Google Analytics 4 is one of the most widely used analytics tools on the planet. It is also, in most cases, the wrong source of data for the most important variable in your Marketing Mix Modeling analysis.
This guide explains why — and what to use instead.
What is the Dependent Variable in MMM?
In Marketing Mix Modeling, your dependent variable is the business outcome you are trying to explain. It answers the fundamental question: “What are we optimizing for?”
For most businesses, this is revenue. Sometimes it’s order volume, leads, or conversions — but the principle is the same. The dependent variable is the number your model uses to measure the impact of every marketing channel, every promotion, every seasonal effect.
If that number is wrong, imprecise, or inconsistent, every insight the model produces is built on a faulty foundation.
The Core Problem with GA4 Revenue Data
GA4 is designed to help you understand user behavior on your website. It is not designed to be the authoritative financial record of your business. That distinction matters enormously in MMM, where model accuracy depends on a clean, complete, and consistent revenue signal over 12 to 24 months.
Here are the specific reasons why GA4 falls short.
1. Cookie Consent Breaks the Measurement Signal
GA4 relies on browser cookies and JavaScript tracking. When a visitor declines cookie consent — which is increasingly common under GDPR and similar frameworks — GA4 records nothing about that session. No pageview, no conversion, no revenue.
In markets with high consent opt-out rates, GA4 can be systematically underreporting 20–50% of actual transactions, with that gap fluctuating month to month depending on consent banner design, traffic source, and regional mix. This does not just reduce accuracy. It introduces a structurally inconsistent signal — the kind that corrupts a time-series model at its core.
2. Tracking Gaps Cannot Be Retroactively Fixed
Website updates, tag manager misconfiguration, third-party script conflicts, and browser security changes routinely cause GA4 tracking to fail silently. Revenue data disappears for days, weeks, or even months — not because the revenue was not there, but because the measurement broke.
Unlike a transactional database, this cannot be corrected after the fact. There is no audit trail. If your tracking was broken during a key promotional period, GA4 will show a dip in revenue that your model will try to explain — and it will assign that dip to the wrong cause.
3. Revenue Attribution is Platform-Specific, Not Factual
GA4 attributes conversions based on its own attribution logic — by default, data-driven or last-click. This means that the revenue figure attached to any given session already carries an interpretation. It is not raw business outcome data. It is revenue filtered through a platform’s view of how much credit to assign to a particular touch.
For MMM, you do not want attributed revenue. You want actual revenue — the money that landed in your business, independent of any channel’s claim over it. These are fundamentally different things.
4. Ad Blockers and Browser-Level Blocking
A large and growing share of web users run ad blockers or browser extensions that prevent GA4 from firing at all. Depending on your audience, this can affect 15–30% of desktop traffic. These users still buy. GA4 simply does not see it.
5. Cross-Device and Cross-Session Gaps
A customer who researches a product on mobile and completes the purchase on desktop may not be correctly unified in GA4 — particularly if they are not logged into a Google account or if your implementation does not use a robust User ID strategy. Revenue from these journeys can be missed, duplicated, or misattributed.
6. Currency and Regional Inconsistencies
For businesses operating across multiple markets or currencies, GA4’s aggregated revenue data often requires significant normalisation before it is analytically usable. Exchange rate handling, VAT treatment, and refund recording vary — and these inconsistencies compound across a 24-month dataset.
What to Use Instead
Your dependent variable should come from the system of record — the authoritative source that captures every transaction, regardless of how the customer arrived, what browser they use, or what consent they gave.
| Source | Why It Works |
|---|---|
| ERP (e.g., SAP, NetSuite, Odoo) | Complete ledger of all revenue, including offline and manual orders. No tracking dependency. |
| E-commerce backend (e.g., Shopify, WooCommerce, Magento) | Transaction-level data, directly tied to actual order creation, deduction of refunds, and correct currency handling. |
| CRM (e.g., Salesforce, HubSpot) | Ideal for B2B or subscription models where revenue is tied to accounts and deal stages, not sessions. |
| Point-of-Sale system | For businesses with physical retail, the only true source of omnichannel revenue. |
| Finance team exports | Monthly or weekly revenue figures from accounting are often the cleanest, most audited data available. |
The rule is simple: if it would appear in your P&L, it is a valid dependent variable. If it depends on a pixel firing correctly, it is not.
Using GA4 the Right Way in MMM
This does not mean GA4 has no role in your MMM project.
GA4 is an excellent source for independent variables — inputs to the model, not the outcome it is measuring. Specifically, it can contribute:
- Session volume per channel (organic, paid, direct)
- Conversion rates over time (as a behavioral signal)
- Traffic mix shifts that may explain revenue movements not captured by spend data alone
Connect GA4 to your MMM Pilot project as a supplementary data source. Use it to enrich the model’s understanding of how traffic behaves — just do not use it to tell the model how much revenue the business made.
Practical Checklist: Validating Your Dependent Variable
Before you lock in your revenue metric, run through this checklist:
- Does the data come from a source that is independent of browser tracking, cookie consent, or ad-block status?
- Is the data consistent over the full analysis period (12–24 months) with no unexplained gaps?
- Does it reflect actual business revenue — including all channels, not just digital?
- Have refunds, cancellations, or returns been accounted for?
- Are all figures in the same currency, at the same VAT treatment, across the entire period?
- Would this number appear in your management accounts?
If you can answer yes to all of these, you have a dependent variable worth building a model on.
The Bottom Line
GA4 is a valuable tool for understanding marketing behavior. It is not a reliable source for the outcome your business ultimately cares about.
Use your backend, your ERP, or your finance team’s export as the dependent variable. Use GA4 to add depth and context. That combination gives your MMM model the clean signal and rich inputs it needs to produce results you can actually act on.
Your model is only as good as the data you build it on. Start with the truth.