Introduction: From ‘Bad Numbers’ to ‘Real Profits’
“When advertising works but no sales/profits remain” — this is the core dilemma facing performance marketing. If you only focus on advertising performance indicators (CTR, CVR, ROAS, etc.), you will get stuck in numbers. A typical ‘data illusion’ phenomenon that misses actual business resultsIt's easy to fall into it.
The core of this problem is Confusion between ‘correlation’ and ‘causation’It is in Marketers often get caught up in the correlation-based hypothesis that if click-through rates go up, conversions will go up as well. However, actual conversion depends on the ‘causal relationship’ of the seller’s trust and the completeness of the proposal.
Goals of this guide
Breaking the illusion of ‘Sales ROAS’, Google Analytics 4 (GA4) Data-driven attribution (DDA) modelclass ‘Gross Margin’ tracking systemBy introducing , we provide a technical and strategic blueprint to build an actual 'profit'-centered decision-making framework for business.
I. Diagnosis of ROAS optical illusion
Why high advertising performance doesn't lead to profits
1.1. The mechanics of ROAS inflation: the bottom-funnel pitfalls
ROAS is essentially Retargeting, brand search, or existing customers who have already decided to buy This is highest when you focus on campaigns that are almost certain to convert. These campaigns may seem efficient, but in reality, you're spending your advertising dollars targeting customers who "would have bought anyway." Bringing only minimal new business growth (incrementality) There are many cases.
Real Life Case: The Shocking Reality of the ROAS Illusion
As a result of analyzing the correlation between marketing expenditures and new customer orders for brands with annual sales of over KRW 160 billion only 0.22I did it. Although the ROAS indicator increased compared to the previous year, the number of new customers and overall sales decreased. This is “The God of ROAS was satisfied, but the business did not grow.” It clearly shows the status.
1.2. Blind spots in measurement: Last clicks and top-funnel collapse
traditional ‘Last Click’ attribution modelis the culprit that technically supports the ROAS optical illusion. This model attributes the contribution of the upper funnel (e.g. social media, display advertising) to making customers ‘aware’ and ‘considered’ of your brand. treated as 0do. Instead, give all the credit to the last interaction right before the purchase (such as clicking on a brand search).
This distorted data Bad decision to cut upper funnel budget because “ROAS is low”It continues. As a result, the pipeline for bringing in new customers dries up, and in the long run, the parameters for retargeting dry up, leading to a worst-case scenario where the entire funnel collapses.
1.3. Absence of LTV and profit: ‘Sales’ are not ‘profit’
ROAS is compared to advertising cost ‘Revenue’It is calculated based on . but Products with low profit marginsIna One-off customers with low lifetime value (LTV)The high sales this generates can be a real bane to your business.
According to one analysis, Customer A had a higher immediate ROAS, but Customer B had a higher immediate ROAS over the long term. Create 8x the valueI did it. A system optimized only for 'Sales ROAS' causes the company to focus only on Customer A, which produces short-term results, thereby losing long-term growth engines.
Key signs of ROAS illusion and GA4-based prescriptions
| Symptom | Root Cause | Prescription (GA4 Solution) |
|---|---|---|
| “ROAS is 150%, but why is there no profit?” | 1. Only track ‘sales’ (X consider profit margin) 2. Target existing customers, not new customers |
1. Track ‘gross margin’ (GA4 item_profit avatar)2. Simultaneous measurement of ‘CAC’ and ‘new customers’ |
| “CTR is good, but CVR is stagnant” | 1. Confusion between correlation (Click) and causality (Conversion) 2. Only top funnel works, bottom funnel collapses |
1. GA4 Funnel Navigation (Bounce Point Analysis) 2. Check internal processes (UX, payment errors, etc.) |
| “If you turn off social advertising, brand search sales will drop.” | 1. Attribution distortion in last click model 2. Underestimating the contribution of top funnel (Meta) |
1. Introduction of data-driven attribution (DDA) model 2. Check credit redistribution with the ‘Model Comparison’ report |
| “I turned off the new customer campaign because the ROAS was low.” | 1. Ignoring LTV buried in short-term ROAS 2. Depletion of new customer acquisition pipeline |
1. Cohort analysis (LTV measurement by channel) 2. Redefine KPIs with LTV-based ROAS |
II. Rebuilding Your Data Foundation: Designing GA4 Event and Profit Tracking
2.1. Why should you track ‘Gross Margin’?
Not all ‘sales’ have the same value. Sales of 100,000 won may have a COGS of 90,000 won or a COGS of 10,000 won. ‘Sales ROAS’ cannot distinguish between high and low margin products., marketers can unknowingly fall into the trap of optimizing the ROAS of products that ‘sell more and lose more’.
The Value of Gross Margin Tracking
‘Gross Margin’ Tracking (Revenue - Cost of Goods Sold (COGS)) is the first step to transforming marketing from a 'cost center' to a 'profit center' and aligning every decision with actual business contribution.
2.2. GA4/GTM Technology Implementation: Building a ‘Gross Margin’ Tracking System
core principles
GA4's purchase event itemsWe take advantage of the fact that all purchased product information can be contained in an array called . Standard parameters (item_name, price etc.) In addition, A custom parameter representing the ‘actual benefit of the product’(yes: item_profit) to this items array.
Step 1: Data Layer Design
At the time of purchase completion, the server (backend) dataLayer.push()The 'purchase' event must be sent via . At this time, each product (item) price(sale price) and cogsBased on (cost) item_profit(Selling Price - Cost) must be calculated and included in the data layer.
dataLayer.push({
event: "purchase",
e-commerce: {
transaction_id: "T_12345",
value: 385000, // total 'sales'
currency: "KRW",
items: [
{
item_id: "SKU_001",
item_name: "Premium leather jacket",
price: 250000,
quantity: 1,
item_profit: 150000 // Selling price 250000 - Cost 100000
},
{
item_id: "SKU_002",
item_name: "Cashmere scarf",
price: 135000,
quantity: 1,
item_profit: 85000 // Selling price 135000 - Cost 50000
}
]
}
});
Step 2: Set up Google Tag Manager (GTM)
- Create a GTM variable: In dataLayer
item_profitGTM variable whose value you want to read, e.g.dlv_item_profit). - GA4 event tag fix: Check 'Send e-commerce data' in the 'GA4 event - purchase' tag, and make sure to properly reference the items array in the 'Event parameters' area.
- GA4 configuration tags: In GTM's 'Google Tags' configuration:
item_profitSet data to be sent as 'item range'.
Step 3: Register GA4 Custom Definitions
item_profitis custom data that GA4 does not know by default, so you must register it manually in GA4 'Management' > 'Data Display' > 'Custom Definition'.
⚠️ Important technical points
item_profit(Profit) is clearly a ‘Number’ and looks like it should be registered as a ‘Metric’. However, GA4's 'custom metrics' only support 'event range' and Custom metrics in ‘item range’ cannot be created directly from the current UI.It works.
Therefore, the technically correct procedure is to use Number in GTM. item_profit Even if you send a value, In GA4, register this as ‘item range custom dimension’This is what we do. This value can be treated as 'Text (String)' in GA4, so in Looker Studio CAST(dim_item_profit AS NUMBER)You'll need a follow-up operation to convert it back to a number, like this:
GA4 Administrator Settings:
- Name: Item Profit
- Scope: Item
- Description: “Gross margin for individual products (Price - COGS)”
- Item parameter:
item_profit
2.3. Redefine ‘Key Event’ based on business goals
GA4 tracks all user actions as ‘events’. Among these, events that are important to business success ‘Key Event’(formerly ‘Conversion’). The 'purchase' event is automatically registered as a key event.
However, the goal of this guide is to track ‘profit’, not ‘sales’. thus The ‘Gross Margin’ value generated by the event is evaluated as more important than the number of occurrences of the ‘purchase’ event.You must do it.
strategic shift
Change the marketing goal from (X) “Increase the number of purchases (conversions)” to (O) “Let’s increase the total Gross Margin”It means a strategic shift to , and is the core philosophy of the GA4 redesign.
III. Ensuring data integrity: UTM tagging rule sets
3.1. Importance of UTM: ‘Garbage In, Garbage Out’
GA4's sophisticated data-driven attribution (DDA) and benefit-based cost analysis are both Rely on accurately identifying incoming trafficdo. Without UTM (Urchin Tracking Module) tagging rules, channel data gets mixed up (e.g., facebook.com and m.facebook.com are all caught as different sources) and paid traffic is misclassified as 'organic' or 'direct', making analysis impossible.
The key role of UTM tagging
UTM tagging goes beyond simple traffic analysis, Role as the ‘Primary Key’ of the databaseDo this. DDA distributes credit based on source/medium/campaign, and when importing non-Google advertising cost data (e.g. Meta, Naver) into GA4 utm_id or utm_campaignYou need to join the data based on . If the UTM rule is broken, DDA and profit-based ROAS calculation itself is impossible.It's done.
3.2. UTM parameter definitions and standard naming conventions
utm_source(essential): The platform or source that sent the traffic (e.g. google, facebook, naver, edm). Always use lowercase letters.utm_medium(essential): Marketing medium or traffic type (e.g. cpc, display, social_paid, email). Don't confuse platform (source) and tactic (medium).utm_campaign(essential): Specific campaign name (e.g. 2025_spring_sale, new_customer_promo). Adhere to consistent naming conventions.utm_term(Optional/Recommended): Paid search keywords (e.g. leather jacket). Can be used to identify target audience (e.g. lookalike_1pct) in Meta, etc.utm_content(Optional/Recommended): For A/B testing. Distinguish creatives or buttons within the same URL (e.g. red_banner, blue_cta_button).utm_id(recommended): Campaign ID. Key values required when importing and joining non-Google ad spend data into GA4.
Master UTM tagging rule set
| Channel | utm_source | utm_medium | utm_campaign |
|---|---|---|---|
| Google Ads | cpc | (Auto-tagging recommended) | |
| Naver SA | naver | cpc | (Campaign name) |
| Naver GFA/DA | naver | display | (Campaign name) |
| Meta (Facebook/IG) | meta | social_paid | (Campaign Name_AdSet Name) |
| YouTube (non-advertising) | youtube | social_organic | (Video topic) |
| Email (EDM) | edm | (Newsletter_Date) | |
| Affiliate | (Affiliate name) | affiliate | (Promotion Name) |
core rules
- Lowercase: Facebook and facebook are counted as different channels. All letters must be written in lowercase.
- No spaces: Use underscores (_) or hyphens (-) instead of spaces.
- Do not tag internal links: If UTM is used for a banner within the company's mall (e.g. "Spring Sale" → sale page), the user's initial inflow source (e.g. google/cpc) will be overwritten with (the company's mall)/internal_banner, making attribution analysis impossible.
IV. Redefining performance attribution: GA4 attribution model comparison and data-driven (DDA) transition
4.1. The obvious limitations of Last Click
The last click (LC) model ignores all other channels (awareness, consideration) through which the user interacted before purchasing. This is “ROAS is high, but new customers are decreasing” phenomenoncauses . Because marketers stop investing in the top of the funnel.
GA4 uses the 'last click on paid and organic channels' model by default, but this is old practice and not optimal.
4.2. How Data Driven Attribution (DDA) Model Works and Its Value
GA4's DDA(Data-Driven Attribution)uses advanced machine learning (based on the Shapley model) to learn both converting paths and non-converting paths. This model counts all touchpoints (up to 50) that actually contributed to a conversion. Distributing ‘fractional credit’do.
- Accuracy: It distributes credit closer to the actual customer journey than the rule-based model (LC, First Click).
- Holistic View: It helps you discover synergies between channels, such as “Email campaigns later drive search conversions.”
- Budget Optimization: Provides a basis for allocating fair budget to undervalued top-funnel channels.
4.3. How to use the GA4 ‘Advertising’ section: Analyzing the impact of DDA
GA4's 'Advertising' > 'Attribution Analysis' > ‘Model comparison’ reportis the key to this change. In this report, you can compare the 'last click' and 'data-driven' models side by side to see for yourself how 'conversions' and 'revenue' are redistributed for each channel.
Strategic value of model comparison reports
The ‘Model Comparison’ report is not just an analysis tool; The most powerful strategic basis for budget negotiationsno see. Executives and other departments may still have a last-click bias that “social media only costs money and doesn’t generate sales.”
The report states, “The Paid Social channel generated 100 conversions based on last click, but based on DDA, Contributed to 350 conversionsI did it. “This means that the channel is undervalued by +250%,” the only quantitative evidence available.
Channel performance changes by attribution model (simulation)
| channel | Conversion (Last Click) | Conversion (Data-Driven) | rate of change | Strategic Implications |
|---|---|---|---|---|
| Paid Social | 50 | 150 | +200% | Underrated - High funnel (awareness) contribution. Consider increasing the budget. |
| Organic Search | 120 | 140 | +16.7% | Sustain - Consistently contribute throughout the journey. |
| Paid Search - Non-Branded | 200 | 180 | -10.0% | Slightly overrated - mid-funnel (consideration) role. |
| Paid Search - Brand | 400 | 200 | -50.0% | Severely overrated - Bottom of the funnel (confirmed) role. The main culprit of LC ROAS. |
| Direct inflow | 230 | 330 | +43.5% | Underrated - the end result of brand awareness. |
V. Performance Report Format: Designing a Profit-Focused KPI Dashboard (Looker Studio)
5.1. Define new key performance indicators (KPIs)
For ‘profit’-driven decision-making, we redefine the KPI stack as follows:
5.2. Looker Studio dashboard implementation
A complete report with ‘Gross Margin ROAS’ and ‘CAC’ is almost impossible to implement inside GA4. The reason is:
- To calculate 'Gross Margin ROAS' you need 'Gross Margin' and 'Gross Advertising Spend'.
- GA4 automatically imports Google Ads costs, but non-Google advertising costs such as Meta and Naver must be imported manually.
- 'Gross Margin' (
item_profit) is stored as an 'item range custom dimension' in GA4, so it is difficult to use it directly as a number (metric) in the GA4 exploration report.
The only solution is to use Looker Studio to blend multiple data sourcesno see.
Step 1: Data Blend and Create ‘Total Ad Spend’
Looker Studio creates a 'Total Cost' field by blending three data sources by date and campaign:
- GA4 data source: sales, campaign, source/medium, dim_item_profit
- Google Ads data sources: Advertising cost (Cost)
- Manual upload source (Google Sheets): Date, campaign ID, source/medium, advertising cost (non-Google cost such as Meta, Naver, etc.)
Step 2: Create a Looker Studio ‘Calculated Field’
The core engine of Profit Driven Dashboards is Looker Studio’s Calculated Fields.
// [Calculated Field 1] Gross Margin
SUM(CAST(dim_item_profit AS NUMBER))
// [Calculated field 2] Gross Margin %
SUM(Gross Margin) / SUM(Purchase Revenue)
// [Calculated Field 3] Margin ROAS
SUM(Gross Margin) / SUM(Total Cost)
// [Calculated Field 4] Revenue ROAS
SUM(Purchase Revenue) / SUM(Total Cost)
// [Calculated Field 5] CAC (Customer Acquisition Cost)
SUM(Total Cost) / COUNT_DISTINCT(User ID)
Performance report layout by campaign-segment
[Area 1] Core KPI scorecard
- Total Advertising Cost
- Total Sales (Purchase Revenue)
- Gross Margin
- CAC (Customer Acquisition Cost)
- Revenue ROAS
- Gross Margin ROAS ⭐
- Gross Margin %
[Area 2] Time Series Chart: Benefits vs Costs
- X-axis: Date
- Y-axis: Total advertising spend, gross margin, total sales (line/bar chart)
[Area 3] Campaign-Segment Performance Table
- Dimension: Campaign (or Source/Medium)
- Metrics: Total Ad Spend, Clicks/Session, Order/CPO, New Customer/CAC, Total Sales/Sales ROAS, Gross Margin, Gross Margin ROAS ⭐
VI. Securing long-term growth drivers: 1-90 day cohort analysis
6.1. Why do you need a cohort? - Importance of LTV
The ‘Margin ROAS’ report designed earlier also has limitations in that it only shows immediate profits. but Even if the initial ROAS is low, securing customers with high LTV (Lifetime Customer Value) is the way to win in the long run.no see.
‘1-90 days cohort regression rate’is the most practical way to measure this LTV.
6.2. Setting up a cohort report using GA4 ‘Exploration’
GA4 Use the ‘Explore’ > ‘Template Gallery’ > ‘Cohort exploration’ report.
Cohort creation (inclusion criteria):
- Criteria: First visit (First touch - acquisition date)
- Reason: Since the goal is to track the LTV of new customers by channel, cohorts are created based on when users ‘first arrived’.
Cohort Return (Based on Return):
- Criteria: Purchase (Any transaction or purchase)
- Reason: Since the purpose of ‘return rate’ is ‘repurchase rate’, set ‘purchase’ event as the return standard.
Other settings:
- Granularity: Monthly (1-90 days is 3 months' worth of data)
- Metric: Active Users or Purchasers
- Breakdown: ⭐ First user source/medium
6.3. Cohort report analysis and utilization
Through the above settings “Among customers who first came in through google/cpc in January 2025, what percentage repurchased in February (Month 1) and what percentage repurchased in March (Month 2)?”You can accurately compare by channel.
Strategic value of cohort analysis
The cohort report becomes the standard for determining whether 'CAC' is a 'cost' or an 'investment'.
For example, assuming that google/cpc's CAC is 50,000 won and meta/social_paid's CAC is 80,000 won, looking at this data alone makes meta's performance look bad.
However, in the cohort report, the 90-day repurchase rate of google/cpc inflow customers is 5%. The 90-day repurchase rate of meta/social_paid incoming customers is 30%If you find that the high CAC of the meta channel is ‘Long-term investment to secure high-value customers’It can be reinterpreted as .
A complete business growth framework
The three core KPIs (CAC, Gross Margin %, and Cohort Reversion Rate) form a complete business growth framework in themselves:
- CAC (Customer Acquisition Cost): “The investment it takes to bring in one new customer.”
- Gross Margin %: “The profit margin we immediately recoup on your first purchase.”
- Cohort regression rate (repurchase rate): “How much and how will the investment be recovered in the future (LTV)?”
The combination of these three KPIs is “A specific campaign (A) has a low CAC (KRW 10,000) and a high margin (40%), making an immediate profit, but the other campaign (B) has a high CAC (KRW 50,000) and a low margin (20%), which results in a loss on the first purchase, but a high cohort repurchase rate (60%), resulting in a gross margin 3 times that of Campaign A within 90 days.”It enables high-level decision-making.
VII. Implementation Roadmap: Strategic Recommendations for Successful Framework Transition
Phase 1: Technical implementation and data verification (0-30 days)
| parking | core action | in charge |
|---|---|---|
| Week 1 | Stakeholder consultation: Marketing, finance/accounting (COGS data), development team (data layer) kickoff meeting | warrior |
| Week 2 | GTM/GA4 design: ‘Event Tree Map’ and ‘UTM Rule Set’ finalized as enterprise standards | marketing |
| Week 3 | Development and implementation: Apply custom parameters such as item_profit to the data layer of the purchase event. GA4 Custom Definitions and GTM Tags Setup Completed | Development Team, Marketing |
| Week 4 | Verification and Debugging: Verify that item_profit values are collected correctly via GA4 DebugView and GTM Preview mode | Marketing, Development Team |
Phase 2: Building a reporting and analysis system (30-60 days)
| parking | core action | in charge |
|---|---|---|
| Week 5 | Cost data integration: Establish a process to manually/automatically import advertising costs from Meta, Naver, etc. into GA4 (or Google Sheets) | marketing |
| Week 6-7 | Building a Looker Studio dashboard: Completed ‘Gross Margin ROAS’ dashboard prototype based on ‘Performance Report Format’ | Marketing, Data Team |
| Week 8 | Ensure data reliability: Cross-verify whether GA4 data matches actual business backend (sales, profit) data | Finance, Marketing |
Phase 3: Cultural Transformation and Optimization (After 60 Days)
| parking | core action | in charge |
|---|---|---|
| Week 9 | Contribution model conversion: Analyze and share the impact of DDA through the GA4 ‘Model Comparison’ report and reach an organizational agreement to change the GA4 attribute default model to ‘DDA’ | marketing, management |
| Week 10 | Introduction of cohort analysis: Start of ‘cohort regression rate by channel’ analysis | Marketing, Data Team |
| Week 11 | Launch a new optimization loop: reallocate budget based on Margin ROAS and 90-day LTV | marketing |
| Week 12 | Internal process inspection: In addition to advertising data, internal factors (UX, payment errors, delivery policy, etc.) that cause a decrease in conversion rate are inspected and improved. | warrior |
Conclusion: Breaking the ‘ROAS illusion’ and moving towards ‘sustainable growth’
The framework (profit tracking, DDA, cohort) proposed in this guide is not a simple GA4 setup guide. This is Upgrading the operating system that changes the marketing success standard from 'fake numbers (sales)' to 'real results (profits)'no see.
success story
These changes occur when technical precision and organizational will are combined. Increase sales by 20%+It leads to practical business results such as: Now marketers who are ‘stuck in numbers’ Strategist who ‘designs profits’It is time to be reborn.
key turning point
- Conversion of measurements: Revenue → Gross Margin
- Conversion of contributions: Last Click → Data Driven Attribution (DDA)
- Shift in perspective: Short-term ROAS → Long-term LTV (Cohort analysis)
- Transform your decision making:
BlueCanvas is with you from building this data-based marketing framework to creating actual business results.