Pulsar · Metrics playbook · No. 13 · Ad set structure & budget. Grounded in a real live campaign — “Bible Cover”, lifetime to Jul 2026 (EGP).
Too Many Ad Sets
How you split campaigns and budget decides whether any ad ever learns — and most accounts split far too much.
Companion to No. 01–12 — the direct sequel to No. 12 · The Learning Phase: that deck set the ~50-a-week exit bar; this one shows how account structure decides whether any ad set can ever reach it. Leans on No. 01 · Efficiency & Fatigue (self-competition inflates your own CPM), No. 05 · Who Made the Sale (which campaign actually earned) and No. 02 · Quantity vs Quality (which event to optimize on).
Contents
01 Structure = budget · 02 The three levels · 03 Fragmentation · 04 Why splitting hurts · 05 CBO vs ABO · 06 The consolidation fix · 07 The budget math
Structure is a budget decision
How you split your account into campaigns and ad sets isn’t org-chart tidiness. It decides how thin your budget — and your conversion signal — get spread, and therefore whether anything ever learns (No. 12 · The Learning Phase).
Every ad set you add divides your budget — and your learning signal — by one more.
No. 12 established the bar: an ad set needs ~50 conversions a week to exit learning. That bar is per ad set — not per account. So the moment you draw the boxes, you’ve already decided how far each box will be from the bar. The rest of this deck walks the hierarchy (section 02), shows what over-splitting did to this real account (section 03), the three mechanics of why it hurts (section 04), who should hold the budget (section 05), the consolidation fix (section 06), and the honest budget math (section 07).
The three levels — the hierarchy
Meta’s account structure has exactly three levels. Beginners obsess over the ad and ignore the ad set — but budget and learning live one level up.
You tune the ad. Meta learns at the ad set. Budget decides both.
Fragmentation, in your own account — two readings of the same 30 days
Here’s a real live campaign done right: “Bible Cover” — one campaign, one ad set, one creative, on a single 225 EGP/day CBO budget. Because the money is concentrated, that one ad set gets ~1,575 EGP a week — enough to clear Meta’s learning bar on its optimization event: ~92 Add-to-Carts a week against a bar of ~50 (No. 12). Lifetime it has bought 300 Add-to-Carts at ~17 EGP and ~115 orders at ~45 EGP on the store’s books. Now imagine the same 225/day split across a dozen ad sets to “test more”: each slice drops to a few EGP a day, none ever reaches the bar, and every result is noise. Structure is budget — splitting a small budget too many ways is the most common way to guarantee nothing learns. The tools below let you split this real budget yourself and watch it happen.
One budget, two shapes — spread thin vs concentrated
An illustration, not a real account. Take Bible Cover’s real budget — 225 EGP/day ≈ 6,750/month — and picture it spread across ~40 ad sets, then concentrated into a few. Same money, two shapes; only one can learn.
Interactive in the HTML version — a toggle regroups one budget (Bible Cover’s 6,750 EGP/month) between a spread of ~40 ad sets (≈ 5 EGP a day each — nothing learns) and a concentrated 3-ad-set shape (≈ 2,250 EGP each). An illustration; the real campaign runs a single concentrated ad set. The printed state is the spread.
Why splitting hurts — three mechanics
Over-splitting isn’t just untidy — it actively damages delivery through three separate mechanisms, and this account is paying all three at once.
The budget splitter
Defaults are Bible Cover’s real budget — 225 EGP/day (6,750/month) — but split 6 ways instead of its real one. Watch it fall under the bar; drag the ad-set slider back toward 1 and it clears at ~92 Add-to-Carts a week, exactly as the real campaign does.
Interactive in the HTML version — two sliders (monthly budget, number of ad sets). The default prints above: 6,750 EGP / month ≈ 225 / day, split across 6 ad sets ≈ 37 EGP per ad set per day — ~15 Add-to-Carts per ad set per week at ~17 EGP each, against a bar of 50. Verdict: Learning Limited. Drag to 1 ad set (the real Bible Cover shape) and it projects ~92 a week — clears the bar.
CBO vs ABO — who holds the budget
The hierarchy in section 02 left one open question: which level holds the money. Meta gives you two answers, and for a small account one of them is clearly right.
| Mode | Who decides the split | Best when | On this account |
|---|---|---|---|
| ABO · ad set budget | You — a fixed budget on each ad set. Maximum control. | You have enough volume in each ad set to compare fairly — structured tests on real budgets. | TOO THIN splitting 176 EGP/day into dozens of hand-set slices would be dozens of starved cells |
| CBO · Advantage campaign budget | Meta — one budget on the campaign, flowed to whichever ad set performs. | Small or uneven budgets — consolidation by default; the algorithm concentrates instead of you pre-splitting. | THE RIGHT CALL let the money find the 2 creatives that convert |
Rule of thumb for a small account: fewer ad sets + CBO + a broad / Advantage+ audience — and let Meta concentrate the money on what works, instead of you pre-splitting it dozens of ways. Control is a luxury of volume; at 176 EGP a day, consolidation is the control.
The consolidation fix — six moves
The way out of section 03’s sprawl isn’t better creatives — it’s fewer, better-fed boxes. Six moves, in order:
Keep the sprawl paused
An archive of 43 old campaigns is fine — the sin is funding them all at once (or judging the account as if you had). Every campaign you actually fund must justify its slice of 176 EGP a day; split that across the whole archive and each box would get ≈ 4 EGP a day — a rounding error, not a budget.
Put 3–5 creatives in ONE ad set
Let Meta pick the winner — don’t give each creative its own campaign. Creative rotation inside one ad set shares the learning signal; 48 separate boxes share nothing.
Concentrate budget
At least one ad set must be able to approach the learning bar (No. 12). One funded ad set beats forty-three starved ones — the splitter in section 04 is the proof.
Go broad
One wide / Advantage+ audience beats many narrow overlapping ones. Broad kills the self-competition of section 04 — your ad sets stop bidding against each other in the same auction (No. 01).
Use CBO
Budget flows to the winner automatically (section 05). You stop guessing the split; Meta reallocates daily toward whatever is actually converting.
Test deliberately
Few variables, enough budget per cell to read the result. A test that can’t reach significance isn’t a test — it’s a pile of starved cells telling you nothing (section 04).
The budget math — tie it together
Be honest about what one ad set needs: ~50 conversions a week to exit learning, at roughly 200 EGP per purchase, is about 10,000 EGP a week for a single ad set optimizing on Purchase — usually out of reach for a small business.
Build only as many ad sets as your budget can actually feed.
| Thread | What structure decides | Where the full read lives |
|---|---|---|
| Learning | SIGNAL ÷ AD SETS whether any ad set can reach the 50-a-week bar | No. 12 · The Learning Phase |
| Self-competition | YOUR OWN CPM overlapping ad sets bidding in the same auction | No. 01 · Efficiency & Fatigue |
| Attribution | WHO EARNED IT which campaign the ~10 real sales actually came from | No. 05 · Who Made the Sale |
| Event choice | WHAT TO LEARN ON the intent-vs-volume trade-off of the optimization event | No. 02 · Quantity vs Quality |
Where Pulsar fits, in one honest line: Pulsar shows real spend and real conversions per creative and per campaign, so you can see the fragmentation — and where the money actually worked (No. 05 · Who Made the Sale) — not just how many boxes you created.