swan-gtm Meta Ads Operating System - B2B SaaS
The single decision framework for running Meta ad accounts for B2B SaaS. Every decision - when to swap an ad, when to graduate it, when to scale budget, how many creatives to produce - flows from this system.
This file drives all operational decisions. Other knowledge-base files provide deeper context on specific topics. When in doubt, follow this OS.
Core Principle
Meta's algorithm is excellent at optimizing delivery (getting people to click and submit forms). But it cannot see lead quality. In B2B, a large percentage of form submissions come from people outside the ICP. The algorithm treats all leads as equal. Our job is to add the quality layer Meta cannot see, and make every decision based on qualified leads - not raw form fills.
1. Set the Target
Every formula depends on one number: TCPL (Target Cost Per Lead) = target cost per qualified lead. All thresholds are derived from TCPL. Without TCPL, you cannot run the Decision Tree, classify ads, or make scaling decisions. Establishing TCPL is always Step 1.
Scenario A: You have a target cost per demo (ideal)
TCPL = Target Cost per Demo x QL-to-Demo Rate
Example: Target cost per demo = EUR 2,000. QL-to-demo rate = 28%. TCPL = EUR 2,000 x 0.28 = EUR 560.
This is the strongest TCPL because it connects ad spend directly to a business outcome. Always pursue this number. If you know your target cost per demo but not the QL-to-demo rate, establishing that rate becomes the first measurement priority.
Scenario B: No target provided, but account has historical data
TCPL = 30-day trailing CPL(QL) x 0.80
Timeline: achieve TCPL within 30 days of engagement start
Logic: The 30-day trailing average represents current performance including waste. A 20% reduction is achievable through operational improvements (cutting zero-QL ads, graduating winners, improving creative mix) without structural changes like new audiences or conversion architecture. It is a realistic first target.
Example: 30-day trailing CPL(QL) = EUR 560. TCPL = EUR 560 x 0.80 = EUR 448. All thresholds derive from EUR 448.
Refinement: Once Scenario A data becomes available (QL-to-demo rate measured), replace the working TCPL with the business-derived TCPL:
TCPL = MIN(
30-day trailing CPL(QL) x 0.80, <- improvement target
Target Cost per Demo x QL-to-Demo Rate <- business target
)
Use whichever is tighter. If the business target is looser than the improvement target, it means the account is closer to healthy than the trailing average suggests - focus on maintaining rather than aggressive optimization.
Scenario C: New account, no historical data
TCPL = Target CAC x expected QL-to-Customer rate
OR
Industry benchmark CPL(QL) as a starting point
B2B SaaS benchmarks for qualified leads on Meta: EUR 300-800 depending on ACV and market. Use as a directional starting point, then switch to Scenario B after 30 days of data. Do not over-optimize against a benchmark TCPL - refine it with real data as quickly as possible.
2. Campaign Structure
2 Campaigns. Both CBO.
| Campaign | Budget | Purpose |
|---|---|---|
| Scaling Campaign (CBO) | ~80% of total budget | Proven/graduated ads only. CBO distributes freely among winners. |
| Testing Campaign (CBO) | ~20% of total budget | New concepts + iterations. Continuous feed of test creative. |
Why Two Campaigns
In a single CBO campaign, proven ads always outcompete test ads for budget. CBO is doing its job - allocating to what converts best. But this starves test ads of the spend they need to be evaluated. Minimum spend floors are a workaround that fights CBO's optimization logic.
Two campaigns solve this by giving testing its own hard budget. Test ads compete only against other test ads, not against proven winners. The testing budget is protected regardless of how well the Scaling Campaign is performing.
Scaling Campaign
- Contains only graduated/proven ads
- CBO distributes budget freely based on performance
- Ads enter only after passing all graduation criteria in the Testing Campaign
- This is where the bulk of qualified leads come from
Testing Campaign
- Contains new concepts and iterations being evaluated
- Fixed daily budget ensures tests always get spend
- Number of active test ads follows the Max Active Ads formula (see below)
- When a test is evaluated (pass or fail), rotate it out and add the next concept
- Winners graduate to the Scaling Campaign
Budget Split
Scaling Campaign daily budget = Total daily budget x 0.80
Testing Campaign daily budget = Total daily budget x 0.20
| Total Monthly Budget | Scaling (80%) | Testing (20%) |
|---|---|---|
| EUR 20K | EUR 16K | EUR 4K |
| EUR 30K | EUR 24K | EUR 6K |
| EUR 50K | EUR 40K | EUR 10K |
| EUR 100K | EUR 80K | EUR 20K |
Ad Count Guidelines
There is no "right" number of ads to run. The number depends on how many winners you have and how many tests you're running. A few strong winners can hold an entire budget. The key is the testing pipeline - finding and graduating winners - not hitting a specific ad count.
Sweet spot: 6-10 active ads. At this range, each ad gets enough daily spend to be properly evaluated. The actual count is: winners (however many you have) + 2-3 test slots.
Ceiling formula - the point where ads start getting starved:
Ceiling = (Daily budget x 14) / (2 x TCPL)
| Daily Budget | TCPL | Ceiling | Sweet Spot |
|---|---|---|---|
| EUR 300 | EUR 500 | 4 | 3-4 |
| EUR 500 | EUR 500 | 7 | 4-6 |
| EUR 1,000 | EUR 500 | 14 | 6-10 |
| EUR 2,000 | EUR 500 | 28 | 10-16 |
| EUR 3,000 | EUR 500 | 42 | 14-22 |
| EUR 1,000 | EUR 300 | 23 | 8-14 |
| EUR 1,000 | EUR 800 | 9 | 5-7 |
When at the ceiling: Queue new tests. Every new ad in means an old ad out. Don't launch more tests if the account is already at capacity - wait for a delivery kill or quality kill to free a slot.
Growing into it: A new account won't start at the ceiling. You grow into it as winners graduate. If you have 3 winners + 2 tests = 5 active ads, that's fine. The ceiling is where you stop adding, not where you need to be.
Ad Sets Within Each Campaign
Each campaign uses 1 ad set (or 2 if you need reporting separation within the campaign). Audience targeting is identical across both campaigns. The separation is about budget control, not audience segmentation.
Structure Rules
Why CBO for both: CBO distributes budget to the best-performing creative within each campaign. In the Scaling Campaign, this maximizes return on proven ads. In the Testing Campaign, this provides an early signal - if CBO deprioritizes a test ad, the creative isn't resonating (see Stage 1: Delivery Check in the Decision Tree).
Why not ABO for testing: ABO forces equal distribution, which wastes budget on creative that Meta's algorithm has already identified as low-performing. CBO's uneven distribution in the Testing Campaign IS useful data - it tells you which creative the algorithm can deliver efficiently.
Why not Advantage+ yet: Requires 50+ conversions/week/ad set. Most B2B accounts don't hit this on qualified leads. Stay Phase 2 until volume supports it.
The Three Phases
| Phase | Budget | Purpose | Transition Trigger |
|---|---|---|---|
| 1: Audience Validation | ABO, 1 Campaign | Test which audiences produce quality leads | Winning audience found (2-4 weeks) |
| 2: Creative Scaling | CBO, 2 Campaigns | Scale through creative testing on validated audience | 50+ conversions/week consistently |
| 3: Automated Scaling | Advantage+ | Max scale with full automation | Ongoing |
Most B2B accounts live in Phase 2. This OS is built for Phase 2.
3. The Decision Tree
The Decision Tree has two stages. Stage 1 uses CBO's delivery behavior as a fast pre-screen. Stage 2 evaluates lead quality and cost for ads that pass Stage 1.
Stage 1: Delivery Check
Before evaluating whether an ad converts, check whether Meta is actually spending on it. CBO decides within the first 48-72 hours which ads it wants to deliver based on predicted engagement (CTR, video views, dwell time). By Day 7, its preference is stable.
IMPORTANT: This check is only valid when the account is at or below the ceiling. If there are more ads active than the ceiling allows, low spend means the ad got crowded out - not that Meta rejected it. Fix the ad count first, then apply this check.
Two delivery kill triggers:
1. Day 7 check (new ads):
Calculate the minimum spend the ad should have after 7 days:
Min spend after 7 days = (Daily campaign budget / Number of active ads in that campaign) x 7 x 0.5
This is half of what the ad would get if budget were split equally. CBO never splits equally - it picks favorites - but an ad getting less than half its fair share has been actively deprioritized.
If an ad spent less than this threshold after 7 days, KILL.
If an ad has EUR 0 spend after 7+ days, Meta predicted poor engagement and chose not to deliver it at all. KILL immediately.
2. Ongoing delivery kill (any ad that has spent >= TCPL):
Both conditions must be true:
- Ad has spent >= TCPL lifetime. CBO gave it a real shot - enough budget to deliver roughly 1 expected QL worth of impressions.
- Last 7d average spend < EUR 10/day (7d total < EUR 70). After that initial delivery, CBO is now pulling budget away.
If an ad hasn't reached TCPL lifetime spend yet, you cannot call it a delivery kill. It may just be crowded out by other ads in the account. Low spend on an under-TCPL ad is not a signal - it's a lack of data.
Why the TCPL threshold matters: Without it, you'd kill ads that CBO never explored - especially in overcrowded accounts where budget is spread too thin. The TCPL threshold ensures the ad got a fair trial before you judge CBO's verdict. Below TCPL, the ad might perform well if it actually got budget.
Why this matters for testing velocity: Delivery kills free up budget faster than quality kills. An ad killed at TCPL spend (instead of 3x TCPL) returns ~2x TCPL to the testing pool, funding the next test. This is why actual testing throughput is higher than the base formula suggests.
What to do with delivery kills: The creative failed at the engagement level - the hook and/or visual don't stop the scroll. Swap with a different hook, visual treatment, or format. Don't iterate on the copy or CTA - the audience never got that far.
Stage 2: Quality Evaluation (ads that pass Stage 1)
Run every Monday for every active ad that passed Stage 1. Pull rolling 14-day data: spend, pixel leads, qualified leads, qualified lead rate, cost per qualified lead, frequency.
Step 1: Enough data?
- Spend < 3x TCPL: WAIT. Not enough data. Check next Monday.
- Spend >= 3x TCPL: Proceed to Step 2.
Logic: If an ad's true cost per QL equals target, you'd expect 3 qualified leads after 3x TCPL spend. Getting zero has a 5% probability (Poisson: e^-3 = 0.0498). At 2x, 13% false negative rate - too high. At 5x, 99% sure but wasted 2x TCPL extra. 3x balances 95% confidence with budget efficiency.
Step 2: Any pixel leads?
- Zero pixel leads: SWAP. Do NOT iterate on this concept. Creative doesn't resonate at all.
- Has pixel leads: Proceed to Step 3.
Step 3: Quality check (the QL layer)
This is the layer Meta cannot see. Campaign optimizes for pixel leads, but we measure qualified leads.
- Zero qualified leads despite pixel leads: SWAP. Keep the format, change the angle. Ad attracts wrong audience.
- Qualified rate < 40%: SWAP. Add ICP-qualifying language. More than 6/10 leads unqualified. True cost per QL is 2.5x+ pixel CPL.
- Qualified rate 40-60%: MONITOR. Borderline. One more week.
- Qualified rate >= 60%: Proceed to Step 4.
Logic (40%/60%): At 40% QL rate with EUR 360 pixel CPL, true cost per QL = EUR 900. At 65% rate, same pixel CPL = EUR 554. The low-quality ad costs 62% more per QL while looking identical in Ads Manager.
Step 4: Cost check
- Cost per QL <= TCPL: POTENTIAL WINNER. Proceed to Step 5.
- Cost per QL 1x-1.5x TCPL: MONITOR. Normal variance (10-20% week-to-week).
- Cost per QL > 1.5x TCPL: SWAP. Same angle, different hook/format/visual.
Logic (1.5x): Normal variance is 10-20%. An ad at 1.5x TCPL over 14 days is structurally underperforming, not experiencing bad luck. At 1.5x on EUR 5K/month, you get 7 QLs instead of 11. Four lost QLs per month from one ad.
Step 5: Graduation (Testing Campaign ads only)
ALL must be true to graduate from Testing Campaign to Scaling Campaign:
- Qualified leads >= 5 (minimum sample for stable rate)
- Qualified lead rate >= 60%
- Cost per QL <= TCPL
- Running >= 14 days (learning phase + stabilization)
- At least 1 QL in last 7 days (still active)
Step 6: Fatigue check (Scaling Campaign ads only)
- Frequency < 3.0 AND cost stable: HEALTHY. Keep running.
- Frequency 3.0-3.5 OR cost up 20%+: WARNING. Start producing 2 iterations NOW. Ready in 14 days.
- Frequency > 3.5 OR cost up 40%+ OR > 1.5x TCPL for 2 weeks: CRITICAL. Swap immediately with iteration.
Frequency logic: At 1.0-2.5 fresh impressions. At 2.5-3.5 diminishing returns, CPM rises. At 3.5+ the convertible audience is exhausted. Based on $25M+ spend data.
4. Swap Rules (Never Pause Without Replacing)
Every SWAP requires a replacement. Pausing without replacing = shrinking.
| Swap Reason | Iterate? | What to Change | Timeline |
|---|---|---|---|
| Delivery failure (Stage 1) | No. Abandon creative. | Different hook/visual/format. The creative doesn't stop the scroll. | Immediate |
| Zero pixel leads (Step 2) | No. Abandon concept. | Completely different angle/format/message. | 48 hours |
| Zero QLs (Step 3) | Partially. Keep format. | Same format, different pain point filtering for ICP. | 48 hours |
| Low quality < 40% (Step 3) | Yes. Add ICP language. | Add revenue figures, industry terms, ICP-specific language. | 48 hours |
| Cost > 1.5x TCPL (Step 4) | Yes. Change execution. | Same angle, different hook/format/visual/CTA. | 48 hours |
| Fatigue critical (Step 6) | Yes. Change surface. | New hook, different format, different customer, different color. | Immediate |
If pipeline is empty: redirect budget to existing Proven ads. Replacement must be live within 7 days max.
5. Creative Production Formula
Tests per month and per week
Tests/month = Testing Campaign monthly budget / (3 x TCPL)
Tests/week = Tests/month / 4
Or as a single formula from total budget:
Tests/week = (Monthly budget x 0.20) / (3 x TCPL) / 4
Logic: The Testing Campaign gets 20% of total budget. Each test needs 3x TCPL to complete Stage 2 quality evaluation. Tests that fail Stage 1 (delivery check) free up budget faster - they get swapped at Day 7, not after 3x TCPL spend.
| Total Monthly Budget | Testing Campaign (20%) | TCPL | Tests/Month | Tests/Week |
|---|---|---|---|---|
| EUR 10K | EUR 2K | EUR 500 | ~1 | 0.25 (1/month) |
| EUR 20K | EUR 4K | EUR 500 | ~3 | ~1 |
| EUR 30K | EUR 6K | EUR 500 | ~4 | ~1 |
| EUR 50K | EUR 10K | EUR 500 | ~7 | ~2 |
| EUR 100K | EUR 20K | EUR 500 | ~13 | ~3-4 |
| EUR 30K | EUR 6K | EUR 300 | ~7 | ~2 |
| EUR 30K | EUR 6K | EUR 800 | ~3 | ~1 |
Actual throughput is higher. Delivery kills (ads killed at EUR 100-300 instead of full EUR 1,530) free up budget for additional tests. If half the tests fail delivery early, actual throughput is roughly 1.5-2x the base formula. At EUR 30K/month, expect 6-8 tests/month (2/week) rather than the base 4.
Win rates
| Type | Win Rate | Tests for 1 Winner |
|---|---|---|
| Iterations on winners | ~25% (1 in 4) | 4 |
| New concepts | ~10% (1 in 10) | 10 |
| Blended (50/50 mix) | ~17% (1 in 6) | 6 |
Production split (50/30/20)
This is a creative production ratio - it defines what you build, not how budget is allocated. CBO handles budget distribution automatically based on performance.
- 50%: Iterations on top performers (same concept, different hook/format/visual)
- 30%: Iterations on other performers (same angle, different execution)
- 20%: Completely new concepts (different angle, format, message)
How 50/30/20 maps to the two-campaign structure:
- All iterations and new concepts launch in the Testing Campaign. Once graduated, they move to the Scaling Campaign.
- The 50/30/20 ratio guides what you produce, not where it lives or how budget is allocated. CBO in the Scaling Campaign distributes budget among proven ads based on performance.
Common mistake: Interpreting 50/30/20 as budget targets. The 80/20 budget split between Scaling and Testing Campaigns is about ensuring tests get evaluated. The 50/30/20 split is about what creative you build.
For each proven ad, produce 10 variations over its lifespan (message-validation.md).
Image-First Testing for New Concepts
When testing a new concept (the 20% of production that is completely new angles), lead with a static image ad before producing video or other formats. Images are cheaper and faster to produce, so you can validate whether the concept resonates before investing in higher-effort formats.
Process:
- Launch the new concept as a static image in the Testing Campaign
- Stage 1 (Delivery Check): Does CBO deliver it? If not, swap at Day 7 - no video production wasted.
- Stage 2 (Quality Evaluation): Run through the Decision Tree (Steps 1-4)
- If the concept passes: produce video, carousel, or AI video versions of the same concept
- If the concept fails: discard without wasting video production time
This does not apply to iterations on proven concepts (the 50% + 30%) where you already know the concept works and the format change is the variable being tested. It also does not apply to inherently video-native concepts (UGC, testimonials, product demos).
Proven ads needed by budget
Minimum proven ads = Monthly budget / EUR 5,000
Logic: Each proven ad absorbs ~EUR 5K/month before frequency causes fatigue. Above EUR 6-8K on one ad, performance degrades.
| Monthly Budget | Min Proven Ads | Tests Needed (17%) | Months to Build |
|---|---|---|---|
| EUR 20K | 4 | ~24 | ~6 |
| EUR 30K | 6 | ~36 | ~8 |
| EUR 50K | 10 | ~60 | ~8 |
| EUR 100K | 20 | ~120 | ~9-10 |
You cannot scale budget ahead of creative.
6. Scaling Protocol
When to scale (ALL must be true)
- Proven ad count in Scaling Campaign >= minimum for next budget level
- Account frequency < 3.0
- Cost per QL at or below TCPL for 2+ consecutive weeks
- Testing Campaign pipeline has 3+ ready-to-launch replacements
How to scale
Scale the Scaling Campaign budget. Testing Campaign budget scales proportionally (maintain 80/20 split).
- Increase: 20% every 5 days (under 30% learning phase reset)
- Maximum: never more than 30% at once
- Rollback trigger: cost > 1.5x TCPL post-scale
- Rollback rate: reduce 20-30% immediately
- Resume after rollback: 10% every 7 days after 2 weeks stable
When scaling hits a wall (frequency > 3.5)
Options: expand lookalike 1% to 2-3%, add new seed lists, test broad targeting, cross-channel retargeting (UTM), reactivate remarketing.
7. Weekly Cadence
| Day | Focus | Actions |
|---|---|---|
| Monday | Decision Day | Pull 14-day data. Run Stage 2 (Decision Tree) for Testing Campaign ads that passed Stage 1. Run Step 6 (fatigue) for Scaling Campaign ads. Execute SWAPs/GRADUATEs. Flag WARNINGs. Update pipeline. |
| Wednesday | Creative Launch + Stage 1 Check | Launch new tests in Testing Campaign. Check Stage 1 (Delivery Check) on ads that have been running 7+ days - swap underdelivering ads. |
| Friday | Budget + Scaling | Check scaling criteria for Scaling Campaign. Apply 20% increase if met. Check rollback triggers. |
| Monthly | Full Review | Creative library audit. Pipeline health across both campaigns. TCPL review. Frequency trend check. |
Quick Reference - All Formulas
| Formula | Calculation | Purpose |
|---|---|---|
| TCPL (Target Cost Per Lead) | Demo cost x QL-to-demo rate | Foundation for all formulas |
| Ad count ceiling | (Daily budget x 14) / (2 x TCPL) | Max ads before starvation. Sweet spot: 6-10 |
| Delivery check (Day 7) | (Daily campaign budget / active ads / 2) x 7 | Min spend after 7 days - below this, Meta rejected the ad |
| Delivery kill (ongoing) | Spent >= TCPL lifetime AND 7d avg < EUR 10/day | CBO gave it a shot and pulled away |
| Min data threshold | 3 x TCPL | Min spend before judging an ad that IS getting spend (95% confidence) |
| Swap cost threshold | 1.5 x TCPL | Cost per QL above which ad is too expensive - pause it |
| Tests per week | (Monthly budget x 0.20) / (3 x TCPL) / 4 | How many new ads to launch per week |
| Min proven ads | Total budget / EUR 5,000 | Proven ads needed to support budget |
| Tests for X winners | X / 0.17 | Tests to plan for finding X winners |
| Budget split | 80% Scaling / 20% Testing | Hard budget control between campaigns |
| Scale rate | 20% every 5 days (max 30%) | Budget increase pace |
| Rollback trigger | Cost > 1.5x TCPL post-scale | When to cut budget |
Decision Flow (Plain Language)
Is the ad getting spend?
Ad has been active 7+ days and spent less than the delivery check threshold. Pause. Meta doesn't want to deliver this creative. Replace with a different hook/visual/format.
Ad has been active 7+ days and spent EUR 0. Pause. Meta rejected the creative entirely. Only valid if account is within Max Active Ads limit. If over the limit, the ad was crowded out - fix the ad count first.
Is the ad converting?
Ad spent more than 3x TCPL with 0 pixel leads. Pause. Creative doesn't resonate at all.
Ad spent more than 3x TCPL, has pixel leads but 0 qualified leads. Pause. Attracts wrong audience.
Ad spent more than 3x TCPL, qualified lead rate below 40%. Pause. Quality too low.
Ad spent more than 3x TCPL, cost per qualified lead above 1.5x TCPL. Pause. Too expensive.
Ad spent more than 3x TCPL, cost per qualified lead at or below TCPL, qualified lead rate above 60%. Potential winner. Check graduation criteria.
Ad spent less than 3x TCPL and IS getting spend. Wait. Not enough data yet. Check again next Monday.
Sources
| Element | Source |
|---|---|
| Two-campaign structure | Separates budget control from CBO optimization |
| Two-stage evaluation | CBO delivery signal + quality evaluation |
| 3x CPA rule | Pilothouse (3-3-3), Foxwell |
| 1.5x threshold | Optimization signals, maintenance rule |
| Frequency thresholds | $25M+ spend data |
| 50/30/20 allocation | Creative fatigue and creative strategy best practices |
| Win rates | Foxwell, Pilothouse ($50K-500K/month) |
| 20% testing budget | Foxwell, AdManage |
| Phase framework | Campaign structure and Advantage+ progression |
| Poisson distribution | Statistical probability (clinical trials standard) |
| Max active ads formula | Derived from 14-day evaluation window + 3x TCPL per ad |
| 14-day evaluation window | B2B standard, Foxwell, multiple B2B agency sources |
| 7-day delivery check | Meta CBO explores ads in 48-72 hours, preference stable by Day 7 |
| Half fair share threshold | Ad getting less than 50% of equal split has been deprioritized by CBO |
| EUR 0 rule | Valid only when Max Active Ads formula is followed |
Related Files (Deeper Context)
- campaign-structure.md - detailed Phase 1/2/3 architecture, naming conventions, campaign types
- creative-fatigue-detection.md - full fatigue detection workflow, rotation system, pipeline management
- creative-strategy.md - creative concepts, copy formulas, placement optimization
- advantage-plus.md - when and how to use Advantage+, setup details
- optimization-playbook.md - full decision trees, benchmarks, seasonal patterns
- message-validation.md - how to score ad quality against revenue, winner scaling
- audience-strategy.md - data hierarchy, lookalikes, third-party sources
- lead-form-optimization.md - lead form setup, work email validation, friction management