery connector, mark connected: false in the channels payload and the canonical surfaces it in the source-footer "Not connected" cluster.
Per-channel timeout
8s per channel. If a channel times out → fail that channel, continue others. Surface in topbar via last_sync_label: "Live · 8 min ago — Reddit delayed (retry 12:13)". Do NOT mock; render — for missing channel cells per Quality Gates below.
STEP 2: Compute Derived Fields
After parallel fetch:
2A. Per-channel KPI snapshot (ch_data)
For each channel.key in {meta, google, linkedin, tiktok, ...}:
ch_data[key] = {
spend: SUM channel.period.spend,
rev: SUM channel.period.revenue,
conv: SUM channel.period.conversions,
leads: SUM channel.period.events.lead,
purch: SUM channel.period.events.purchase,
ac: SUM channel.period.events.add_to_cart,
pv: SUM channel.period.events.page_view,
imp: SUM channel.period.impressions,
prevSpend: SUM channel.prev.spend,
prevRev: SUM channel.prev.revenue,
prevConv: SUM channel.prev.conversions,
prevLeads: SUM channel.prev.events.lead,
prevPurch: SUM channel.prev.events.purchase
}
2B. Channels list (top-level array consumed by the channel matrix, donut, scatter)
For each connected channel:
channels[i] = {
key: channel.key, # 'meta', 'google', ...
name: channel.display_name, # 'Meta Ads', 'Google Ads', ...
sub: <account_name OR "Multi-account (N)" if N>=2>,
logo: channel.key, # matches <symbol id="lg-meta"> etc.
color: PLATFORM_BRAND_COLOR[channel.key],
connected: true,
spend: ch_data[key].spend,
impr: ch_data[key].imp,
clicks: SUM channel.period.clicks,
conv: ch_data[key].conv,
revenue: ch_data[key].rev,
prevSpend: ch_data[key].prevSpend,
prevROAS: ch_data[key].prevRev / ch_data[key].prevSpend if prevSpend > 0 else 0
}
For each disconnected platform from the active-source list (LinkedIn / YouTube / etc.):
channels[i] = { key, name, sub: "Not connected", logo, color, connected: false, spend: 0, ... }
2C. Daily trend (trend_daily)
Build one row per day in [date_start, date_end]:
trend_daily[d] = {
day: d_index_1based,
dt: "MM-DD" of that day,
dow: day_of_week_0_to_6,
spend: SUM channels.daily[d].spend across selected channels,
rev: SUM channels.daily[d].revenue across selected channels,
conv: SUM channels.daily[d].conversions across selected channels
}
2D. Funnel-stage classification (per-campaign auto-classification)
For each campaign, classify into Awareness / Consideration / Purchase by objective + KPI structure:
| Stage | Meta objective | Google campaign_type | LinkedIn objectiveType | TikTok objective_type |
|---------------|-----------------------------------------|-----------------------------|------------------------|----------------------------|
| Awareness | OUTCOME_AWARENESS, BRAND_AWARENESS, | VIDEO, DEMAND_GEN, | BRAND_AWARENESS, | REACH, VIDEO_VIEWS |
| | REACH, VIDEO_VIEWS | DISPLAY w/ reach goal | VIDEO_VIEW | |
| Consideration | OUTCOME_TRAFFIC, OUTCOME_ENGAGEMENT, | SEARCH (non-brand), | WEBSITE_VISIT, | TRAFFIC, ENGAGEMENT, |
| | LINK_CLICKS, POST_ENGAGEMENT, MESSAGES | DISPLAY w/ click goal, | ENGAGEMENT, | LEAD_GENERATION |
| | | PMAX w/o purchase signals | LEAD_GENERATION | |
| Purchase | OUTCOME_SALES, CONVERSIONS, | SEARCH (brand+commercial), | WEBSITE_CONVERSION, | CONVERSIONS, CATALOG_SALES |
| | CATALOG_SALES, OFFLINE_CONVERSIONS | SHOPPING, PMAX (purchase) | JOB_APPLICANT | |
Auto-derive on ambiguity: classify by attached conversion event (purchase event > traffic objective; lead event > engagement objective).
Conditional Purchase column: if zero campaigns in workspace have purchase events attached → set `goals.roas = 0` AND set `ch_data[*].purch = 0` AND `channels[*].revenue = 0` so the canonical's KPI engine omits the Revenue/ROAS tile values gracefully (renders as `—`). The funnel classification is still useful for Awareness vs Consideration narrative.
2E. Goals (workspace plan ceilings)
Try in priority order. **For each ClickHouse priority, run the three-query gate (Bugs #2–#4 fix — MANDATORY):**
```sql
-- Query A (Bug #2): resolve table existence — no schema prefix (internal_analytics.* does not
-- exist for tenant workspaces; a prefix yields UNKNOWN_TABLE + cross-tenant suggestions)
SELECT table_name FROM mrt_database_tables
WHERE table_name LIKE '%target%' OR table_name LIKE '%plan%' OR table_name LIKE '%marketing_target%';
-- Query B (Bug #4): resolve actual column names BEFORE writing the SELECT clause
SELECT column_name, type FROM mrt_database_columns
WHERE table_name = '<resolved_table_from_A>' ORDER BY position;
-- Query C (Bug #3): freshness gate - never trust an aggregate without checking date range
SELECT MIN(date) AS first_date, MAX(date) AS last_date FROM <resolved_table>;
-- If MAX(date) < period_start_minus_1d → table is stale; surface "plan data N days stale"
-- banner and FALL THROUGH to the next priority below. Do NOT silently render 0 rows.
```
Column names vary across goal tables (`spend` vs `cost`, `impressions` vs `imps`, `conversions` vs `conv`) — never paste column names from memory.
Priority list:
1. ClickHouse `biz_marketing_targets` for the workspace_id (no prefix; if present, table fresh, columns resolved)
2. Notion "Marketing Plan" page for the workspace
3. Infer: prior-period actuals × {1.10 spend growth, 1.20 revenue growth} as a soft target
4. Fall back to zeros — canonical renders flat goal bars
goals = { spend, rev, roas, conv, cpa }
2F. Alerts (anomaly feed for "Needs your attention" — agent-derived)
Run these detectors on the fetched data:
| Detector | Severity | Alert payload |
|-------------------------------------------------------------------|------------|--------------------------------------------------------------|
| Channel ROAS in bottom quartile + spend ≥ 2× workspace median | warn | meta_left = "$<X>K · <pct>% of <ch>", meta_right = "Funnel mix" |
| Per-campaign CPA > 5× channel-average AND spend > $1K | bad | meta_left = "$<X>K wasted", meta_right = "Underperformer" |
| Per-campaign ROAS > 1.5× channel-average AND budget-util < 80% | good | meta_left = "+$<X>K opportunity", meta_right = "Top campaign" |
| Channel mix concentration > 80% on one platform | null/info | meta_left = "Channel mix", meta_right = "Heavy <ch>" |
| Spend ≥ 2× median AND ROAS in bottom quartile (per Quality Gate) | warn | meta_left = "$<X>K at risk" |
Cap at 6 visible alerts; rank by severity then |$-impact|. Each → one entry in the `alerts` block.
2G. Journeys + role bars + journey stats
Source priority:
1. GA4 `path-data` API via the workspace's GA4 connection (if connected)
2. ClickHouse `biz_attribution` for the workspace_id (no schema prefix; last-365d), rolled up by `path × position`
3. Neither available → set `journeys = []`, `roles = []`, `journey_stats = {}` (canonical renders the empty state)
journeys[i] = { filter: 'all'|'new'|'ret', share, conv, steps:[{t, ch?, conv?}], time, touches }
roles[i] = { ch, name, first: 0..100, last: 0..100 }
journey_stats = { avg_touches, avg_time, multi_channel_pct, cross_device_pct }
STEP 3: Discovery API → INJECT-marker mapping
This is the CONNECTING CONTRACT between Appendix A's INJECT markers and live data. For every INJECT point, the source endpoint + transformation + format.
Inline placeholders ({{INJECT:name}})
All ten markers fill text content of <span id="…"> elements in the topbar / title / footer.
| Marker | Source | Transformation → Format | Example |
|---|
agency_name | Step 0E | string | Apex Outdoors |
mark_letter | Step 0E | first letter of agency_name, uppercased | A |
agency_meta | Step 0E | "agency_id <id> · <N> sub-brands" | agency_id 4827 · 3 sub-brands |
last_sync_label | Step 1 post-fetch | "Live · <N> min ago" | Live · 8 min ago |
avatar_initials | Step 0E | first letters of impersonator's first+last name | JS |
owner_name | Step 0E | string | Jamie Sullivan |
owner_role | Step 0E | string | Improvado AE |
period_label | Step 0E | "<MMM D> – <MMM D>, <YYYY>" | Apr 4 – May 4, 2026 |
period_compare | Step 0E | "<MMM D> – <MMM D> vs <prev MMM D> – <prev MMM D>, <YYYY>" | Apr 4 – May 4 vs Mar 5 – Apr 3, 2026 |
trend_desc | Step 0E | "<N> days (<from> → <to>) · <attribution> attribution" | 30 days (2026-04-04 → 2026-05-04) · Data-driven attribution |
foot_source | Step 0E | string (default "Source: Improvado Discovery API · live") | Source: Improvado Discovery API · live |
Block: channels (JS const ALL_CHANNELS — variable channel count)
For each active channel from Step 2B, emit one entry. For each known-connected-platform-not-active, emit a connected: false placeholder so the source-footer "Not connected" cluster renders.
const ALL_CHANNELS = [
{
key: string,
name: string,
sub: string,
logo: string,
color: string,
connected: boolean,
spend: number,
impr: number,
clicks: number,
conv: number,
revenue: number,
prevSpend: number,
prevROAS: number
},
];
Block: trend_daily (JS const trendDaily — daily trend rows)
Length = (date_end − date_start) + 1 days. Each row is one day across the selected channels (Step 2C).
var trendDaily = [
{ day: 1, dt: '04-04', dow: 4, spend: 0, rev: 0, conv: 0 },
{ day: 2, dt: '04-05', dow: 5, spend: 0, rev: 0, conv: 0 },
{ day: N, dt: '<MM-DD>', dow: <0..6>, spend: <num>, rev: <num>, conv: <num> }
];
Block: ch_data (JS const CH_DATA — per-channel KPI snapshot, keyed by channel.key)
The KPI engine consumes this for dropdown-driven recomputation (channel filter × conversion event × attribution model). All values are sums over the period.
var CH_DATA = {
meta: { spend, rev, conv, leads, purch, ac, pv, imp, prevSpend, prevRev, prevConv, prevLeads, prevPurch },
google: { spend, rev, conv, leads, purch, ac, pv, imp, prevSpend, prevRev, prevConv, prevLeads, prevPurch },
};
Block: goals (JS const GOAL_30D — plan ceilings for goal bars)
Source per Step 2E. If unavailable, all-zeros is OK (canonical renders flat goal bars).
var GOAL_30D = { spend: 0, rev: 0, roas: 0, conv: 0, cpa: 0 };
Block: alerts (JS const DD_ALERTS — agent-derived anomaly feed)
Each entry is one card in the "Needs your attention" feed. Rank by severity then |$-impact|; cap at 6 (Step 2F).
var DD_ALERTS = [
{
severity: 'good' | 'warn' | 'bad' | null,
meta_left: string,
meta_right: string,
title: string,
body: string,
actions: [
{ label: 'Pause campaign →', outline: false },
{ label: 'Compare to control', outline: true }
]
},
];
Block: journeys (JS const JOURNEYS_ALL — top customer-journey paths)
Source: Step 2G. Filter values: 'all' (all paths), 'new' (new visitors only), 'ret' (returning).
var JOURNEYS_ALL = [
{
filter: 'all' | 'new' | 'ret',
share: string,
conv: string,
steps: [{ t: 'Meta', ch: 'meta' }, { t: 'Direct' }, { t: 'Convert', conv: true }],
time: string,
touches: string
},
];
Block: journey_stats (JS const DD_JOURNEY_STATS — single object)
Strings (not numbers) so the canonical renders units verbatim.
var DD_JOURNEY_STATS = {
avg_touches: '2.6',
avg_time: '2.3d',
multi_channel_pct: '23%',
cross_device_pct: '41%'
};
Block: roles (JS const roleData — first-touch / last-touch shares)
var roleData = [
{ ch: 'meta', name: 'Meta', first: 48, last: 31 },
{ ch: 'google', name: 'Google', first: 42, last: 55 },
{ ch: 'linkedin', name: 'LinkedIn', first: 7, last: 2 }
];
v1.1 TODO — CH_DAILY block (Bug #12)
Known limitation in v1.0: the canonical's period-segmented control (7d / 14d / 30d / QTD / YTD) computes longer-window totals via periodSliceFactor() — a multiplier against the 30-day totals — instead of slicing real per-channel daily data. QTD and YTD show scaled estimates, not actual cumulative numbers.
v1.1 fix (planned): add a CH_DAILY injection block — per-channel daily arrays (≥90 days, 365 ideal for YTD coverage) — and refactor computeKPIs() to slice CH_DAILY[key] by the active period instead of multiplying CH_DATA[key] by pf.f. Until v1.1 lands, surface "QTD/YTD shown as estimates from 30-day pacing" in the dashboard subtitle when the user picks those periods.
When v1.1 lands, this section gets a new block in STEP 3 mirroring trend_daily but per-channel:
var CH_DAILY = {
meta: { spend: [<365 floats>], rev: [<365>], conv: [<365>], leads: [<365>], purch: [<365>] },
google: { spend: [<365>], rev: [<365>], conv: [<365>], leads: [<365>], purch: [<365>] },
// … one entry per channel.key in CH_DATA …
};
STEP 1 v1.1 also extends per-platform discoveryRequestTool calls to fetch a 365-day window (instead of 30-day) so the period control can operate on real data without a re-fetch on every period switch.
Optional: role_insight (single string, plain text)
Wrap in a dedicated inline marker if you want a callout under the role bars. Set to empty string ("") to hide the callout.
| Marker | Format | Example |
|---|
role_insight | plain text, ≤ 240 chars (HTML-escaped by canonical) | Direct/Last-click dominates (70% of paths). Cross-device data needed for full multi-touch view. |
STEP 4: Build Custom Widget HTML
computeKPIs() return-contract (Bug #11 fix — DO NOT REWRITE THIS FUNCTION)
The canonical's KPI engine has an under-documented contract between computeKPIs() and the downstream renderers (renderKPIs, setKpi, the per-tile delta chips). If you regenerate computeKPIs from scratch and miss any required key, the tiles render undefined% deltas silently. Substitute data, never the function body. The function MUST return all 12 of these keys:
{
pf: { f: <number>, days: <number>, label: <string> },
spend: <number>,
rev: <number>,
conv: <number>,
roas: <number>,
cpa: <number>,
prevSpend: <number>,
prevRev: <number>,
prevConv: <number>,
prevRoas: <number>,
prevCpa: <number>,
keys: <string[]> // selectedChannelKeys() output for label rendering
}
If any of these is missing, the corresponding KPI tile will show undefined% for the Δ chip or a blank value. The fix is to substitute Discovery API data into the canonical, not to rewrite the function.
Substitution rules
Take the canonical HTML from Appendix A. Find every INJECT marker. Substitute per Step 3 mapping.
For each {{INJECT:name}} placeholder:
Find the literal string {{INJECT:name}} in canonical
Replace with computed value (string)
For each block marker pair:
// === INJECT: <name> ===
<existing default content (empty array OR placeholder)>
// === END INJECT: <name> ===
Replace the inner content with the computed JSON / JS literal.
KEEP the marker pair intact (so future regen can find them).
HARD RULE: if you find yourself writing <style> or restructuring <div class="kpi"> — STOP. You are regenerating, not substituting. Re-load Appendix A and substitute only at marked points. The canonical is NEVER edited per-render.
After substitution, the entire HTML is ONE string — assign it as componentCode of a single custom-component widget.
STEP 5: Save Dashboard via /business-intelligence-editor
Assemble dashboard config:
{
"dashboardTitle": "Cross-Channel Performance",
"dashboardSubtitle": "{N_active_channels} channels | {period_label} | Updated {last_sync_label}",
"dashboardUrl": "clients/template/dashboards/CrossChannelEditableDashboard.tsx",
"isMenuItem": true,
"defaultTimePeriod": "30",
"editState": {
"schemaVersion": 2,
"appearance": {
"hideTitle": true,
"hideFilters": true,
"colorMode": "light"
},
"persona": "executive | growth | marketer | analyst",
"widgets": [
{
"id": "main-grid-1",
"type": "custom-component",
"props": {
"type": "custom-component",
"renderMode": "html",
"componentCode": "<the entire substituted HTML from Step 4>",
"gridWidth": 12,
"gridHeight": "huge",
"chromeless": true,
"showTitle": false,
"title": "Cross-Channel Performance",
"preloadLibraries": [],
"inheritFiltersFromDashboard": false,
"customSqlEnabled": false
}
}
],
"layout": {
"items": [
{ "id": "main-grid-1", "x": 0, "y": 0, "w": 12, "h": 28 }
]
}
}
}
Schema gotchas (Bugs #5–#8 fixes — validated 2026-05-08):
editState must be nested inside the config root (sibling to dashboardTitle/dashboardUrl), NOT at the top level.- The widget discriminator is
"type": "custom-component" — NOT "widgetType". The discriminator must appear at the widget level AND be duplicated inside props (the schema requires both for the custom-component shape).
- Layout items use
"id": — NOT "i": (the latter is react-grid-layout's internal field name; BIE's wrapping schema renames it).
- Validate before save: run
npx tsx .claude/skills/business-intelligence-editor/skill-cli.ts validate-widget --config '<full_config_json>' and stop on any non-zero exit. Treat all four schema bugs as a single "untested schema" antibody.
appearance.hideTitle + hideFilters + colorMode: "light" matches the precedent set by weekly-creative-performance v6 and daily-performance-report v2 — the canonical owns its own header (workspace name + period chip + date picker). The dashboard chrome would duplicate / conflict.
dashboardUrl: "clients/template/dashboards/CrossChannelEditableDashboard.tsx" — emit the canonical host-TSX path directly (precedent: full-marketing-audit/dashboard-template.json:104 and weekly-creative-performance/dashboard-template.json:112 v7.4.1+). The Miras viewer (useRepoDashboardData in main/components/repo/hooks/useRepoDashboardData.ts:55) reads dashboard_url verbatim as the file path on /experimental/agent/api/repo/file?repo_id=dashboard&path=<dashboardUrl>; emitting the canonical path removes the slug→path indirection that previously depended on BIE skill-cli's HARDCODED_DASHBOARD_URL override (skill-cli.ts:14). Earlier drafts shipped the slug "cmo-cross-channel"; that worked only when the save flowed through skill-cli — any save that bypassed it (direct curl/fetch, pre-2026-01-30 PR #415 historical save, fixture/migration) wrote the slug into the DB and the viewer 404'd with Failed to fetch repo file: 500 (Dashboard file API 404 Not Found). There is no CmoCrossChannelDashboard.tsx host TSX — CrossChannelEditableDashboard.tsx renders this skill's dashboard via settings_id.
appearance.persona rides through BIE's schema via .passthrough(). The host TSX reads it on mount to decide which tabs to show (executive → Overview only; others → all 3).
Layout: ONE widget covering full width × tall enough for the entire dashboard (12 cols × ~28 rows in the BIE grid). The dashboard is intentionally a single tall widget — there is NO second widget below it.
Save:
Skill('business-intelligence-editor') with the assembled JSON
# CLI form: npx tsx .claude/skills/business-intelligence-editor/skill-cli.ts save --config '<full_config_json>'
Open preview (always with --production):
python3 frontend-cli.py open-preview --production "clients/template/dashboards/CrossChannelEditableDashboard.tsx?settings_id={NEW_ID}"
Present to user with:
- Dashboard URL
- 1-line summary (e.g. "Meta = 60% of spend, ROAS 2.36×; LinkedIn lagging at 0.31× — $28K/mo at risk")
- Auto-refresh suggestion (per § Output Mode → Auto-Refresh Strategy)
STEP 6: Auto-Refresh Suggestion
After the user sees the dashboard, offer:
"Want me to refresh this every Monday at 8 AM with last-30d numbers?"
If yes → scheduleChatTool with cadence weekly-monday-08:00-user-tz. Each scheduled run re-executes the skill: re-fetches Discovery API, re-substitutes Appendix A, overwrites the saved dashboard at the same dashboardUrl. Channel logos / palette / structure stay frozen — only the data churns.
If onboarding said "monthly leadership review" → suggest 1st-of-month-08:00 instead of weekly.
Anti-patterns (HARD STOP if you find yourself doing these)
- Regenerating HTML from scratch. The canonical (Appendix A) is the source of truth. Only substitute at INJECT markers.
- Patching the live widget in place instead of fixing § Appendix A. (Bug #13.) If you find a defect at render time (e.g., a leftover
rng() reference, a missing --c color), fix the canonical at § Appendix A and re-run from STEP 4 — do not patch the saved widget directly. The next refresh re-substitutes from § Appendix A and silently overwrites your patch. Always either fix at the source or batch all live patches into one save AFTER re-fetching the current widget state.
- SQL composed without first reading
mrt_database_tables + mrt_database_columns. (Bugs #2–#4.) Three-query gate before any aggregate: tables → columns → MIN/MAX(date). Memorized table or column names are stale; verify on every run.
- Schema fields hand-typed from memory instead of
validate-widget-checked. (Bugs #5–#8.) Run validate-widget CLI on the assembled config before save. Stop on non-zero exit.
- Leftover helper references surviving canonical regeneration. (Bug #10.) Search the assembled HTML for
rng(, genDaily, seed( — any helper that isn't defined will throw ReferenceError at render. Replace with Math.random() or the actual helper definition before save.
- Decomposing the dashboard into multiple BIE widgets. Use ONE custom-component widget. Built-in
kpi-widget / chart / data-table MUST NOT appear in the dashboard composition.
- Querying data outside Discovery API. Per Rule 0, every fetch goes through
discoveryRequestTool. No clickhousePalantirTool for marketing data. The single permitted ClickHouse use is for workspace plan/goal lookup (biz_workspace_plans / biz_marketing_targets).
- Pasting client-specific values into the canonical.
{{INJECT:agency_name}} is the slot for the agency name; never hardcode.
- Hand-rolled recharts grid. This skill does NOT produce a recharts radar / sankey / funnel composition. The canonical's exact chart set (KPI strip, daily trend, donut, scatter, channel matrix, journey paths, role bars) IS the deliverable.
- More than 6 alerts in the feed. The 6-card cap is the value — forces severity × $-impact selection.
- Alerts without provenance. Every alert MUST be derivable from the Discovery API response. No fabricated or memorized client text.
- Editing dashboard chrome.
hideTitle: true, hideFilters: true, colorMode: "light" — required for visual contract.
- Mocked or static data in production. If Discovery API fails — surface the failure honestly with
— placeholders + retry banner. Do NOT fall through to a fixture.
- Falling back to
daily-performance-report for "where is my budget"-style queries. Cross-channel funnel intent → CMO dashboard, period.
Tone Rules — scoped to Alerts feed only
Apply to the alerts feed cards (NOT to KPI tiles, NOT to channel matrix — those are factual/structural):
- Confident verbs only — "Pause", "Scale", "Cut", "Shift", "Rebalance". Not "consider pausing".
- No hedging. Forbidden: "may want to", "could potentially", "if applicable".
- Money-tied or skip. Each alert's
meta_left MUST show $-value (gain or savings) when the detector is dollar-based. If you can't quantify, drop the card.
- Calibration threshold: ≥7d data AND (≥30 conversions OR ≥$500 spend) on the drill entity. Below threshold → skip the candidate.
- 6-card cap — the constraint is the value.
- No education. Audience knows the metrics.
Failure modes — what to render when something goes wrong
Per Rule 0 — never mock, never fall through. Behavior matrix:
| Failure mode | UI behavior |
|---|
| Connection missing for a channel | Drop that channel from channels; if active count drops <2 → fall back to Mode C single-channel chart |
| API returns NULL/empty for a metric on one channel | That channel's contribution = 0; blended KPI still renders; channel-row in the matrix shows — for missing cells |
| Whole API call fails (timeout, 5xx) | Topbar last_sync_label reads "Live · <T> min ago — <channel> delayed (retry <T>)". Render last-known cached values if <2h old; else — placeholders WITH explicit error indicator (not the canonical's INJECT placeholders) |
biz_marketing_targets row missing | Use last-period actuals × growth target as plan baseline; show banner "No plan set for {period} — using last-period actuals × 1.10. [Set plan]" |
biz_marketing_targets table date range pre-dates the requested period (Bug #3) | Surface "plan data N days stale" banner; fall through to the next priority source (Notion plan page → prior-period × growth target). Do NOT silently render goals = {0,0,0,0,0} and call it "no plan". |
| Workspace has only 1 active channel | Dashboard requires ≥2 channels. Surface message; offer Mode C single-channel chart instead. |
| GA4 path-data + biz_attribution both unavailable | journeys = [], roles = [], journey_stats = {} — canonical renders the Customer Journey tab's empty states cleanly. |
Quality gates
- No empty tiles in the KPI strip. If a metric is fundamentally unavailable on every active channel — render the tile but with
— and an explicit reason (#k-spend-pre text).
- Conditional Purchase column. If zero campaigns in workspace have purchase events attached → set
goals.roas = 0 AND ch_data[*].purch = 0 AND channels[*].revenue = 0. The canonical's KPI engine handles the cascade — Revenue and ROAS tiles render —.
- Comparative analysis on the scatter. Each bubble is sized by conversions; the diagonal split (low/high spend, low/high ROAS) labels Scale/Keep/Test/Fix quadrants relatively, not against an absolute threshold.
- $-impact callouts. If any channel has spend ≥ 2× the median AND ROAS in the bottom quartile → emit a single
alerts entry with meta_left:"$<X>K at risk".
- Empty state, not error state. When Discovery returns 0 rows for a platform, render the canonical's "No data — connect a channel via Discovery API" skeleton — never an unstyled error.
- Data freshness. Live-API call timestamp shown in topbar via
last_sync_label. If stale >2h, banner: "Data delayed — last fetch <HH:mm UTC>".
- Visual contract. Inter typography, indigo
#4b5cf2 accent, platform brand colors per channel. Do not deviate.
Appendix A — Canonical Widget HTML (single source of truth)
This is the canonical visual contract for the CMO Cross-Channel Dashboard. The widget's componentCode (per Step 5) is THIS HTML with every INJECT marker substituted by Discovery API data (per Step 3).
LLM agents performing this skill: load this section as raw text, find INJECT markers, substitute per § STEP 3 Discovery API mapping, output the result as the widget's componentCode string.
Mirrored at algorithms/Design/web/output/marketing-os/uc-cmo-1-template.html in the team-internal Marketing OS / tcs-chrome-extension repo (with the window.__DASHBOARD_DATA__ injection variant — the JS-global pattern is the design-iteration source). The marker-style HTML below is the runtime contract for this skill. Edit BOTH atomically when the structural template changes.
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=1280">
<title>UC-CMO-1 — Cross-Channel Performance Template</title>
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800&family=JetBrains+Mono:wght@400;500&display=swap" rel="stylesheet">
<style>
:root{
--bg:#f5f6fa; --surface:#ffffff; --surface-2:#fafbfd; --surface-3:#f4f5f9;
--border:#e7eaf0; --border-strong:#d4d9e3; --border-soft:#eef0f5;
--ink:#0b1220; --ink-2:#3a4358; --ink-3:#6b7589; --ink-4:#9aa3b6;
--accent:#0F4C81; --accent-2:#E8B43F; --accent-soft:#eef0ff; --accent-deep:#0a3b66;
--good:#0fae74; --good-soft:#e6f7ef; --good-deep:#0a8054;
--bad:#e1453f; --bad-soft:#fdebea; --bad-deep:#c63631;
--warn:#d68a13; --warn-soft:#fff5e1; --warn-deep:#a86a06;
--grid:#eef0f5;
--shadow:0 1px 2px rgba(15,23,42,.04), 0 4px 14px rgba(15,23,42,.05);
--shadow-lg:0 8px 28px rgba(15,23,42,.08);
--radius:14px; --radius-sm:10px;
}
*{box-sizing:border-box}
html,body{margin:0;padding:0;background:var(--bg);color:var(--ink);font-family:'Inter',-apple-system,sans-serif;font-feature-settings:"cv11","ss01","tnum";font-size:13px;line-height:1.45;-webkit-font-smoothing:antialiased}
a{color:inherit}
button{font-family:inherit}
.app{max-width:1480px;margin:0 auto;padding:0 28px 60px}
.topbar{display:flex;align-items:center;gap:14px;padding:14px 0 12px;flex-wrap:wrap}
.brand{display:flex;align-items:center;gap:10px;font-weight:700;font-size:14px;letter-spacing:-.01em}
.brand-mark{width:26px;height:26px;border-radius:7px;background:linear-gradient(135deg,#0F4C81,#E8B43F);display:grid;place-items:center;color:#fff;font-weight:800;font-size:12px;box-shadow:0 4px 10px rgba(75,92,242,.3)}
.brand-sub{color:var(--ink-3);font-weight:500;font-size:12px}
.crumbs{display:flex;align-items:center;gap:6px;color:var(--ink-3);font-size:12px;margin-left:6px}
.crumbs span{opacity:.6}
.crumbs b{color:var(--ink-2);font-weight:600}
.topbar .spacer{flex:1}
.iconbtn{height:32px;padding:0 10px;border-radius:7px;background:var(--surface);border:1px solid var(--border);display:inline-flex;align-items:center;gap:5px;font-size:12px;font-weight:500;color:var(--ink-2);cursor:pointer;transition:.15s}
.iconbtn:hover{border-color:var(--border-strong);background:#fff}
.iconbtn svg{width:13px;height:13px}
.live{display:inline-flex;align-items:center;gap:6px;padding:0 10px;height:32px;border-radius:7px;background:var(--good-soft);color:var(--good-deep);font-size:11.5px;font-weight:600;border:1px solid #cde9d9}
.live .dot{width:6px;height:6px;border-radius:50%;background:var(--good);box-shadow:0 0 0 3px rgba(15,174,116,.18);animation:p 2s ease-in-out infinite}
@keyframes p{50%{box-shadow:0 0 0 6px rgba(15,174,116,0)}}
.avatar{width:30px;height:30px;border-radius:50%;background:linear-gradient(135deg,#22d3ee,#4b5cf2);color:#fff;display:grid;place-items:center;font-weight:600;font-size:11px}
.tabs-nav{display:flex;align-items:center;gap:2px;border-bottom:1px solid var(--border);margin-bottom:18px;overflow-x:auto;flex-wrap:wrap}
.tab-btn{position:relative;height:38px;padding:0 16px;border:0;background:transparent;font-size:13px;font-weight:500;color:var(--ink-3);cursor:pointer;display:inline-flex;align-items:center;gap:6px;border-bottom:2px solid transparent;margin-bottom:-1px;transition:.15s;white-space:nowrap}
.tab-btn:hover{color:var(--ink-2)}
.tab-btn.on{color:var(--ink);border-bottom-color:var(--accent);font-weight:600}
.tab-btn .badge{display:inline-grid;place-items:center;min-width:18px;height:18px;padding:0 5px;border-radius:9px;background:var(--accent-soft);color:var(--accent-deep);font-size:10px;font-weight:700}
.title-row{display:flex;align-items:center;justify-content:space-between;gap:18px;margin-bottom:14px;flex-wrap:wrap}
.title-row h1{margin:0;font-size:22px;font-weight:700;letter-spacing:-.02em;display:flex;align-items:center;gap:10px}
.title-row .meta-line{display:flex;align-items:center;gap:10px;color:var(--ink-3);font-size:12px;flex-wrap:wrap}
.filters{display:flex;align-items:center;gap:6px;flex-wrap:wrap;margin-bottom:18px}
.seg{display:inline-flex;background:var(--surface);border:1px solid var(--border);border-radius:8px;padding:2px;height:32px;box-shadow:var(--shadow)}
.seg button{height:26px;padding:0 11px;border:0;background:transparent;font-size:11.5px;font-weight:500;color:var(--ink-3);border-radius:6px;cursor:pointer}
.seg button.on{background:var(--ink);color:#fff;box-shadow:0 1px 2px rgba(0,0,0,.05)}
.fgroup{display:flex;align-items:center;gap:6px;padding:0 10px;background:var(--surface);border:1px solid var(--border);border-radius:7px;font-size:12px;color:var(--ink-2);cursor:pointer;height:32px;box-shadow:var(--shadow)}
.fgroup:hover{border-color:var(--border-strong)}
.fgroup b{color:var(--ink);font-weight:600}
.fgroup svg.ic{width:12px;height:12px;color:var(--ink-4)}
.fgroup svg.ch{width:10px;height:10px;color:var(--ink-4);margin-left:2px}
.fgroup.active{background:var(--accent-soft);border-color:#c5cbff;color:var(--accent-deep)}
.fgroup.active b{color:var(--accent-deep)}
.filters .right{margin-left:auto;display:flex;gap:6px;align-items:center}
.compare-tag{display:inline-flex;align-items:center;gap:6px;padding:0 10px;height:32px;border-radius:7px;background:var(--accent-soft);color:var(--accent-deep);font-size:11.5px;font-weight:600;border:1px solid #d6dafd}
.compare-tag svg{width:11px;height:11px}
.tab-section{display:none}
.tab-section.on{display:block;animation:fadeIn .25s ease}
@keyframes fadeIn{from{opacity:0;transform:translateY(4px)}to{opacity:1;transform:translateY(0)}}
.kpis{display:grid;grid-template-columns:repeat(5,1fr);gap:12px;margin-bottom:16px}
.kpi{background:var(--surface);border:1px solid var(--border);border-radius:var(--radius);padding:16px;box-shadow:var(--shadow);position:relative;display:flex;flex-direction:column}
.kpi-head{display:flex;align-items:center;justify-content:space-between;margin-bottom:12px;gap:6px}
.kpi-label{font-size:11px;text-transform:uppercase;letter-spacing:.06em;color:var(--ink-3);font-weight:600;letter-spacing:.05em}
.kpi-val{display:flex;align-items:baseline;line-height:1;margin-bottom:6px}
.kpi-val .num{font-size:30px;font-weight:700;letter-spacing:-.022em;color:var(--ink);line-height:1}
.kpi-val .currency{font-size:14px;font-weight:600;color:var(--ink-4);margin-right:2px;align-self:flex-start;margin-top:3px}
.kpi-val .suffix{font-size:14px;font-weight:600;color:var(--ink-3);margin-left:2px;align-self:flex-end;margin-bottom:3px}
.kpi-prev{font-size:11px;color:var(--ink-4);font-weight:500;margin-bottom:10px;line-height:1.4}
.delta{display:inline-flex;align-items:center;gap:3px;padding:2px 7px;border-radius:6px;font-size:11px;font-weight:600;flex-shrink:0}
.delta.up{background:var(--good-soft);color:var(--good-deep)}
.delta.down{background:var(--bad-soft);color:var(--bad-deep)}
.delta svg{width:9px;height:9px}
.kpi-goal{margin-top:auto;display:flex;flex-direction:column;gap:4px}
.kpi-goal-bar{height:5px;background:var(--surface-3);border-radius:99px;overflow:hidden}
.kpi-goal-bar i{display:block;height:100%;background:linear-gradient(90deg,var(--accent),var(--accent-2));border-radius:99px}
.kpi-goal-bar i.good{background:linear-gradient(90deg,var(--good),#3bc792)}
.kpi-goal-meta{display:flex;justify-content:space-between;font-size:10px;color:var(--ink-3);font-weight:500}
.grid{display:grid;gap:14px;margin-bottom:14px}
.grid.t1{grid-template-columns:1.55fr 1fr}
.grid.t2{grid-template-columns:1.3fr 1fr}
.grid.c-2{grid-template-columns:repeat(2,1fr)}
.grid.c-3{grid-template-columns:repeat(3,1fr)}
.card{background:var(--surface);border:1px solid var(--border);border-radius:var(--radius);padding:18px;box-shadow:var(--shadow)}
.card.flush{padding:0}
.card-head{display:flex;align-items:flex-start;justify-content:space-between;margin-bottom:14px;gap:12px;flex-wrap:wrap}
.card-head h3{margin:0 0 2px;font-size:14px;font-weight:700;letter-spacing:-.01em}
.card-head .desc{font-size:12px;color:var(--ink-3)}
.card-head .actions{display:flex;align-items:center;gap:6px;flex-shrink:0;flex-wrap:wrap}
.ctabs{display:inline-flex;background:var(--surface-3);border-radius:7px;padding:2px}
.ctabs button{font-family:inherit;height:24px;padding:0 9px;border:0;background:transparent;font-size:11px;font-weight:500;color:var(--ink-3);border-radius:5px;cursor:pointer;transition:.15s}
.ctabs button.on{background:#fff;color:var(--ink);box-shadow:0 1px 2px rgba(0,0,0,.06)}
.ctabs button:hover:not(.on){color:var(--ink-2)}
.legend-key{display:inline-flex;align-items:center;gap:5px;font-size:11px;color:var(--ink-3);font-weight:500}
.legend-key i{width:10px;height:10px;border-radius:3px}
.trend-stats{display:grid;grid-template-columns:repeat(5,1fr);gap:14px;margin-bottom:6px;padding:10px 0;border-bottom:1px solid var(--border-soft)}
.trend-stat .lab{display:flex;align-items:center;gap:6px;font-size:10.5px;color:var(--ink-3);font-weight:600;text-transform:uppercase;letter-spacing:.04em;margin-bottom:3px}
.trend-stat .lab i{width:10px;height:10px;border-radius:3px;display:inline-block}
.trend-stat .v{font-size:17px;font-weight:700;letter-spacing:-.01em}
.trend-svg{width:100%;height:300px;display:block;margin-top:12px}
.feed{display:flex;flex-direction:column;gap:9px;max-height:560px;overflow-y:auto;padding-right:4px}
.feed::-webkit-scrollbar{width:4px}
.feed::-webkit-scrollbar-thumb{background:var(--border-strong);border-radius:2px}
.feed-item{display:grid;grid-template-columns:38px 1fr;gap:11px;padding:12px;border-radius:11px;background:var(--surface-2);border:1px solid var(--border);transition:.15s;cursor:pointer;position:relative}
.feed-item:hover{border-color:var(--border-strong);transform:translateX(-1px);box-shadow:var(--shadow)}
.feed-item.warn{background:linear-gradient(135deg,#fffaf0 0%,#fff5e1 100%);border-color:#fce4b1}
.feed-item.bad{background:linear-gradient(135deg,#fef5f4 0%,#fde7e6 100%);border-color:#fac9c7}
.feed-item.good{background:linear-gradient(135deg,#f1faf5 0%,#e6f7ef 100%);border-color:#c2e8d2}
.feed-item .priority-bar{position:absolute;left:0;top:8px;bottom:8px;width:3px;border-radius:0 2px 2px 0;background:var(--accent)}
.feed-item.warn .priority-bar{background:var(--warn)}
.feed-item.bad .priority-bar{background:var(--bad)}
.feed-item.good .priority-bar{background:var(--good)}
.feed-icon{width:38px;height:38px;border-radius:10px;background:#fff;display:grid;place-items:center;border:1px solid var(--border);color:var(--accent);box-shadow:0 1px 2px rgba(0,0,0,.04)}
.feed-icon.warn{color:var(--warn-deep)}
.feed-icon.bad{color:var(--bad-deep)}
.feed-icon.good{color:var(--good-deep)}
.feed-icon svg{width:18px;height:18px}
.feed-body .h{font-size:12.5px;font-weight:700;color:var(--ink);line-height:1.35;margin-bottom:2px}
.feed-body .meta{display:flex;gap:8px;font-size:10px;color:var(--ink-4);font-weight:500;text-transform:uppercase;letter-spacing:.04em;margin-bottom:5px}
.feed-body .meta b{color:var(--ink-3);font-weight:600}
.feed-body .b{font-size:11.5px;color:var(--ink-2);line-height:1.5}
.feed-body .a{margin-top:7px;display:flex;gap:8px;font-size:11px}
.feed-body .a a{color:var(--accent-deep);text-decoration:none;font-weight:600;padding:3px 8px;background:rgba(75,92,242,.08);border-radius:5px;cursor:pointer}
.feed-body .a a:hover{background:rgba(75,92,242,.15)}
.feed-body .a a.outline{background:transparent;border:1px solid var(--border-strong);color:var(--ink-3)}
.donut-card{display:grid;grid-template-columns:auto 1fr;gap:24px;align-items:center;min-height:300px}
.donut-big{position:relative;width:240px;height:240px;flex-shrink:0;display:grid;place-items:center}
.donut-big svg{width:240px;height:240px}
.donut-big-label{position:absolute;text-align:center;line-height:1.1}
.donut-big-label .v{font-size:24px;font-weight:700;letter-spacing:-.02em}
.donut-big-label .v .currency{font-size:14px;font-weight:600;color:var(--ink-4);margin-right:1px;vertical-align:0.35em}
.donut-big-label .l{font-size:11px;color:var(--ink-3);text-transform:uppercase;letter-spacing:.05em;font-weight:600;margin-top:2px}
.donut-big-label .sub{font-size:10.5px;color:var(--ink-4);margin-top:1px;font-weight:500}
.donut-legend{flex:1;display:flex;flex-direction:column;gap:6px}
.donut-legend .row{display:grid;grid-template-columns:24px 1fr auto auto;align-items:center;gap:10px;padding:7px 8px;border-radius:7px;cursor:pointer;font-size:12px;border:1px solid transparent}
.donut-legend .row:hover{background:var(--surface-3);border-color:var(--border)}
.donut-legend .pc{color:var(--ink-3);font-weight:500;width:46px;text-align:right;font-size:11.5px}
.donut-legend .va{color:var(--ink);font-weight:700;width:60px;text-align:right;font-size:12.5px}
.scatter-svg{width:100%;height:340px;display:block}
.scatter-legend{display:flex;justify-content:space-between;align-items:center;margin-top:8px;padding:8px 0 0;border-top:1px solid var(--border-soft);font-size:11px;color:var(--ink-3)}
.scatter-legend .quadrant-key{display:grid;grid-template-columns:repeat(4,1fr);gap:10px;flex:1}
.qk{display:flex;flex-direction:column;gap:1px}
.qk b{color:var(--ink);font-size:11.5px;font-weight:700}
.qk small{font-size:10.5px;color:var(--ink-3);font-weight:500}
.ch-table-wrap{overflow-x:auto}
table{width:100%;border-collapse:collapse;font-size:12px}
th{text-align:left;font-weight:600;font-size:10.5px;text-transform:uppercase;letter-spacing:.05em;color:var(--ink-3);padding:11px 9px;border-bottom:1px solid var(--border);background:var(--surface-3);white-space:nowrap;cursor:pointer;user-select:none}
th.r,td.r{text-align:right}
th .sort-arrow{display:inline-block;color:var(--ink-4);margin-left:3px;font-size:9px}
th.sorted .sort-arrow{color:var(--accent-deep)}
td{padding:13px 9px;border-bottom:1px solid var(--border);vertical-align:middle}
tr:last-child td{border-bottom:0}
tr:hover td{background:var(--surface-2)}
td.ch{font-weight:600;color:var(--ink)}
.ch-cell{display:flex;align-items:center;gap:10px}
.ch-cell .info{display:flex;flex-direction:column}
.ch-cell .nm{font-weight:600;font-size:13px;color:var(--ink);letter-spacing:-.01em}
.ch-cell .sub{font-size:10.5px;color:var(--ink-4);font-weight:500;margin-top:1px}
.roas-tag{display:inline-flex;align-items:center;gap:3px;padding:3px 8px;border-radius:6px;font-weight:700;font-size:11.5px;font-variant-numeric:tabular-nums}
.roas-tag.up{background:var(--good-soft);color:var(--good-deep)}
.roas-tag.mid{background:var(--accent-soft);color:var(--accent-deep)}
.roas-tag.warn{background:var(--warn-soft);color:var(--warn-deep)}
.roas-tag.down{background:var(--bad-soft);color:var(--bad-deep)}
.spark{height:26px;width:100px;display:block}
.conn-status{display:inline-flex;align-items:center;gap:6px;padding:5px 10px;border-radius:7px;background:var(--warn-soft);border:1px solid #fce4b1;color:var(--warn-deep);font-size:11px;font-weight:600}
.conn-status svg{width:12px;height:12px}
.conn-status .link{color:var(--warn-deep);text-decoration:underline;cursor:pointer}
.lg{width:28px;height:28px;border-radius:7px;display:grid;place-items:center;flex-shrink:0;overflow:hidden}
.lg svg{width:18px;height:18px;display:block}
.lg.sm{width:20px;height:20px;border-radius:5px}
.lg.sm svg{width:13px;height:13px}
.lg.xs{width:16px;height:16px;border-radius:4px}
.lg.xs svg{width:10px;height:10px}
.lg.meta{background:#1877F2}
.lg.google{background:#fff;border:1px solid var(--border)}
.lg.tiktok{background:#000}
.lg.pinterest{background:#E60023}
.lg.snapchat{background:#FFFC00}
.lg.youtube{background:#FF0000}
.lg.reddit{background:#FF4500}
.lg.dv360{background:#34A853}
.lg.klaviyo{background:#1F2024}
.lg.linkedin{background:#0A66C2}
.lg.x{background:#0F1419}
.lg.bing{background:#00A4EF}
.journey-list{display:flex;flex-direction:column;gap:11px}
.journey-row{padding:13px;border:1px solid var(--border);border-radius:10px;background:var(--surface-2)}
.jr-head{display:flex;align-items:center;justify-content:space-between;margin-bottom:9px}
.jr-share{font-size:14px;font-weight:700;letter-spacing:-.01em}
.jr-share .conv{font-size:11px;color:var(--ink-3);font-weight:500;margin-left:5px}
.jr-meta{font-size:10.5px;color:var(--ink-3);font-weight:500;display:flex;gap:8px}
.jr-meta b{color:var(--ink-2);font-weight:700}
.jr-path{display:flex;align-items:center;gap:7px;flex-wrap:wrap}
.jr-step{display:inline-flex;align-items:center;gap:6px;padding:5px 10px;background:var(--surface);border:1px solid var(--border);border-radius:7px;font-size:11.5px;font-weight:500;color:var(--ink-2)}
.jr-step.conv{background:var(--good-soft);border-color:#cde9d9;color:var(--good-deep);font-weight:700}
.jr-step.conv svg{width:11px;height:11px}
.jr-arrow{color:var(--ink-4);display:flex;align-items:center}
.jr-arrow svg{width:12px;height:12px}
.jr-direct{background:#f1f3f7;color:var(--ink-3);font-weight:600;font-style:italic}
.stat-grid{margin-top:14px;padding:14px;background:var(--surface-2);border-radius:11px;display:grid;grid-template-columns:repeat(4,1fr);gap:14px;border:1px solid var(--border)}
.stat-grid .lab{font-size:10px;color:var(--ink-3);text-transform:uppercase;letter-spacing:.05em;font-weight:600;margin-bottom:3px}
.stat-grid .v{font-size:20px;font-weight:700;letter-spacing:-.02em}
.stat-grid .v small{font-size:11px;color:var(--ink-3);font-weight:500;margin-left:3px}
.demo-grid{display:grid;grid-template-columns:repeat(3,1fr);gap:14px}
.demo-block{background:var(--surface);border:1px solid var(--border);border-radius:11px;padding:14px;box-shadow:var(--shadow)}
.demo-block h4{margin:0 0 12px;font-size:12.5px;font-weight:700;display:flex;align-items:center;gap:7px;letter-spacing:-.01em}
.demo-block h4 svg{width:14px;height:14px;color:var(--ink-3)}
.demo-bar{display:grid;grid-template-columns:90px 1fr 90px;align-items:center;gap:10px;font-size:11.5px;padding:5px 0}
.demo-bar .nm{color:var(--ink-2);font-weight:500}
.demo-bar .bar{height:8px;background:var(--surface-3);border-radius:99px;overflow:hidden}
.demo-bar .bar i{display:block;height:100%;border-radius:99px}
.demo-bar .v{font-weight:700;color:var(--ink);text-align:right;font-variant-numeric:tabular-nums}
.demo-bar .v small{display:block;color:var(--ink-3);font-weight:500;font-size:10px;text-transform:uppercase;letter-spacing:.03em}
.new-vs-ret{display:grid;grid-template-columns:1fr 1fr;gap:12px;margin-bottom:14px}
.nvr-card{padding:14px;border-radius:11px;border:1px solid var(--border);background:var(--surface);display:flex;flex-direction:column;gap:6px;position:relative;overflow:hidden}
.nvr-card.new{border-color:#c5cbff;background:linear-gradient(135deg,#fff 0%,#eef0ff 100%)}
.nvr-card.ret{border-color:#cde9d9;background:linear-gradient(135deg,#fff 0%,#e6f7ef 100%)}
.nvr-card .lab{font-size:11px;text-transform:uppercase;letter-spacing:.05em;color:var(--ink-3);font-weight:600}
.nvr-card .v{font-size:24px;font-weight:700;letter-spacing:-.02em;display:flex;align-items:center;gap:8px}
.nvr-card .meta-line{display:flex;justify-content:space-between;font-size:11px;color:var(--ink-2);font-weight:500;flex-wrap:wrap;gap:4px}
.nvr-card .meta-line b{color:var(--ink);font-weight:700}
.foot{display:flex;align-items:center;justify-content:space-between;color:var(--ink-3);font-size:11px;padding:18px 0 0;border-top:1px solid var(--border);margin-top:24px;flex-wrap:wrap;gap:12px}
.foot .src{display:inline-flex;gap:10px;flex-wrap:wrap;align-items:center}
.foot .src .lg{width:18px;height:18px;border-radius:4px;opacity:.5}
.foot .src .lg.connected{opacity:1}
.foot .src .lg svg{width:11px;height:11px}
.axis-label{font-size:10px;fill:var(--ink-4);font-family:inherit}
.axis-label.bold{font-weight:700;fill:var(--ink)}
.grid-line{stroke:var(--grid);stroke-width:1}
.grid-line.q{stroke:var(--border-strong);stroke-dasharray:4 4}
</style>
</head>
<body>
<svg width="0" height="0" style="position:absolute" aria-hidden="true">
<defs>
<symbol id="lg-meta" viewBox="0 0 24 24"><path fill="#fff" d="M6.915 4.03c-1.968 0-3.683 1.28-4.871 3.113C.704 9.208 0 11.883 0 14.449c0 .706.07 1.369.21 1.973a6.624 6.624 0 0 0 .265.86 5.297 5.297 0 0 0 .371.761c.696 1.159 1.818 1.927 3.593 1.927 1.497 0 2.633-.671 3.965-2.444.76-1.012 1.144-1.626 2.663-4.32l.756-1.339.186-.325c.061.1.121.196.183.3l2.152 3.595c.724 1.21 1.665 2.556 2.47 3.314 1.046.987 1.992 1.22 3.06 1.22 1.075 0 1.876-.355 2.455-.843a3.743 3.743 0 0 0 .81-.973c.542-.939.861-2.127.861-3.745 0-2.72-.681-5.357-2.084-7.45-1.282-1.912-2.957-2.93-4.716-2.93-1.047 0-2.088.467-3.053 1.308-.652.57-1.257 1.29-1.82 2.05-.69-.875-1.335-1.547-1.958-2.056-1.182-.966-2.315-1.303-3.454-1.303zm10.16 2.053c1.147 0 2.188.758 2.992 1.999 1.132 1.748 1.647 4.195 1.647 6.4 0 1.548-.368 2.9-1.839 2.9-.58 0-1.027-.23-1.664-1.004-.496-.601-1.343-1.878-2.832-4.358l-.617-1.028a44.908 44.908 0 0 0-1.255-1.98c.07-.109.141-.224.211-.327 1.12-1.667 2.118-2.602 3.358-2.602zm-10.201.553c1.265 0 2.058.791 2.675 1.446.307.327.737.871 1.234 1.579l-1.02 1.566c-.757 1.163-1.882 3.017-2.837 4.338-1.191 1.649-1.81 1.817-2.486 1.817-.524 0-1.038-.237-1.383-.794-.263-.426-.464-1.13-.464-2.046 0-2.221.63-4.535 1.66-6.088.454-.687.964-1.226 1.533-1.533a2.264 2.264 0 0 1 1.088-.285z"/></symbol>
<symbol id="lg-google" viewBox="0 0 24 24"><path fill="#4285F4" d="M22.56 12.25c0-.78-.07-1.53-.2-2.25H12v4.26h5.92c-.26 1.37-1.04 2.53-2.21 3.31v2.77h3.57c2.08-1.92 3.28-4.74 3.28-8.09z"/><path fill="#34A853" d="M12 23c2.97 0 5.46-.98 7.28-2.66l-3.57-2.77c-.98.66-2.23 1.06-3.71 1.06-2.86 0-5.29-1.93-6.16-4.53H2.18v2.84C3.99 20.53 7.7 23 12 23z"/><path fill="#FBBC04" d="M5.84 14.09c-.22-.66-.35-1.36-.35-2.09s.13-1.43.35-2.09V7.07H2.18C1.43 8.55 1 10.22 1 12s.43 3.45 1.18 4.93l2.85-2.22.81-.62z"/><path fill="#EA4335" d="M12 5.38c1.62 0 3.06.56 4.21 1.64l3.15-3.15C17.45 2.09 14.97 1 12 1 7.7 1 3.99 3.47 2.18 7.07l3.66 2.84c.87-2.6 3.3-4.53 6.16-4.53z"/></symbol>
<symbol id="lg-tiktok" viewBox="0 0 24 24"><path fill="#fff" d="M12.525.02c1.31-.02 2.61-.01 3.91-.02.08 1.53.63 3.09 1.75 4.17 1.12 1.11 2.7 1.62 4.24 1.79v4.03c-1.44-.05-2.89-.35-4.2-.97-.57-.26-1.1-.59-1.62-.93-.01 2.92.01 5.84-.02 8.75-.08 1.4-.54 2.79-1.35 3.94-1.31 1.92-3.58 3.17-5.91 3.21-1.43.08-2.86-.31-4.08-1.03-2.02-1.19-3.44-3.37-3.65-5.71-.02-.5-.03-1-.01-1.49.18-1.9 1.12-3.72 2.58-4.96 1.66-1.44 3.98-2.13 6.15-1.72.02 1.48-.04 2.96-.04 4.44-.99-.32-2.15-.23-3.02.37-.63.41-1.11 1.04-1.36 1.75-.21.51-.15 1.07-.14 1.61.24 1.64 1.82 3.02 3.5 2.87 1.12-.01 2.19-.66 2.77-1.61.19-.33.4-.67.41-1.06.1-1.79.06-3.57.07-5.36.01-4.03-.01-8.05.02-12.07z"/></symbol>
<symbol id="lg-pinterest" viewBox="0 0 24 24"><path fill="#fff" d="M12.017 0C5.396 0 .029 5.367.029 11.987c0 5.079 3.158 9.417 7.618 11.162-.105-.949-.199-2.403.041-3.439.219-.937 1.406-5.957 1.406-5.957s-.359-.72-.359-1.781c0-1.663.967-2.911 2.168-2.911 1.024 0 1.518.769 1.518 1.688 0 1.029-.653 2.567-.992 3.992-.285 1.193.6 2.165 1.775 2.165 2.128 0 3.768-2.245 3.768-5.487 0-2.861-2.063-4.869-5.008-4.869-3.41 0-5.409 2.562-5.409 5.199 0 1.033.394 2.143.889 2.741.099.12.112.225.085.345-.09.375-.293 1.199-.334 1.363-.053.225-.172.271-.401.165-1.495-.69-2.433-2.878-2.433-4.646 0-3.776 2.748-7.252 7.92-7.252 4.158 0 7.392 2.967 7.392 6.923 0 4.135-2.607 7.462-6.233 7.462-1.214 0-2.354-.629-2.758-1.379l-.749 2.848c-.269 1.045-1.004 2.352-1.498 3.146 1.123.345 2.306.535 3.55.535 6.607 0 11.985-5.365 11.985-11.987C23.97 5.39 18.592.026 11.985.026L12.017 0z"/></symbol>
<symbol id="lg-snapchat" viewBox="0 0 24 24"><path fill="#000" d="M12.206.793c.99 0 4.347.276 5.93 3.821.529 1.193.403 3.219.299 4.847l-.003.06c-.012.18-.022.345-.03.51.075.045.203.09.401.09.3-.016.659-.12 1.033-.301.165-.088.344-.104.464-.104.182 0 .359.029.509.09.45.149.734.479.734.838.015.449-.39.839-1.213 1.168-.089.029-.209.075-.344.119-.45.135-1.139.36-1.333.81-.09.224-.061.524.12.868l.015.015c.06.136 1.526 3.475 4.791 4.014.255.044.435.27.42.509 0 .075-.015.149-.045.225-.24.569-1.273.988-3.146 1.271-.059.091-.12.375-.164.57-.029.179-.074.36-.134.553-.076.271-.27.405-.555.405h-.03c-.135 0-.313-.031-.538-.074-.36-.075-.765-.135-1.273-.135-.3 0-.599.015-.913.074-.6.104-1.123.464-1.723.884-.853.599-1.826 1.288-3.294 1.288-.06 0-.119-.015-.18-.015h-.149c-1.468 0-2.427-.675-3.279-1.288-.599-.42-1.107-.779-1.707-.884-.314-.045-.629-.074-.928-.074-.54 0-.958.089-1.272.149-.211.043-.391.074-.54.074-.374 0-.523-.224-.583-.42-.061-.192-.09-.389-.135-.567-.046-.181-.105-.494-.166-.57-1.918-.222-2.95-.642-3.189-1.226-.031-.063-.052-.15-.055-.225-.015-.243.165-.465.42-.509 3.264-.54 4.73-3.879 4.791-4.02l.016-.029c.18-.345.224-.645.119-.869-.195-.434-.884-.658-1.332-.809-.121-.029-.24-.074-.346-.119-1.107-.435-1.257-.93-1.197-1.273.09-.479.674-.793 1.168-.793.146 0 .27.029.383.074.42.194.789.3 1.104.3.234 0 .384-.06.465-.105l-.046-.569c-.098-1.626-.225-3.651.307-4.837C7.392 1.077 10.739.807 11.727.807l.419-.015h.06z"/></symbol>
<symbol id="lg-youtube" viewBox="0 0 24 24"><path fill="#fff" d="M23.498 6.186a3.016 3.016 0 0 0-2.122-2.136C19.505 3.545 12 3.545 12 3.545s-7.505 0-9.377.505A3.017 3.017 0 0 0 .502 6.186C0 8.07 0 12 0 12s0 3.93.502 5.814a3.016 3.016 0 0 0 2.122 2.136c1.871.505 9.376.505 9.376.505s7.505 0 9.377-.505a3.015 3.015 0 0 0 2.122-2.136C24 15.93 24 12 24 12s0-3.93-.502-5.814zM9.545 15.568V8.432L15.818 12l-6.273 3.568z"/></symbol>
<symbol id="lg-reddit" viewBox="0 0 24 24"><path fill="#fff" d="M12 0C5.373 0 0 5.373 0 12c0 3.314 1.343 6.314 3.515 8.485l-2.286 2.286C.775 23.225 1.097 24 1.738 24H12c6.627 0 12-5.373 12-12S18.627 0 12 0Zm4.388 3.199c1.104 0 1.999.895 1.999 1.999 0 1.105-.895 2-1.999 2-.946 0-1.739-.657-1.947-1.539v.002c-1.147.162-2.032 1.15-2.032 2.341v.007c1.776.067 3.4.567 4.686 1.363.473-.363 1.064-.58 1.707-.58 1.547 0 2.802 1.254 2.802 2.802 0 1.117-.655 2.081-1.601 2.531-.088 3.256-3.637 5.876-7.997 5.876-4.361 0-7.905-2.617-7.998-5.87-.954-.447-1.614-1.415-1.614-2.538 0-1.548 1.255-2.802 2.803-2.802.645 0 1.239.218 1.712.585 1.275-.79 2.881-1.291 4.64-1.365v-.01c0-1.663 1.263-3.034 2.88-3.207.188-.911.993-1.595 1.959-1.595Zm-8.085 8.376c-.784 0-1.459.78-1.506 1.797-.047 1.016.64 1.429 1.426 1.429.786 0 1.371-.369 1.418-1.385.047-1.017-.553-1.841-1.338-1.841Zm7.406 0c-.786 0-1.385.824-1.338 1.841.047 1.017.634 1.385 1.418 1.385.785 0 1.473-.413 1.426-1.429-.046-1.017-.721-1.797-1.506-1.797Zm-3.703 4.013c-.974 0-1.907.048-2.77.135-.147.015-.241.168-.183.305.483 1.154 1.622 1.964 2.953 1.964 1.33 0 2.47-.81 2.953-1.964.057-.137-.037-.29-.184-.305-.863-.087-1.795-.135-2.769-.135Z"/></symbol>
<symbol id="lg-dv360" viewBox="0 0 24 24"><circle cx="12" cy="12" r="9" fill="#fff"/><path fill="#34A853" d="M12 5l5 7-5 7-5-7z" opacity=".9"/><circle cx="12" cy="12" r="3" fill="#34A853"/></symbol>
<symbol id="lg-klaviyo" viewBox="0 0 24 24"><path fill="#fff" d="M3 6l9 12 9-12-9 7-9-7zm9 8l-7-5h14l-7 5z"/></symbol>
<symbol id="lg-linkedin" viewBox="0 0 24 24"><path fill="#fff" d="M5 3a2 2 0 100 4 2 2 0 000-4zm-2 6h4v12H3V9zm6 0h4v2c.6-1 2-2 4-2 3 0 4 2 4 5v7h-4v-6c0-1.5-.5-3-2-3s-2 1.5-2 3v6H9V9z"/></symbol>
<symbol id="lg-bing" viewBox="0 0 24 24"><path fill="#fff" d="M5 2v17l5-2 6-3-5-2-2-5-4-5z"/></symbol>
</defs>
</svg>
<div class="app">
<header class="topbar">
