Turn a performance problem into the next action.
Practical notes for narrowing down CPA, ROAS, budget, creative, and measurement issues — then checking the answer with your own data.
Turn this question into a decision
Every guide and definition leads to a matching free analysis tool. No sign-up; raw data stays in your browser.
See what actually moved performanceNarrow the cause with channel, campaign, and creative contributions.Analyze performance change →Find creative fatigue and swap priorityIdentify what to replace and what to produce next.Analyze creative fatigue →Start with the weekly movement mapUse a darker-is-newer path to spot repeated opposite movement, then continue to detailed diagnosis when needed.Open movement map →Know where the next dollar belongsSimulate channel scale-ups and pull-backs with marginal CPA, ROAS, and headroom.Plan budget allocation →Review CPA, ROAS, and pacing togetherUse one CSV to check the weekly operating state and anomalies.Open operations dashboard →Holdout and pre/post incrementalityEstimate what advertising added beyond attributed conversions.Analyze incrementality →
User Churn Timing: Find When Users Leave With Survival Curves
Dropping users who have not left yet pulls churn forward. Use a survival curve and monthly churn risk to find when users actually leave.
Simpson's Paradox: Every Campaign Improved, Yet Conversion Fell
When every campaign's conversion rate rises but the total falls, split the change into volume shifting between campaigns and change within campaigns.
Weekly Marketing Report Template: From Metrics to Decisions
Build a weekly marketing report with three weeks of campaign data. Follow a worked example from comparison to a saved decision and next-week review.
Marketing Report Automation: BigQuery, Google Sheets and CSV
Connect BigQuery and Connected Sheets refreshes to analysis-ready CSVs. Check permissions, costs, aggregation grain and the manual import boundary.
CAC Payback Period: Calculation and Cash Recovery
Calculate CAC payback from a new-customer cohort’s monthly contribution. Distinguish refunds, churn, unrecovered costs and revenue-based tool estimates.
App Store A/B Testing: Designing and Reading PPO Experiments
Before-and-after gets icon and screenshot tests wrong. How Apple PPO and Play listing experiments differ, and how to read them.
App Store Conversion Rate Dropped: Page Problem or Traffic Mix?
The blended rate hides the cause. Split conversion by source and share to separate two problems whose fixes are opposites.
SKAN 4 Strategy: Which SKAdNetwork 4.0 Window to Open
The order to settle measurement windows, value resolution, and campaign structure when moving to SKAN 4.
Why SKAN Attribution and MMP Numbers Disagree
SKAN and an MMP count different things at different times under different rules. Which source to trust for which question.
SKAN Conversion Value Schema: What Fits in 64 Slots
How to decide which behavior goes into six bits, and the design mistakes that leave half the slots empty.
Measuring TV and Offline Ad Impact Online
Compare online demand across exposed and unexposed windows to estimate offline ad impact, and label its limits.
Ad Creative Regression: Compare Thumbnails Without A/B Tests
When you cannot run an experiment, use the creatives you already served to compare the effect of each attribute.
Ad Creative Performance Analysis: Hook, Length, or Format?
Instead of changing one production element at a time, hold several elements constant and compare them all in one pass.
Ad Budget Scaling: Find the Limit Before CPA Rises
Compare marginal CPA with average CPA to separate channels with headroom from channels already saturated.
Measuring Brand Campaign Lift Without Clicks
Estimate brand campaign impact from brand search and direct traffic time series when clicks cannot be attributed.
Advertising Uplift: Measure Net Lift With a Holdout Test
Measure advertising uplift with a holdout test, learn when good CPA or ROAS is not enough to scale, and interpret net lift in order.
Performance Marketing Analysis: Choose the Right Method
Choose between a dashboard, variance decomposition, saturation, A/B testing, incrementality, and MMM based on the question and data you actually have.
Multicollinearity Before MMM: How to Read High VIF
Check multicollinearity with VIF before MMM and learn what to change first when the data cannot support channel-level contribution.
Apple Search Ads Keyword Optimization: Terms, Negatives, Bids
Use ASA search-term reports to find Exact-promotion candidates and adjust CPT bids from budget pacing and target CPA.
ROAS Dropped: Causes and Fixes Before You Cut Budget
When ROAS falls, separate data maturity, channel movement, conversion leakage, and marginal efficiency before changing budget.
Ad Creative Testing: How Many to Run, and How to Judge
Calculate how many ad creatives your budget can support, define one hypothesis, and classify winners, holds, and losers without relying on CTR alone.
Ad Optimization Events: Use the Aha Event, Not Installs
Use the early action that predicts retention to connect ad optimization, onboarding, and durable growth.
Performance Marketing Metrics: Read CTR, CVR, CPA, and ROAS in Order
Review CTR, CVR, CPA, and ROAS in order to locate whether a performance problem sits before or after conversion.
Conversion Rate Optimization: Find the Funnel Step That Leaks
Find the highest-priority funnel drop-off, then check message match, friction, traffic quality, and offer strength before validating one change at a time.
Ad Performance Drop: Why CPA Rises and CTR Falls
Before replacing creative when CPA rises or CTR and conversion rate fall, diagnose measurement, channel mix, efficiency, and the funnel in four steps.
Marketing Budget Allocation: Split Channels by Marginal CPA and ROAS
Compare budget increases, reallocations, and cuts with marginal ROAS or CPA, then check the assumptions and limits of the estimated response curve.
Why Retargeting Looks Better Than It Is
Why retargeting belongs in its own campaign, how to segment the audience, and how to distinguish deep links from deferred installation flows.
Postback Errors: Diagnose Zero Installs and Lost Events
Troubleshoot postback problems from the last verified point in the app-to-MMP-to-network path. Includes SAN vs. S2S context and a practical zero-data checklist.
Meta Advantage+ Diagnosis: Reading iOS vs Android
The parts a marketer can actually adjust in an automated Meta Advantage+ App (AAP) campaign — OS split, event priority, and bidding stages.
The 3-Second Hook: Reduce Video Ad Drop-off
Four hook patterns to test that stop the scroll in a video ad's first three seconds, plus a checklist for the opening frame.
Event Taxonomy & SDK Errors: Names, Gaps, and QA
A practical event taxonomy guide for SDK event names, parameters, user properties, platform mapping, versioning, and developer QA.
ASA Keyword Expansion: When to Promote Discovery to Exact
Split the campaign that finds terms from the campaign that spends, and promote only terms that clear both volume and CPA.
Why Creatives Get Rejected or Cropped by Networks
Ignore each platform's image and video specs and UI safe zones, and creative gets rejected in review or has its core message cropped.
GA4 Numbers Don't Match: 7 Reasons They Differ
Check seven reasons GA4 numbers differ: metrics, dates, processing, thresholds, sampling, the other row and attribution. Start with the symptom.
Attribution Data Mismatch: Why GA4, MMP, and Media Disagree
Meta, GA4, MMP, and your order DB all report different conversion counts — because of attribution windows, view-through, credit date, and last-touch overlap.
LTV:CAC Ratio Explained: How to Calculate It Correctly
More important than the 3:1 benchmark is whether the number is right. Denominator, revenue vs margin, channel split: three common mistakes.
iOS Measurement: ATT, SKAN, and Conversion Value Explained
iOS performance may not be bad, just invisible. ATT cuts user-level measurement and SKAN gives delayed aggregate signal. How to judge anyway.
When Google UAC Slips: Diagnose Learning and Bids
The levers a marketer actually controls inside an automated Google App Campaign — bidding, assets, and conversion events.
CPM vs CPC vs CPI vs CPA: Meaning, Formulas, and Which to Use
At 1% CTR, CPC equals CPM divided by 10. Compare CPI, CPA, CPM and CPC denominators and formulas, then inspect changes without assuming a cause.
Cohort Analysis: Reading D1, D7, and D30 Retention Cohorts
Compare only mature cohorts when reading D1, D7, and D30 retention. Handle incomplete cohorts, retention definitions, sample size, and heatmaps correctly.
Ad Cannibalization: Did More Ads Shrink Organic Traffic?
Compare weekly paid and organic changes together, with the data conditions and example that separate co-movement from real paid or internal cannibalisation.
Campaign Anomaly Detection: Is Today's CPA Spike Real?
When CPA or conversions swing on a given day, how to separate noise from a real anomaly and trace the cause.
ASO Strategy: Keywords, Screenshots, and Store Conversion
An ASO strategy fixes both search visibility and the store conversion after an ad click. Icon, screenshots, ratings, then keywords — in order.
How to Find the Aha Moment: Early Actions and Retention
Find Aha Moment candidates by comparing action, timing, count, reach, Lift, and F1—then test whether the behavior causes retention before operationalizing it.
The AI-Era Performance Marketer: Skills That Still Matter
AI is said to replace marketers, but what machines took over and what is still yours are different things. Here is what operators still need to verify.
Ad Learning Phase: Why CPA Swings While the Model Learns
Misread the learning phase and your budget leaks. Why CPA bounces, how changes affect evaluation, and why the early numbers mislead.
Broad vs Narrow Targeting: How to Choose an Ad Audience
Compare broad and narrow audiences with a CPM, CTR and CPA example. Check reach, frequency, new customers and test conditions before changing budgets.
Performance Marketer Skills: What to Learn, In Order
The growth path from operations to data, decision-making, and causal inference, and why tools should be learned by function, not by name.
Marketing Mix Modeling: Measure Channel Contribution
Last-click credits whichever channel stood last, which is why brand search always looks heroic. How MMM back-calculates model-based contribution estimates.
Incrementality Measurement: Validate Ad Lift With Holdouts and DiD
Dashboard ROAS should only be half-trusted. Incrementality measurement and three holdout designs for measuring what ads actually create.
Correlation vs Causation: Verify Lift With Experiments
Distinguish correlation from causation, identify confounding and reverse causality, and check the assumptions behind experiments and causal claims.
A/B Testing: Sample Size, Significance, and Reading Results
From random assignment, sample size, and statistical significance to the early-stopping (peeking) trap to watch for.