If Shopify, GA4, Google Ads, and your advertising platforms show different numbers, the first question is usually:

Which one is correct?

I don't think that's the right place to start.

Different ecommerce platforms can report different numbers because they are measuring different things.

They may use different:

• definitions
• attribution rules
• tracking mechanisms
• time windows
• event models
• identities
• filters
• reporting logic

So two systems can disagree without either one being fundamentally wrong.

The more useful question is:

What exactly is each system measuring, and why should these measurements match?

That changes the problem from choosing a number to understanding the evidence.

Why Shopify and GA4 Can Show Different Numbers

Imagine a store reports:

Shopify:
120 orders

GA4:
105 purchases

Google Ads:
72 conversions

Meta:
83 conversions

It is tempting to decide which number looks most believable.

But these numbers may represent different measurements.

Shopify may be reporting orders recorded by the commerce system.

GA4 may be reporting purchase events captured by analytics tracking.

Google Ads may be reporting conversions attributed to Google Ads.

Meta may be doing the same according to its own attribution and tracking configuration.

These are related measurements.

They are not automatically identical.

That distinction matters before you start diagnosing the business.

Define the Measurement Before Comparing the Number

Before comparing two numbers, write down what each number means.

For example:

Shopify orders

The question might be:

How many orders did the commerce system record?

GA4 purchases

The question might be:

How many purchase events did the analytics system capture?

Google Ads conversions

The question might be:

How many conversions were attributed to Google Ads under its configured conversion settings?

Those questions are different.

If the questions are different, different answers are not necessarily a data problem.

This is one of the most common mistakes I see in ecommerce analysis:

comparing numbers before defining them.

Measurement Boundaries Matter

Every analytics system observes the business from a particular boundary.

A commerce platform primarily records transactions.

An analytics platform observes user interactions and events.

An advertising platform is interested in advertising interactions and attributed conversions.

You can think of them as different windows into the same business.

For example:

Customer clicks an ad
        ↓
Visits the store
        ↓
Leaves
        ↓
Returns through organic search
        ↓
Purchases

The commerce system primarily cares about:

Did the order happen?

The advertising platform may care about:

Did this advertising interaction receive credit for the conversion?

The analytics platform may care about:

What happened during the customer's journey?

All three can describe the same customer differently.

Attribution Creates Another Layer of Difference

Suppose someone clicks a Google ad on Monday.

They don't buy.

On Wednesday, they return through an organic search result.

They purchase.

Should Google Ads receive credit?

That depends on the attribution configuration.

Should GA4 describe the earlier advertising interaction?

That depends on what data was captured and how the report is configured.

Should Shopify record an order?

Yes, assuming the transaction completed successfully.

The order itself is not necessarily in dispute.

The attribution is.

This is why comparing:

Shopify orders

with:

Google Ads conversions

as if they were the same metric can create confusion.

They answer different questions.

Tracking Problems Can Create Real Discrepancies

Not every difference is expected.

Analytics tracking can be affected by:

• consent settings
• ad blockers
• browser restrictions
• missing events
• JavaScript errors
• incorrect configuration
• cross-domain problems
• tagging errors
• checkout implementation

Imagine:

Shopify orders: 120
GA4 purchases: 42

That is a much more interesting discrepancy than:

Shopify orders: 120
GA4 purchases: 105

The second difference may be explainable by measurement and tracking boundaries.

The first deserves an investigation.

But even here, the right question isn't:

Which platform should I trust?

It is:

Where could the missing measurement have disappeared?

Time Windows Can Make Good Data Look Wrong

Before investigating technical problems, align the comparison.

Check:

• date range
• timezone
• attribution window
• conversion event
• filters
• currency
• event definition
• reporting delay

Suppose Shopify reports 100 orders on September 10.

Google Ads reports 67 conversions.

That does not automatically mean Google Ads missed 33 orders.

Only some orders may be attributable to Google Ads.

And the two platforms may use different reporting windows.

The comparison needs to be made on equivalent terms before the difference becomes meaningful.

Definitions Matter More Than Labels

The same word can mean different things.

For example, two systems may both show:

Revenue

But one may include or exclude:

• discounts
• taxes
• shipping
• refunds
• canceled orders
• currency conversion
• timing differences

The label isn't enough.

You need the definition.

This is especially important when someone exports multiple datasets and gives them to AI.

If the model isn't told what each column actually represents, it can compare numbers that should never have been compared in the first place.

A Better Way to Diagnose Analytics Discrepancies

I would use this sequence:

Different Numbers
      ↓
Identify Each System
      ↓
Define What Each Measures
      ↓
Align Date and Time
      ↓
Compare Event Definitions
      ↓
Compare Attribution Rules
      ↓
Check Tracking Coverage
      ↓
Identify Expected Differences
      ↓
Investigate Unexpected Differences

The important step is separating:

expected difference

from:

unexpected difference.

Not every discrepancy needs to be fixed.

Some discrepancies simply reflect how the systems were designed.

When the Difference Suddenly Changes, Pay Attention

The absolute difference isn't always the most interesting signal.

The change in the difference can be more useful.

Imagine:

Previous month:

Shopify → 100
GA4 → 94

Difference → 6%

Then:

Current month:

Shopify → 100
GA4 → 62

Difference → 38%

The question is now:

What changed?

Maybe:

• checkout tracking changed
• consent configuration changed
• a new domain was introduced
• a tracking script broke
• an event name changed
• a campaign changed
• the analytics implementation was modified

The discrepancy itself has become a diagnostic signal.

Use the Right System for the Right Question

I don't think ecommerce teams need one universal “source of truth” for every question.

Different systems can be useful for different questions.

For example:

Shopify

Useful for questions about:

• orders
• products
• sales
• commerce operations

GA4

Useful for questions about:

• user behavior
• acquisition
• sessions
• events
• navigation patterns

Advertising platforms

Useful for questions about:

• campaigns
• advertising audiences
• attributed conversions
• advertising performance

The mistake is forcing one platform to answer every question.

The better approach is:

Use the evidence source that matches the question.

AI Can Help Investigate Discrepancies

This is also where AI can become useful.

But I wouldn't simply upload four reports and ask:

“Why are these numbers different?”

Give AI the measurement context.

For example:

Shopify:
120 completed orders

GA4:
105 purchase events

Google Ads:
72 attributed conversions

Time period:
September 1–7

Known change:
Checkout tracking was modified on September 4.

Investigate why these systems may produce different numbers.

For each possible explanation:

1. Explain which measurement boundary it affects.
2. Identify evidence that supports the explanation.
3. Identify evidence that contradicts it.
4. Identify what should be checked next.

Do not assume that one platform is automatically correct.

Now AI has an investigation task.

It isn't merely comparing numbers.

Don't Let AI Turn a Measurement Problem Into a Business Diagnosis

This distinction matters.

Suppose GA4 shows fewer purchases than Shopify.

You should not immediately conclude:

“The conversion rate is falling.”

That may be completely unrelated.

You first need to establish whether the measurement changed.

Otherwise you can end up diagnosing the business based on a tracking problem.

The sequence should be:

Measurement discrepancy
        ↓
Understand the discrepancy
        ↓
Establish whether the underlying business metric changed
        ↓
Then diagnose the business problem

This is one reason I prefer evidence-first ecommerce diagnosis.

You don't want AI to explain a problem that exists only inside the reporting layer.

Sometimes the Data Is Not the Problem

There is another possibility.

The data may be doing exactly what it was designed to do.

The problem may be that you're asking it a question it cannot answer.

For example:

“Why did customers stop trusting this product?”

A dashboard probably isn't enough.

You may need:

• customer reviews
• support conversations
• product questions
• customer comparisons
• survey responses
• social discussions

The answer isn't always another metric.

Sometimes you need another type of evidence.

This is closely related to why ecommerce diagnosis needs customer evidence.

A Simple Checklist

When two ecommerce systems show different numbers, ask:

1. Are the date ranges identical?
2. Are the timezones identical?
3. Are the metric definitions identical?
4. Are the conversion events identical?
5. Are the attribution rules identical?
6. Are the attribution windows identical?
7. Are the filters identical?
8. Are the identity assumptions identical?
9. Is tracking coverage comparable?
10. Did anything technical change?

If the answer to several of these is no, the discrepancy may already be explainable.

If they are aligned and the difference remains unexpectedly large, then you have a real investigation.

The Goal Is Not to Make Every Number Match

This is the part I think is easy to miss.

The goal isn't:

Make Shopify, GA4, Google Ads, and Meta show the same number.

The goal is:

Understand what each number means well enough to use it correctly.

A discrepancy can be normal.

A discrepancy can also reveal a tracking problem.

And sometimes the most important finding is that the systems were never measuring the same thing in the first place.

Before asking:

“Which number should I trust?”

ask:

“What question is this number actually answering?”

That is a much better starting point for ecommerce diagnosis.

Related Ecommerce Diagnosis Research

• Why Is My Ecommerce Conversion Rate Dropping? A Step-by-Step Diagnosis
• How to Use AI to Diagnose Ecommerce Conversion Problems With Hypotheses
• When the Numbers Aren't Enough: Why Ecommerce Diagnosis Needs Customer Evidence
• Ecommerce Analytics for Small Businesses: What to Do When You Don't Have Enough Data

Official Documentation

For implementation-specific definitions and tracking behavior, always check the current documentation for the platforms you use.

The Miyeta Approach

Don't pick the number you like.

First understand what each number represents.

Then decide which evidence is appropriate for the question you're actually trying to answer.