Is Your Meta ROAS Real? What Conversion Lift Reveals
Panto Source
Your Meta dashboard shows 500 conversions and 4x ROAS. But here's the question attribution can't answer: How many of those customers would have bought anyway — without ever seeing your ad?
Meta Conversion Lift tests answer that question.
Attribution tells you who converted after seeing your ad. Conversion Lift tells you who converted because of your ad. That's the difference between correlation and causation — and it's the difference between wasting budget and scaling profitably.
This guide walks through exactly how Meta Conversion Lift works, when to use it, how to set it up, and how to interpret results that actually inform budget decisions.
How Meta Conversion Lift Works
A Conversion Lift test is a randomized controlled experiment. Meta splits your target audience into two statistically identical groups:
Treatment group: Eligible to see your ads
Holdout (control) group: Not shown your ads
After the test period, Meta compares conversion rates between the groups. The difference represents your incremental lift — conversions that wouldn't have happened without advertising.
META CONVERSION LIFT:THE EXPERIMENT═══════════════════════════════════════════════════════════════════════YOUR TARGET AUDIENCE(100,000 users)│┌───────────┴───────────┐│ META RANDOMIZES │└───────────┬───────────┘│┌───────────┴───────────┐▼ ▼┌───────────────┐ ┌───────────────┐│ TREATMENT │ │ HOLDOUT ││ GROUP │ │ GROUP ││(85,000)│ │(15,000)││ │ │ ││ ✓ See ads │ │ ✗ No ads ││ │ │ ││ Conversions:│ │ Conversions:││ 850│ │ 120││(1.0% rate)│ │(0.8% rate)││ │ │ │└───────────────┘ └───────────────┘│┌───────────┴───────────┐│ LIFT CALCULATION │└───────────┬───────────┘│▼┌─────────────────────────────────────────────────────────────────────┐│ ││ Treatment conversion rate:1.0% ││ Holdout conversion rate:0.8% ││ ───────────────────────────────── ││ Incremental lift:25% ││ ││ MEANING:25% of conversions were CAUSED by your ads. ││ 75% would have happened anyway(organic demand). ││ │└─────────────────────────────────────────────────────────────────────┘
META CONVERSION LIFT:THE EXPERIMENT═══════════════════════════════════════════════════════════════════════YOUR TARGET AUDIENCE(100,000 users)│┌───────────┴───────────┐│ META RANDOMIZES │└───────────┬───────────┘│┌───────────┴───────────┐▼ ▼┌───────────────┐ ┌───────────────┐│ TREATMENT │ │ HOLDOUT ││ GROUP │ │ GROUP ││(85,000)│ │(15,000)││ │ │ ││ ✓ See ads │ │ ✗ No ads ││ │ │ ││ Conversions:│ │ Conversions:││ 850│ │ 120││(1.0% rate)│ │(0.8% rate)││ │ │ │└───────────────┘ └───────────────┘│┌───────────┴───────────┐│ LIFT CALCULATION │└───────────┬───────────┘│▼┌─────────────────────────────────────────────────────────────────────┐│ ││ Treatment conversion rate:1.0% ││ Holdout conversion rate:0.8% ││ ───────────────────────────────── ││ Incremental lift:25% ││ ││ MEANING:25% of conversions were CAUSED by your ads. ││ 75% would have happened anyway(organic demand). ││ │└─────────────────────────────────────────────────────────────────────┘
META CONVERSION LIFT:THE EXPERIMENT═══════════════════════════════════════════════════════════════════════YOUR TARGET AUDIENCE(100,000 users)│┌───────────┴───────────┐│ META RANDOMIZES │└───────────┬───────────┘│┌───────────┴───────────┐▼ ▼┌───────────────┐ ┌───────────────┐│ TREATMENT │ │ HOLDOUT ││ GROUP │ │ GROUP ││(85,000)│ │(15,000)││ │ │ ││ ✓ See ads │ │ ✗ No ads ││ │ │ ││ Conversions:│ │ Conversions:││ 850│ │ 120││(1.0% rate)│ │(0.8% rate)││ │ │ │└───────────────┘ └───────────────┘│┌───────────┴───────────┐│ LIFT CALCULATION │└───────────┬───────────┘│▼┌─────────────────────────────────────────────────────────────────────┐│ ││ Treatment conversion rate:1.0% ││ Holdout conversion rate:0.8% ││ ───────────────────────────────── ││ Incremental lift:25% ││ ││ MEANING:25% of conversions were CAUSED by your ads. ││ 75% would have happened anyway(organic demand). ││ │└─────────────────────────────────────────────────────────────────────┘
The Key Insight
If 75% of your "attributed" conversions would have happened without ads, your true ROAS is dramatically lower than what Meta reports. A 4x reported ROAS might be 1x incremental ROAS — meaning you're breaking even, not scaling profitably.
💡 The Core Concept
Attribution answers: "Who converted after seeing an ad?" Conversion Lift answers: "Who converted because of an ad?"
The difference between these questions is the difference between wasting money on customers who'd buy anyway and actually driving growth.
When to Run a Conversion Lift Test
Conversion Lift tests answer specific strategic questions. Run one when:
Question
Why Lift Testing Helps
"Should I scale or cut Meta spend?"
Proves whether Meta is driving real growth
"Is my retargeting truly incremental?"
Tests if you're paying for customers who'd convert anyway
"Manual campaigns vs. Advantage+?"
Compares true lift between strategies
"Are we over-crediting Meta?"
Validates (or invalidates) your attribution model
"What's our real Meta ROAS?"
Shows incremental return, not just attributed return
⚠️ When NOT to Run a Lift Test
Conversion volume is too low (need 100+ conversions in test period)
You're about to run major promotions or launch new products
Test period overlaps with Black Friday, holidays, or other anomalies
You can't maintain stable creative and budget during the test
Requirements and Prerequisites
Before setting up a Meta Conversion Lift test, confirm you have:
Technical Requirements
Requirement
Details
Meta Pixel
Installed and tracking conversions reliably
Conversions API (CAPI)
Recommended for improved match rates
Event Match Quality
EMQ score of 6.0+ for primary conversion events
Conversion volume
100+ conversions expected during test period
Account access
Admin or advertiser access to Meta Ads Manager
Test Design Requirements
Parameter
Recommended Setting
Holdout size
10-20% of audience (larger = stronger signal, more revenue loss)
Test duration
2-4 weeks minimum (depends on conversion volume)
Campaigns
Established campaigns with consistent performance
Budget stability
No major increases/decreases during test
Creative stability
No major refreshes during test
Signal Quality: The Hidden Test Killer
🚨 CRITICAL: Your Lift Test Is Only as Good as Your Signals
If your Pixel/CAPI is feeding low-quality events — bot traffic, unverified leads, or inflated micro-conversions — your lift test will measure the incrementality of junk, not real business outcomes.
Use the right conversion events:
Event Type
Use for Lift Testing?
Why
Purchase
✅ Yes
Measures actual revenue — the gold standard
Qualified Lead
✅ Yes
Verified leads tied to real pipeline
Add to Cart
⚠️ Cautiously
Can inflate lift without revenue correlation
Lead (unqualified)
❌ No
Often includes bots, spam, low-intent users
Link Click
❌ No
No conversion intent — meaningless for lift
Page View
❌ No
Engagement ≠ incremental business value
Before running a lift test:
Audit your conversion events for signal quality
Ensure CAPI is sending verified, server-side events
Check EMQ scores (6.0+ minimum, 8.0+ ideal)
Remove or filter known bot traffic from your events
Use Purchase or Qualified Lead as your primary optimization event
A lift test on "Lead" events might show 40% lift — but if 60% of those leads are spam, you've learned nothing actionable.
How to Set Up a Conversion Lift Test
Option 1: Through Meta Ads Manager
Go to Experiments in Meta Ads Manager
Select Conversion Lift
Choose campaigns to include in the test
Configure holdout percentage (start with 10-15%)
Set test duration based on Meta's recommendations
Launch the study
Option 2: With a Meta Account Representative
For larger accounts, working with your Meta rep provides:
Custom test configuration
Sandbox cells for complex setups
Help with multi-cell comparisons
Guidance on statistical power requirements
Option 3: Through a Measurement Platform
Third-party platforms can create, monitor, and analyze Meta Conversion Lift tests within unified dashboards — connecting results to your broader measurement strategy.
LIFT TEST SETUP CHECKLIST═══════════════════════════════════════════════════════════════════════BEFORE LAUNCH:☐ Pixel/CAPI tracking verified and stable☐ EMQ score checked(target 6.0+)☐ Sufficient conversion volume confirmed☐ No major promotions planned during test period☐ Holdout size selected(10-20% recommended)☐ Test duration set(2-4 weeks minimum)DURING TEST:☐ No major creative changes☐ No significant budget shifts☐ No newcampaigns targeting same audience☐ Monitor forexternal factors(competitor activity,news)AFTER TEST:☐ Wait forstatistical significance☐ Calculate iROAS(incremental ROAS)☐ Compare to attributed ROAS☐ Document findings forfuture tests
LIFT TEST SETUP CHECKLIST═══════════════════════════════════════════════════════════════════════BEFORE LAUNCH:☐ Pixel/CAPI tracking verified and stable☐ EMQ score checked(target 6.0+)☐ Sufficient conversion volume confirmed☐ No major promotions planned during test period☐ Holdout size selected(10-20% recommended)☐ Test duration set(2-4 weeks minimum)DURING TEST:☐ No major creative changes☐ No significant budget shifts☐ No newcampaigns targeting same audience☐ Monitor forexternal factors(competitor activity,news)AFTER TEST:☐ Wait forstatistical significance☐ Calculate iROAS(incremental ROAS)☐ Compare to attributed ROAS☐ Document findings forfuture tests
LIFT TEST SETUP CHECKLIST═══════════════════════════════════════════════════════════════════════BEFORE LAUNCH:☐ Pixel/CAPI tracking verified and stable☐ EMQ score checked(target 6.0+)☐ Sufficient conversion volume confirmed☐ No major promotions planned during test period☐ Holdout size selected(10-20% recommended)☐ Test duration set(2-4 weeks minimum)DURING TEST:☐ No major creative changes☐ No significant budget shifts☐ No newcampaigns targeting same audience☐ Monitor forexternal factors(competitor activity,news)AFTER TEST:☐ Wait forstatistical significance☐ Calculate iROAS(incremental ROAS)☐ Compare to attributed ROAS☐ Document findings forfuture tests
How to Interpret Results
Meta Conversion Lift provides several key metrics:
Core Metrics
Metric
What It Means
Lift %
Percentage increase in conversions caused by ads
Incremental Conversions
Number of conversions that wouldn't have happened without ads
Confidence Interval
Statistical reliability of results (higher = more trustworthy)
This is your true return — what you gained that wouldn't have happened without advertising.
THE ROAS REALITY CHECK:ATTRIBUTED vs. INCREMENTAL═══════════════════════════════════════════════════════════════════════YOUR META DASHBOARD SAYS:YOUR LIFT TEST REVEALS:┌─────────────────────────┐ ┌─────────────────────────┐│ │ │ ││ ATTRIBUTED ROAS │ │ INCREMENTAL ROAS ││ │ │ ││ 4.0x │ │ 1.5x ││ ██████████████████ │ │ ███████ ││ ██████████████████ │ │ ███████ ││ ██████████████████ │ │ ││ ██████████████████ │ │ ││ │ │ ││ "Great performance!"│ │ "Breaking even..."││ │ │ │└─────────────────────────┘ └─────────────────────────┘WHY THE GAP?
├── 62% of "attributed"conversions would have happened anyway├── Retargeting is capturing existing demand,not creating it├── Branded search is getting credit fororganic intent└── Only 38% of conversions were truly CAUSED by your ads
THE ROAS REALITY CHECK:ATTRIBUTED vs. INCREMENTAL═══════════════════════════════════════════════════════════════════════YOUR META DASHBOARD SAYS:YOUR LIFT TEST REVEALS:┌─────────────────────────┐ ┌─────────────────────────┐│ │ │ ││ ATTRIBUTED ROAS │ │ INCREMENTAL ROAS ││ │ │ ││ 4.0x │ │ 1.5x ││ ██████████████████ │ │ ███████ ││ ██████████████████ │ │ ███████ ││ ██████████████████ │ │ ││ ██████████████████ │ │ ││ │ │ ││ "Great performance!"│ │ "Breaking even..."││ │ │ │└─────────────────────────┘ └─────────────────────────┘WHY THE GAP?
├── 62% of "attributed"conversions would have happened anyway├── Retargeting is capturing existing demand,not creating it├── Branded search is getting credit fororganic intent└── Only 38% of conversions were truly CAUSED by your ads
THE ROAS REALITY CHECK:ATTRIBUTED vs. INCREMENTAL═══════════════════════════════════════════════════════════════════════YOUR META DASHBOARD SAYS:YOUR LIFT TEST REVEALS:┌─────────────────────────┐ ┌─────────────────────────┐│ │ │ ││ ATTRIBUTED ROAS │ │ INCREMENTAL ROAS ││ │ │ ││ 4.0x │ │ 1.5x ││ ██████████████████ │ │ ███████ ││ ██████████████████ │ │ ███████ ││ ██████████████████ │ │ ││ ██████████████████ │ │ ││ │ │ ││ "Great performance!"│ │ "Breaking even..."││ │ │ │└─────────────────────────┘ └─────────────────────────┘WHY THE GAP?
├── 62% of "attributed"conversions would have happened anyway├── Retargeting is capturing existing demand,not creating it├── Branded search is getting credit fororganic intent└── Only 38% of conversions were truly CAUSED by your ads
Interpreting Lift Results
UNDERSTANDING YOUR LIFT TEST RESULTS═══════════════════════════════════════════════════════════════════════SCENARIO A:STRONG INCREMENTAL LIFT┌─────────────────────────────────────────────────────────────────────┐│ Lift:35% │ Confidence:95% │ Attributed ROAS:4x │ iROAS:3x││ ││ ✓ Ads are driving meaningful incremental conversions ││ ✓ Scale withconfidence ││ ✓ True efficiency is ~75% of reported(3x vs 4x)│└─────────────────────────────────────────────────────────────────────┘SCENARIO B:LOW OR NO LIFT┌─────────────────────────────────────────────────────────────────────┐│ Lift:5% │ Confidence:90% │ Attributed ROAS:4x │ iROAS:0.5x││ ││ ⚠️ Most conversions would happen without ads ││ ⚠️ You're paying for customers who'd convert anyway ││ ⚠️ Consider reallocating budget or testing different strategies │└─────────────────────────────────────────────────────────────────────┘SCENARIO C:INCONCLUSIVE┌─────────────────────────────────────────────────────────────────────┐│ Lift:15% │ Confidence:60% │ Not statistically significant ││ ││ ⚠️ Results are directional but not conclusive ││ ⚠️ Consider extending test duration or increasing holdout ││ ⚠️ Don't make major budget decisions on inconclusive data │
└─────────────────────────────────────────────────────────────────────┘
UNDERSTANDING YOUR LIFT TEST RESULTS═══════════════════════════════════════════════════════════════════════SCENARIO A:STRONG INCREMENTAL LIFT┌─────────────────────────────────────────────────────────────────────┐│ Lift:35% │ Confidence:95% │ Attributed ROAS:4x │ iROAS:3x││ ││ ✓ Ads are driving meaningful incremental conversions ││ ✓ Scale withconfidence ││ ✓ True efficiency is ~75% of reported(3x vs 4x)│└─────────────────────────────────────────────────────────────────────┘SCENARIO B:LOW OR NO LIFT┌─────────────────────────────────────────────────────────────────────┐│ Lift:5% │ Confidence:90% │ Attributed ROAS:4x │ iROAS:0.5x││ ││ ⚠️ Most conversions would happen without ads ││ ⚠️ You're paying for customers who'd convert anyway ││ ⚠️ Consider reallocating budget or testing different strategies │└─────────────────────────────────────────────────────────────────────┘SCENARIO C:INCONCLUSIVE┌─────────────────────────────────────────────────────────────────────┐│ Lift:15% │ Confidence:60% │ Not statistically significant ││ ││ ⚠️ Results are directional but not conclusive ││ ⚠️ Consider extending test duration or increasing holdout ││ ⚠️ Don't make major budget decisions on inconclusive data │
└─────────────────────────────────────────────────────────────────────┘
UNDERSTANDING YOUR LIFT TEST RESULTS═══════════════════════════════════════════════════════════════════════SCENARIO A:STRONG INCREMENTAL LIFT┌─────────────────────────────────────────────────────────────────────┐│ Lift:35% │ Confidence:95% │ Attributed ROAS:4x │ iROAS:3x││ ││ ✓ Ads are driving meaningful incremental conversions ││ ✓ Scale withconfidence ││ ✓ True efficiency is ~75% of reported(3x vs 4x)│└─────────────────────────────────────────────────────────────────────┘SCENARIO B:LOW OR NO LIFT┌─────────────────────────────────────────────────────────────────────┐│ Lift:5% │ Confidence:90% │ Attributed ROAS:4x │ iROAS:0.5x││ ││ ⚠️ Most conversions would happen without ads ││ ⚠️ You're paying for customers who'd convert anyway ││ ⚠️ Consider reallocating budget or testing different strategies │└─────────────────────────────────────────────────────────────────────┘SCENARIO C:INCONCLUSIVE┌─────────────────────────────────────────────────────────────────────┐│ Lift:15% │ Confidence:60% │ Not statistically significant ││ ││ ⚠️ Results are directional but not conclusive ││ ⚠️ Consider extending test duration or increasing holdout ││ ⚠️ Don't make major budget decisions on inconclusive data │
└─────────────────────────────────────────────────────────────────────┘
What to Do With Your Results
If lift is strong (20%+ with high confidence):
Document your iROAS as your true Meta efficiency
Scale spend with confidence that you're driving real growth
Use iROAS (not attributed ROAS) for budget planning
If lift is weak or negative:
Don't panic — this is valuable information
Test different campaign types (prospecting vs. retargeting often show different lift)
Consider whether Meta is better at awareness than conversion
Reallocate budget to channels with proven incrementality
If results are inconclusive:
Extend the test duration if possible
Increase holdout percentage for the next test
Ensure you have sufficient conversion volume before testing again
Common Mistakes to Avoid
Mistake 1: Ending the Test Early
Stopping a test before reaching statistical significance produces unreliable results. Let it run the full recommended duration.
Mistake 2: Running During Promotions
Major sales, new product launches, or seasonal peaks distort results. Test during stable performance periods.
Mistake 3: Changing Variables Mid-Test
Creative refreshes, budget shifts, or audience changes during the test introduce confounding variables. Keep conditions stable.
Mistake 4: Too Small a Holdout
A 5% holdout may not generate enough data for statistical significance. Start with 10-15% unless conversion volume is very high.
Mistake 5: Ignoring Contamination
If other campaigns target the same audience, users in your holdout may still see ads — contaminating results. Pause or exclude overlapping campaigns during the test.
Mistake 6: Testing the Wrong Campaigns
New campaigns with limited data won't produce reliable lift estimates. Test established campaigns with proven performance.
Conversion Lift vs. Other Measurement Methods
Understanding where Conversion Lift fits in your measurement stack:
THE MEASUREMENT HIERARCHY:GROUND TRUTH TO DAILY NAVIGATION═══════════════════════════════════════════════════════════════════════┌─────────────────────────────────────────────────────────────────────┐│ GROUND TRUTH ││(Causal Validation)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ CONVERSION LIFT / INCREMENTALITY │ ││ │ │ ││ │ • Proves what CAUSED conversions │ ││ │ • Randomized experiment = causal evidence │ ││ │ • Use for:Major budget decisions,model validation │ ││ │ • Frequency:Quarterly or when strategy changes │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘││ VALIDATES▼┌─────────────────────────────────────────────────────────────────────┐│ STRATEGIC PLANNING ││(Budget Allocation)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ MMM │ ││ │ │ ││ │ • Aggregate channel contribution │ ││ │ • Accounts foroffline and external factors │ ││ │ • Use for:Quarterly/annual budget planning │ ││ │ • Calibrated BY incrementality results │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘││ INFORMS▼┌─────────────────────────────────────────────────────────────────────┐│ DAILY NAVIGATION ││(Operational Decisions)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ ATTRIBUTION(GA4 / Platform MTA)│ ││ │ │ ││ │ • Shows who converted after touchpoints │ ││ │ • Real-time performance signals │ ││ │ • Use for:Daily optimization,creative testing │ ││ │ • VALIDATED by incrementality and MMM │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘
THE MEASUREMENT HIERARCHY:GROUND TRUTH TO DAILY NAVIGATION═══════════════════════════════════════════════════════════════════════┌─────────────────────────────────────────────────────────────────────┐│ GROUND TRUTH ││(Causal Validation)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ CONVERSION LIFT / INCREMENTALITY │ ││ │ │ ││ │ • Proves what CAUSED conversions │ ││ │ • Randomized experiment = causal evidence │ ││ │ • Use for:Major budget decisions,model validation │ ││ │ • Frequency:Quarterly or when strategy changes │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘││ VALIDATES▼┌─────────────────────────────────────────────────────────────────────┐│ STRATEGIC PLANNING ││(Budget Allocation)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ MMM │ ││ │ │ ││ │ • Aggregate channel contribution │ ││ │ • Accounts foroffline and external factors │ ││ │ • Use for:Quarterly/annual budget planning │ ││ │ • Calibrated BY incrementality results │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘││ INFORMS▼┌─────────────────────────────────────────────────────────────────────┐│ DAILY NAVIGATION ││(Operational Decisions)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ ATTRIBUTION(GA4 / Platform MTA)│ ││ │ │ ││ │ • Shows who converted after touchpoints │ ││ │ • Real-time performance signals │ ││ │ • Use for:Daily optimization,creative testing │ ││ │ • VALIDATED by incrementality and MMM │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘
THE MEASUREMENT HIERARCHY:GROUND TRUTH TO DAILY NAVIGATION═══════════════════════════════════════════════════════════════════════┌─────────────────────────────────────────────────────────────────────┐│ GROUND TRUTH ││(Causal Validation)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ CONVERSION LIFT / INCREMENTALITY │ ││ │ │ ││ │ • Proves what CAUSED conversions │ ││ │ • Randomized experiment = causal evidence │ ││ │ • Use for:Major budget decisions,model validation │ ││ │ • Frequency:Quarterly or when strategy changes │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘││ VALIDATES▼┌─────────────────────────────────────────────────────────────────────┐│ STRATEGIC PLANNING ││(Budget Allocation)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ MMM │ ││ │ │ ││ │ • Aggregate channel contribution │ ││ │ • Accounts foroffline and external factors │ ││ │ • Use for:Quarterly/annual budget planning │ ││ │ • Calibrated BY incrementality results │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘││ INFORMS▼┌─────────────────────────────────────────────────────────────────────┐│ DAILY NAVIGATION ││(Operational Decisions)││ ┌───────────────────────────────────────────────────────────────┐ ││ │ ATTRIBUTION(GA4 / Platform MTA)│ ││ │ │ ││ │ • Shows who converted after touchpoints │ ││ │ • Real-time performance signals │ ││ │ • Use for:Daily optimization,creative testing │ ││ │ • VALIDATED by incrementality and MMM │ ││ └───────────────────────────────────────────────────────────────┘ │└─────────────────────────────────────────────────────────────────────┘
Method
Role
Use For
Limitation
Conversion Lift
Ground Truth
Proving causation, validating models
Periodic, not real-time
MMM
Strategic Planning
Budget allocation, forecasting
Aggregate only, no tactical detail
Attribution (GA4/MTA)
Daily Navigation
Daily optimization, creative testing
Shows correlation, not causation
When They Disagree
Attribution says scale, Lift says no lift: Trust lift — attribution is over-crediting
MMM says cut, Lift says strong lift: Trust lift — MMM may have modeling errors
Both Attribution and Lift agree: High confidence — act on the signal
Conversion Lift is your "source of truth" — use it to validate whether your daily attribution signals are trustworthy.
The Bottom Line
Meta Conversion Lift tests answer the question attribution cannot: Did your ads actually cause these conversions?
Key takeaways:
Lift tests prove causation — not just correlation
Signal quality matters — test on Purchase/Qualified Lead events, not Link Clicks
10-20% holdout is ideal for most tests
2-4 weeks minimum to reach statistical significance
iROAS is your true return — attributed ROAS often overstates impact
Don't end early — let tests complete for reliable results
Use lift as Ground Truth — validate your daily attribution with periodic lift tests
If you're making significant Meta budget decisions without lift testing, you're flying blind. Attribution tells you what happened. Lift testing tells you why.
Get Started
Start Tracking Every Sale Today
Join 1,389+ e-commerce stores. Set up in 5 minutes, see results in days.
