Introduction: Why Model Transparency Matters
When brands adopt advanced measurement like Fospha, your executive team will inevitably ask: "How does it work? How did you get to these numbers?" Model Spotlight ensures it's not a black box—giving you complete confidence to explain our methodology internally.
This transparency isn't just nice-to-have; it's essential for getting buy-in on the strategic budget shifts our data will reveal. When you see significant differences between Fospha's and other measurement tools (GA4, pixel-based measurement or even a traditional MMM), you need to understand exactly why those differences exist and how we calculate them.
How to Use the Dashboard
Accessing the Model Waterfall
Navigate to the Attribution Comparison Dashboard
Click on the "Explain" tab
Select your desired date range and channel
Review each step of the model waterfall
1. Explore the Model Waterfall
Start with GA4’s baseline numbers to see what your team is used to.
Step through each stage (validation, click measurement, reconciliation, impression modelling) to understand how Fospha builds on GA4.
Use the tooltips and visualizations to pinpoint what changes at each step and why.
2. Answer Stakeholder Questions with Evidence
When presenting Fospha data internally:
Start with GA4 baseline: "Here's what last-click measurement shows us..."
Walk through each step: "Here's how our numbers change at each stage..."
Highlight the incremental value: "This is the additional impact we discover through full-funnel measurement..."
Connect to business outcomes: "This transparency helps us make confident budget decisions..."
Examples of how to answer tough questions:
CFO: “Why are Fospha’s numbers different from GA4?” → Show the Reconciliation step that accounts for missing sales in GA4 and aligns to eCommerce actuals (your finance source of truth).
CMO: “How do we know brand-building channels are working?” → Show the Impression Modelling step, where Fospha quantifies the revenue impact of awareness campaigns that GA4 can’t see.
Channel Lead: “Why should I trust reallocating budget?” → Show the Click Measurement step where Fospha reassigns credit from last-click winners (Direct/Brand PPC) to the channels that actually drove demand.
3. Build Confidence to Make Budget Decisions
You can request Model Spotlight validation metrics (e.g. NRMSE) to see how reliable Fospha’s outputs are.
Reassure finance and leadership that every number is scientifically validated — not just a black-box estimate.
Make budget decisions knowing you’re not guessing: you can show how accurate that number is, how it was derived, and the level of confidence behind.
Use case: When debating shifting $100K from one channel to the other, show how Fospha’s outputs are validated daily against unseen data, and present confidence intervals to prove the reallocation is backed by science.
Understanding the Model Waterfall
The Model Waterfall shows you exactly how your data transforms at each stage, from GA4's starting point to Fospha's complete full-funnel measurement. Let's walk through each step:
Step 0: GA4 Last-Click (Starting Point)
Every measurement journey begins with raw conversion data from GA4—the last-non-direct-click view of customer journeys. These numbers show final touchpoints but miss most of the influence from paid media that created the demand in the first place.
Many sales manifest in channels that didn't influence that sale—Direct, Organic Search, Brand PPC—because of privacy constraints and attribution limitations. This is where last-click measurement stops and declares victory. For us, it's just the starting point for revealing what actually drives your growth.
Step 1: GA4 Validation & Cleaning
Before any modeling begins, we ensure GA4 data is clean and modeling-ready:
UTM Standardization: Parameters that weren't tagged correctly get reassigned to the right channels
Duplicate Removal: Transactions are deduplicated to prevent double-counting
Order Validation: Test orders and staff purchases are filtered out
Transaction Matching: Order IDs are matched to ensure we're only modeling valid transactions
This foundation ensures our modeling reflects real customer behavior, not tracking errors or internal activity.
Step 2: Click Measurement (Daily MMM)
We move beyond last-click using a "wisdom of crowds" approach—combining four different models:
Last-click: Captures immediate conversion drivers (deterministic signal)
Data-driven attribution: Leverages Google's device graph for cross-device insights
Linear regression: Identifies quality engagement patterns that correlate with sales
XGBoost: Detects complex, non-linear relationship patterns
By averaging these approaches, we create a more accurate view of which clicks actually drove sales, giving previously under-credited channels the recognition they deserve. This robust click measurement creates a solid foundation for full-funnel measurement while staying grounded in deterministic signals.
Step 3: Reconciliation with eCommerce Data
GA4 typically misses 5-30% of transactions due to cookie consent, ad blockers, and privacy restrictions. We use your eCommerce platform as the source of truth—the actual orders and revenue your finance team sees.
We calculate the gap between GA4-tracked orders and eCommerce-reported orders, then proportionally redistribute this "missing" revenue based on channel credit from our click measurement stage. This ensures we're modeling 100% of your actual sales, creating alignment between marketing measurement and finance reporting.
Step 4: Post-Purchase Attribution
We integrate deterministic, zero-party data—what customers actually tell you about their journey. When customers complete post-purchase surveys answering "How did you hear about us?" or use discount codes tied to specific channels, we tie those sales back to the right source.
This is especially powerful for influencer campaigns where click data is weak (influencers aren't great at UTM management) but impact is real. This deterministic signal helps channels that create demand but don't capture clicks get the recognition they deserve.
Step 5: Impression Measurement (Daily MMM)
This is where the magic happens—and why brands choose Fospha. Brand-building channels like YouTube, TikTok, and Snapchat drive significant demand but rarely get credit when sales later come through Direct, Organic, or Brand PPC.
We use XGBoost—a machine learning model that handles complex relationships—to measure true impression impact daily without needing massive data volumes. We apply Shapley values (game theory-based method) to assign attribution credit because our testing showed it delivers the most stable, fair results over time.
We model impact down to campaign objective level, validate every relationship for statistical significance, then distribute credit to the ad level. This reveals how upper-funnel impressions create the demand that converts later.
Key Insight: Every marketer who's invested in paid social channels intuitively knows that as their investment increases, they see growth in "non-attributed" channels like Direct, Organic Search, and Brand PPC. Fospha quantifies this relationship and credits back to the impression sources where demand was actually created.
Model Validation & Confidence Metrics
Every step is visible, scientifically validated, and tailored to your business — so you can see how the numbers are built, understand why they differ from GA4 or pixels, and trust them to guide real budget decisions.
Real-Time Validation
Daily Retraining: Your model retrains with latest data and undergoes rigorous testing on unseen data
RMSE Monitoring: We monitor prediction accuracy using metrics like Root Mean Square Error
Adaptive Modeling: Our Daily MMM automatically adapts to the changes in your data, optimizing training windows and segment splits
Conclusion
Model Spotlight transforms Fospha from a measurement tool into a growth intelligence platform. By understanding exactly how we arrive at our attribution, you can make strategic decisions with complete confidence, get internal buy-in for budget shifts, and unlock the full potential of your marketing investment.
Remember: this isn't about inflated numbers—it's about complete measurement that accounts for how modern marketing actually works. You now have a unified view trusted by marketing, finance, and leadership to make confident, growth-driving decisions based on the full picture of marketing performance.