Orchestration (Sumeru)
The 4 capabilities
One auth per platform, unified billing view. Google + Meta + TikTok in one console.
AI-drafted multi-platform plans — audiences, creatives, bids. Approve once, push to all.
Hourly rebalancing within approved guardrails. Underperformers throttled; winners boosted.
Conversational diagnostics + recommendations — "why did X happen, what should I do?"
What this typically unlocks
| Outcome | Typical lift |
|---|---|
| Cross-platform ROAS | +25% at flat spend |
| Hours/week on ad operations | −15h vs. multi-tool stack |
| Time from idea → live campaign | 2 hours vs. 2 days |
| Budget reallocation cadence | daily vs. weekly |
| Ad-cost surprises (overspend) | near 0 — budget gates |
| Agency dependency | optional — copilot handles 80% |
How it powers every decision
| Decision | What orchestration unlocks |
|---|---|
| "Should we shift budget to Meta?" | Real cross-platform ROAS comparison + auto-rebalance suggestion |
| "Which creative wins?" | Multivariate test across platforms with unified attribution |
| "Are we wasting on cannibalizing audiences?" | Holdout-based incrementality on ad spend |
| "What's our weekly ad budget?" | Budget gates that prevent overspend |
| "Why did Q2 ROAS drop?" | Copilot diagnoses with platform-by-platform breakdown |
| "How should I bid on this audience?" | LTV-aware bid caps from Customer 360 + predictive LTV |
How it works
The unified plane
Once your ad accounts are connected, you stop working in three separate consoles. Orchestration shows:
- Spend across all three platforms in one chart
- Conversions + revenue attributed unified via Customer 360 — no double-counting
- Audience overlap visualized so you don't pay two platforms to reach the same person
AI-drafted plans
Set the goal ("$50K spend next month, target ROAS 4×, focus on new customer acquisition + win-back"). The campaign-plan copilot drafts a multi-platform allocation. You review, approve, and one click pushes to Google + Meta + TikTok with consistent UTM tagging.
Continuous optimization
The budget optimizer reads ROAS hourly across platforms and rebalances spend within your approved guardrails.
Copilot
Ask "why did Meta ROAS drop this week?" → get a structured answer: "Audience X saturated; CPM up 22%; CTR down 18%; recommend shifting $8K/week to TikTok."
Real merchant scenarios
Scenario A — DTC brand replaces $4K/month agency
$3M brand, agency at $4K/month for Google + Meta. 90-day result:
| Metric | Pre (agency) | Post (orchestration) |
|---|---|---|
| Total ad spend | $48K | $48K |
| Attributed revenue | $192K | $241K |
| ROAS | 4.0× | 5.0× |
| Hours/week brand-side | 4 | 2 |
| Agency cost | $4K | $0 |
Net gain: $49K more revenue + $4K saved + 2h/week back. Annualized ~$640K.
Scenario B — Mid-market scales spend 3× without losing efficiency
$15M brand, $40K/mo → $120K/mo. 90-day result:
| Month | Spend | Revenue | ROAS |
|---|---|---|---|
| Pre (m0) | $40K | $200K | 5.0× |
| m1 | $80K | $360K | 4.5× |
| m2 | $100K | $470K | 4.7× |
| m3 | $120K | $580K | 4.8× |
ROAS dipped slightly during scale but stabilized — orchestration diversified into TikTok new-customer audiences which had headroom Meta didn't.
Scenario C — Agency manages 12 brands from one console
Previously 36 ad accounts in 36 consoles. With orchestration: single agency-tier dashboard with per-brand ROAS, cross-brand learnings, automated weekly reports. Agency took on 4 more clients with the same headcount.
Best practices
✅ Connect all platforms even if you're not active on all — cross-platform attribution is much more valuable with complete data.
✅ Use the copilot for weekly reviews, not as a dashboard substitute.
✅ Set budget gates before scaling — prevent the "card billed $10K overnight" moment.
❌ Don't let the optimizer run uncontested for the first month — set guardrails (max ±20% reallocation/day).
❌ Don't ignore audience overlap warnings — paying twice for the same person.
Plan tiers
| Capability | Free | Starter | Pro | Agency | Enterprise |
|---|---|---|---|---|---|
| Connect 1 ad platform | — | — | ✓ | ✓ | ✓ |
| Connect 3 ad platforms | — | — | ✓ | ✓ | ✓ |
| AI campaign plans | — | — | ✓ | ✓ | ✓ |
| Budget optimizer | — | — | ✓ | ✓ | ✓ |
| Copilot | — | — | ✓ | ✓ | ✓ |
| Cross-platform attribution | — | — | ✓ | ✓ | ✓ |
| Audience overlap detection | — | — | ✓ | ✓ | ✓ |
| Holdout-based incrementality on ads | — | — | — | ✓ | ✓ |
| Multi-brand orchestration | — | — | — | ✓ | ✓ |
| Custom optimizer rules | — | — | — | — | ✓ |
See also
- Ad account management
- Sales engine — Customer 360 + predictive LTV feed audience targeting
- Attribution & revenue — multi-touch attribution underpins ROAS
Why this exists — the long version
Paid ads are the largest controllable line item in most e-commerce P&Ls — and the most fragmented. Google has its own console, its own bid logic, its own pixel. Meta has its own. TikTok has its own. Each platform optimizes for itself, not for your business outcome. Your agency (or in-house ad lead) spends 60% of their time stitching dashboards together, and the strategic question — "given my budget, what's the best allocation across platforms, audiences, and creatives this week?" — never gets a clean answer.
Orchestration (Sumeru) is the answer. It connects all three platforms under one unified plane, plans campaigns across them with budget-aware AI, optimizes spend continuously based on your outcome (not platform-internal metrics), and gives you a copilot that turns ambiguous "what should I do?" questions into specific "approve this plan" actions. The result is that paid ad spend becomes predictable, defensible, and ~25% more efficient at the same budget — without hiring an agency or adding a full-time ad ops headcount.
The deeper shift: ads stop being a tactical spend that you re-fight every Monday and become a strategic lever you set quarterly and let optimize. Marketing leaders get their time back; CFOs get a number they can stand behind.