feat: sync Meta metrics, objetivos ppto/consecución y simulador por tenant

Rellena ad_daily desde Meta, KPIs con fallback a daily, objetivos de presupuesto vs consecución, y simulador sin TikTok anclado a métricas del cliente.

Co-authored-by: Cursor <cursoragent@cursor.com>

src/lib/simulate.ts

@@ -0,0 +1,143 @@
+/** Deterministic budget simulation from tenant ad_daily + kpi_weekly. */
+
+import { AdDailyRow, KpiWeeklyRow } from '@/lib/metrics';
+import { normalizePlatform } from '@/lib/ad-tracker-range';
+
+export type SimChannel = 'meta' | 'google' | 'blended';
+
+export type SimBaseline = {
+  historical_cpl: number;
+  conversion_rate_pct: number;
+  avg_ticket: number;
+  sample_spend: number;
+  sample_leads: number;
+  sample_deals: number;
+  sample_revenue: number;
+  channel: string;
+  source: 'tenant_metrics';
+};
+
+export type SimBand = {
+  leads: number;
+  deals: number;
+  revenue: number;
+  roas: number;
+};
+
+export type SimResult = {
+  target_channel: string;
+  baseline: SimBaseline;
+  conservative_p10: SimBand;
+  expected_p50: SimBand;
+  optimistic_p90: SimBand;
+};
+
+function n(v: unknown): number {
+  const x = Number(v);
+  return Number.isFinite(x) ? x : 0;
+}
+
+function matchChannel(platform: string | undefined, ch: SimChannel): boolean {
+  if (ch === 'blended') return true;
+  const p = normalizePlatform(platform);
+  if (ch === 'meta') return p === 'meta' || p === 'other';
+  if (ch === 'google') return p === 'google';
+  return false;
+}
+
+function band(
+  spend: number,
+  cpl: number,
+  convRate: number,
+  ticket: number,
+  cplMult: number,
+  convMult: number
+): SimBand {
+  const effectiveCpl = Math.max(1, cpl * cplMult);
+  const leads = Math.round(spend / effectiveCpl);
+  const deals = Math.round(leads * Math.min(0.95, convRate * convMult));
+  const revenue = deals * ticket;
+  const roas = spend > 0 ? Math.round((revenue / spend) * 100) / 100 : 0;
+  return { leads, deals, revenue: Math.round(revenue * 100) / 100, roas };
+}
+
+/** Build simulation from this tenant's warehouse rows. */
+export function simulateFromTenantMetrics(opts: {
+  spend: number;
+  channel: SimChannel;
+  daily: AdDailyRow[];
+  weekly: KpiWeeklyRow[];
+}): SimResult {
+  const { spend, channel, daily, weekly } = opts;
+  const dailyCh = daily.filter((r) => matchChannel(r.platform, channel));
+  const srcDaily = dailyCh.length ? dailyCh : daily;
+
+  const sampleSpend = srcDaily.reduce((a, r) => a + n(r.spend), 0);
+  const sampleLeads = srcDaily.reduce(
+    (a, r) => a + n(r.platform_leads ?? r.conversions),
+    0
+  );
+
+  const weeklyCh = weekly.filter((r) => {
+    if (channel === 'blended') return true;
+    const c = (r.channel || '').toLowerCase();
+    if (channel === 'meta') {
+      return (
+        c.includes('meta') ||
+        c.includes('facebook') ||
+        c.includes('instagram') ||
+        c.includes('fb')
+      );
+    }
+    if (channel === 'google') {
+      return c.includes('google') || c.includes('search') || c.includes('pmax');
+    }
+    return true;
+  });
+  const srcWeekly = weeklyCh.length ? weeklyCh : weekly;
+
+  const wSpend = srcWeekly.reduce((a, r) => a + n(r.total_spend), 0);
+  const wLeads = srcWeekly.reduce((a, r) => a + n(r.total_leads), 0);
+  const wDeals = srcWeekly.reduce((a, r) => a + n(r.cohort_deals_won), 0);
+  const wRevenue = srcWeekly.reduce((a, r) => a + n(r.cohort_revenue_won), 0);
+
+  const spendBase = sampleSpend > 0 ? sampleSpend : wSpend;
+  const leadsBase = sampleLeads > 0 ? sampleLeads : wLeads;
+  let historicalCpl =
+    leadsBase > 0
+      ? spendBase / leadsBase
+      : sampleSpend > 0
+        ? sampleSpend / Math.max(1, sampleLeads)
+        : 25;
+  if (!Number.isFinite(historicalCpl) || historicalCpl <= 0) historicalCpl = 25;
+
+  const convRate =
+    wLeads > 0
+      ? wDeals / wLeads
+      : wDeals > 0 && leadsBase > 0
+        ? wDeals / leadsBase
+        : 0.12;
+  const avgTicket =
+    wDeals > 0 ? wRevenue / wDeals : wRevenue > 0 ? wRevenue : 800;
+
+  const baseline: SimBaseline = {
+    historical_cpl: Math.round(historicalCpl * 100) / 100,
+    conversion_rate_pct: Math.round(Math.max(0.01, convRate) * 1000) / 10,
+    avg_ticket: Math.round(avgTicket * 100) / 100,
+    sample_spend: Math.round(spendBase * 100) / 100,
+    sample_leads: Math.round(leadsBase),
+    sample_deals: Math.round(wDeals),
+    sample_revenue: Math.round(wRevenue * 100) / 100,
+    channel: channel === 'blended' ? 'blended' : channel,
+    source: 'tenant_metrics',
+  };
+
+  const cr = Math.max(0.01, convRate);
+  return {
+    target_channel: baseline.channel,
+    baseline,
+    conservative_p10: band(spend, historicalCpl, cr, avgTicket, 1.35, 0.75),
+    expected_p50: band(spend, historicalCpl, cr, avgTicket, 1.0, 1.0),
+    optimistic_p90: band(spend, historicalCpl, cr, avgTicket, 0.75, 1.2),
+  };
+}