Campaign structure & bidding basics
Two decisions shape results more than any headline or targeting tweak: how many campaigns and ad sets you run, and whether you or the algorithm sets the bid.
Both platforms have moved toward consolidation. On Meta, Advantage+ sales campaigns (the AI campaign type, formerly Advantage+ Shopping) pull most audiences into one structure. On Google, Performance Max replaces ad groups with asset groups and runs alongside a lean Standard Shopping setup.
AI can design a starter structure. The learning thresholds must come from each platform's own help page, never a guess.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are my paid-media architect. Platform: [Meta / Google]. Product: [product]. Monthly budget: [budget]. Expected monthly conversions: [number if known].
1. Open the platform's help page and confirm the current learning threshold (Meta: ~50 optimization events per ad set per rolling 7 days; Google: ~20-30 conversions/month per campaign). Do NOT invent it.
2. Design the MINIMAL starter structure: the fewest campaigns and ad sets/asset groups that match a real business split (funnel stage, market, hard budget), so my budget clears that threshold instead of spreading thin.
3. Recommend a starting bid strategy: lowest-cost / maximize-conversions first to gather data, moving to cost caps or target-ROAS only after ~30+ conversions/month with clean tracking.
Output ONE markdown table: Campaign | Ad set / asset group | Audience or theme | Bid strategy | Why. State the learning threshold and its source above the table.
Do not guess or estimate any number. The thresholds must come from the platform page; cite it. If you can't browse, leave them blank and name the help page to open.
Or do it in 5 steps
- Pick ONE platform and consolidate. On Meta, top accounts run 3-5 campaigns total, not per product. Start with 1-3 ad sets per campaign; add more only for a hard budget split, a genuinely different audience, or a reporting need. Merge audiences overlapping ~25%+.
- Match structure to conversion volume. Google PMax needs roughly 20-30 conversions/month per campaign. Under ~100 conversions/month, one consolidated campaign with 2-3 asset groups beats five thin ones that all stay stuck in learning and compete in the same auctions.
- Don't run PMax and Standard Shopping on the same products at once. Per Google, Performance Max takes priority over Standard Shopping in the same account, so it cannibalizes rather than adds; use one as the lead and keep the other tightly scoped.
- Start with lowest-cost bidding. Use lowest-cost / maximize-conversions first so the algorithm gathers a clean signal. Meta targets ~50 optimization events per ad set in a rolling 7-day window before delivery stabilizes; below that, don't just wait, consolidate audiences or broaden the optimization event.
- Add cost caps or target-ROAS later. Once you're at ~30+ conversions/month with clean tracking (server-side CAPI on Meta, verified conversions on Google), automated bidding with caps almost always beats tight manual bids. Below that floor, caps just choke delivery.
Done looks like: one platform, the fewest campaigns that clear the learning threshold, lowest-cost bidding running, and caps deferred until you have the data.
Worked example: a minimal Meta starter structure (illustrative)
| Campaign | Ad sets | Audience | Bid strategy |
|---|
| Prospecting | 1-2 | broad + one lookalike | lowest cost / maximize conversions |
| Retargeting | 1 | site visitors + cart abandoners | lowest cost |
| (later) Advantage+ sales | 1 | let Meta's AI allocate | maximize conversions, cost cap once 30+/mo |
Three campaigns, not thirty. The structure and thresholds are the honest part; confirm current numbers on Meta's Learning Phase help page and Google's automated-bidding help page before you build.
Revisit structure only when volume clearly outgrows it.