If your Meta account has dozens of ad sets and nothing scales, the problem is usually structure, not budget or creative. Consolidate into a few campaigns organized by funnel layer, give each ad set enough conversions to exit the learning phase, and fix the conversion signal before you touch targeting. Split only when a campaign genuinely needs its own budget, optimization event or success metric.
Why simpler structures usually win
Meta’s delivery system learns per ad set. Meta’s long-standing guidance is that an ad set exits the learning phase after roughly 50 optimization events within seven days of its last significant edit. Until then, results are less stable and usually more expensive. Every ad set you add divides the same conversions into smaller piles.
A worked example with made-up round numbers:
- Monthly spend: $40,000
- Cost per purchase: $80
- Purchases per month: 500, or about 115 a week
- Ad sets: 40, so roughly 3 purchases per ad set per week
None of those ad sets will ever exit learning. With the same spend in two ad sets, each would get close to 60 purchases a week.
Fragmentation causes quieter problems too:
- Overlapping ad sets cannibalize each other. When several of your ad sets qualify for the same auction, Meta enters only one, so the rest underdeliver.
- Every read is noise. Three purchases per ad set per week can’t tell you which audience works.
- Manual targeting isn’t the lever it was. With a clean conversion signal, broad and Advantage+ audiences find buyers on their own, and the creative does most of the targeting.
| Symptom | What it usually means |
|---|---|
| Many ad sets marked “Learning limited” | Too few conversions per ad set |
| Budget increases spike CPA and it never settles | Small ad sets restarting learning |
| High audience overlap between ad sets | Ad sets cannibalizing each other’s delivery |
| The “winning” ad set changes every week | Sample sizes too small to trust |
| One ad set per interest, age band or placement | Manual segmentation Meta already does itself |
When I audit accounts that “stopped scaling,” this pattern usually shows up before any creative or bidding problem.
Prospecting, retargeting and retention layers
Organize campaigns around who you’re reaching and why, not how you’re targeting them. Three layers cover most ecommerce and lead generation accounts.
| Layer | Who it reaches | Job | Budget share (starting point, adjust to your data) |
|---|---|---|---|
| Prospecting | People who haven’t engaged with you | Find new customers | The large majority, often 70-90% |
| Retargeting | Recent site visitors, add-to-carts, video viewers, social engagers | Close people already considering you | Often 5-20%, capped by audience size |
| Retention | Existing customers | Repeat purchase, launches, cross-sell | Small, and only if email and SMS can’t do the job |
One nuance: broad prospecting doesn’t exclude warm users unless you tell it to, and automated campaigns happily spend on past visitors and existing customers because they convert cheaply. That flatters prospecting results.
Define your existing customers and engaged audiences as audience segments in your ad account settings, so reporting can split results into new, engaged and existing customers. If a large share of “prospecting” revenue comes from existing customers, you’re paying for sales that email might have delivered anyway.
Advantage+ vs manual campaigns
Meta keeps folding its automation into standard campaign setup: a sales campaign with automated audience, placements and budget is treated as Advantage+, the successor to Advantage+ shopping campaigns. The names change often; the trade-off doesn’t.
| Advantage+ (automated) | Manual | |
|---|---|---|
| Audience | Meta decides; your inputs are mostly suggestions | You set inclusions, exclusions and hard limits |
| Best for | Ecommerce with steady purchase volume and a clean signal | Isolated tests, hard exclusions, niche or regulated audiences, B2B lead gen |
| Main risk | Spend drifts to existing customers and warm users | Fragmentation and over-narrow targeting |
My default for ecommerce is one Advantage+ campaign as the core prospecting engine, plus manual campaigns only where I need control: a testing campaign, or a new-customer campaign with purchasers hard-excluded.
Judge them on blended numbers and new-customer revenue, not in-platform ROAS, because every campaign claims credit generously. The reasons platform numbers disagree with your analytics are covered in why GA4 and ad platforms never match.
For B2B lead gen, manual control matters more. Broad targeting optimized to a raw form fill will find people who fill in forms, not buyers.
Budget allocation: CBO vs ABO
Campaign budget optimization (CBO, since renamed under Meta’s Advantage branding) sets one budget per campaign and shifts it between ad sets toward the cheapest results. Ad set budgets (ABO) fix spend for each ad set.
| Use CBO when | Use ABO when |
|---|---|
| Ad sets share an optimization event and similar audience sizes | Something small needs guaranteed spend, like retargeting or a new market |
| You’re scaling a proven layer | You’re testing and each variant needs a fair budget |
| There’s enough conversion volume for Meta to compare ad sets | Ad sets differ so much that “cheapest result” would misallocate |
The most common mistake is mixing cold and warm audiences in one CBO campaign. Meta chases the cheapest result right now, which can mean saturating a small retargeting pool or starving it completely. Keep each layer in its own campaign.
If CBO pours most of the budget into one ad set, that’s usually a signal to merge, not a bug to fight with ad set spend limits.
Large budget changes can send an ad set back into learning. I raise budgets in steps, commonly around 20-30% every few days on a stable ad set. I don’t duplicate ad sets to scale; that recreates the overlap problem.
Signal quality: conversion events and CAPI
Structure decides how conversions are grouped. Signal quality decides whether Meta learns from the right ones.
Choose the optimization event
Optimize for the deepest event that reaches about 50 per ad set per week. Purchase if you can. If you can’t, consolidate further before moving up the funnel, because optimizing for add-to-cart buys add-to-carts.
For lead gen, send qualified stages from your CRM, such as sales-accepted leads or booked meetings, back to Meta as conversion events. Optimize to them once volume allows.
Set up the Conversions API properly
The browser pixel loses events to ad blockers, browser privacy features and consent choices. The Conversions API (CAPI) sends events from your server. Run both and deduplicate. On Shopify, the native Meta integration can send server events; check its data-sharing level.
- Pixel and CAPI send the same key events with matching event names and a shared event ID, so Meta deduplicates them
- Purchase events carry the real order value and currency
- Server events include customer information parameters: hashed email and phone, external ID, fbp and fbc cookie values, IP address and user agent
- Event Match Quality in Events Manager is reviewed per event
- Events fire only where your consent setup allows
- Test events are checked after every site or checkout change
- Diagnostics are clear of duplication and missing-parameter warnings
Event Match Quality is a 0-10 score for how well the customer information on your server events can be matched to Meta accounts. Better matching means more conversions attributed and more data for delivery to learn from.
When to split campaigns
Use one test: split only when the new campaign needs a different budget, optimization event or success metric. If the answer is no, add ads to an existing ad set instead.
| Justified split | Not a reason to split |
|---|---|
| Different optimization event (purchase vs qualified lead) | Interest groups |
| Markets with separate budgets, currencies or languages | Age and gender bands |
| Product lines with very different margin or order value | Placements |
| A special ad category, such as housing, employment or financial products | One ad set per creative |
| An isolated test with a fixed budget and end date | Duplicating a winner to scale |
| A warm or customer pool that needs guaranteed spend | “The old campaign got tired” |
To consolidate a fragmented account without a crash, build the new structure with your best-performing ads, shift budget across over one to two weeks, and pause old ad sets only once the new ones are delivering. How to find and rotate new creative is its own process, covered in the ad creative testing framework.
Example structures by budget level
All three examples are for an ecommerce brand at a made-up $60 cost per purchase. Lead gen follows the same logic with qualified leads as the event.
| Monthly spend | Purchases per week | Campaigns | Ad sets |
|---|---|---|---|
| $5,000 | about 20 | 1 | 1 |
| $30,000 | about 115 | 3 | 3-4 |
| $150,000 | about 575 | 6-10 | 8-15 |
Around $5,000 a month
- One prospecting campaign, one ad set, broad or Advantage+ audience, optimized for purchase, with four to eight ads.
- No separate retargeting. Broad targeting already reaches warm visitors.
- Accept “Learning limited” rather than fragmenting further. Switch to initiate checkout only if purchases are too sparse, and judge it on purchases per dollar.
Around $30,000 a month
- Core prospecting: one Advantage+ campaign with most of the budget.
- Testing: one manual ABO campaign with a fixed slice, often 10-20%, for new concepts.
- Retargeting: one small ABO campaign for recent visitors and engagers, sized so frequency stays reasonable.
- Retention stays mostly in email and SMS.
Around $150,000 a month
- Split by market or business line only where budgets are managed separately.
- Within each: an Advantage+ core, a manual new-customer campaign with purchasers excluded, a testing campaign and a retargeting campaign.
- Add a retention campaign for launches and replenishment aimed at customer lists.
- Validate with incrementality tests, such as conversion lift studies or geo holdouts.
Even this runs fewer ad sets than the 40-ad-set account. This consolidation and signal work is where most of my performance marketing engagements start.
Get it built
If your account has more ad sets than its conversions can feed, I’ll restructure it, fix the signal and run it with you. Most engagements start with a fixed-price Growth Audit, credited if we continue. See pricing or get in touch.