OPTIMISIUMCOMMERCE
All insights

Customers Don’t Follow Funnels: How to Read Ecommerce Attribution and Global ROAS

Learn why Meta, Google, GA4, and store data report different results—and how to interpret attribution, MER, and global ROAS together.

01

Customers Don’t Follow Funnels: How to Read Ecommerce Attribution and Global ROAS

A customer sees an Instagram ad. It doesn't click. Two days later, he searches for the brand on Google. Reviews a product page. Then he signs up for email. Three days later, he asks for stock on WhatsApp. Finally, he/she goes directly to the site and buys.

Who owns this sale?

The commodity can make a profit for itself. Google Ads may show a different conversion. GA4 can highlight another channel. The email platform can attribute the last message to revenue. The store only knows that the order has been fulfilled.

The problem is not that one of the tools is necessarily “wrong.” Each vehicle has a different observation area, model, window and identity analysis method.

So tying the advertising decision to a single platform's ROAS number means forcing the customer's actual journey into a single line.

QUICK ANSWER

Attribution in e-commerce is the method that determines to what extent the sale will be linked to which marketing theme. Platform metrics may differ from each other because they do not use the same source and the same model. For a sound decision, platform ROAS, store revenue, total marketing spend, new customer rate, margin and global ROAS/MER should be evaluated together. Before closing a channel, the secondary impact of the customer on other channels should be examined.

IN 60 SECONDS

The essentials first.

What you need to know before acting on this guide.

  • Customers do not follow the planned funnel in a linear manner.
  • Meta, Google, GA4, Klaviyo and store reports may use different attribution models.
  • You cannot obtain real total revenue by adding up the revenues associated with the platforms.
  • Global ROAS or MER shows the high-level relationship between total revenue and total marketing spend.
  • Platform ROAS is still valuable for creative and campaign optimization; However, the company alone does not make the decision.
  • Channel decisions should be made with stock, margin, new customer rate, repurchase and measurement limitations.
04

What is Attribution?

Attribution is the model for which contact or channel a conversion will be attributed to.

Simple models:

  • First click
  • Last click
  • Non-direct last click
  • Post-view conversion
  • Data-driven attribution
  • Platform specific attribution

Each model answers a different question.

For example, “what was the last measurable click before purchase?” and “Which channel created the demand first?” It is not the same question.

05

Why are platform reports different?

01

Different attribution windows

One platform may consider seven days after the click, another system may consider a shorter or different window.

02

Post-view conversion

Even if the user does not click on the ad, they may have purchased it after seeing the ad. Some platforms include this in their performance.

03

ID and device difference

The customer can see the ad on the phone and buy it on the desktop. Not every system can combine this journey with the same accuracy.

04

Cookie and permission limitations

Browser, device, consent, and data sharing conditions change the scope of measurement.

05

Different conversion definitions

One system may use the “purchase” event, another may use the store order after payment is completed. Cancellation or refund transactions may also be reflected differently.

06

Biggest mistake: Collecting platform revenues

If Meta reports $100,000 in revenue, Google reports $80,000 in revenue, and the email platform reports $40,000 in revenue, it cannot be concluded that the actual revenue is $220,000.

The same order may be associated by more than one platform.

The financial reference point for actual orders and revenue should be the store or ordering system. Advertising and lifecycle platforms are used to analyze which contacts this income is related to.

07

What are Global ROAS and MER?

In practice, concepts can be used with different names. The basic rate is:

Total store revenue / total ad spend

Example:

  • Total store revenue: 1.200.000 TL
  • Total advertising expenditure: 300,000 TL
  • Blended MER: 4.0

This ratio shows the company's overall revenue efficiency versus wage growth. But it does not mean profitability.

Because the following variables are still missing:

  • product cost,
  • shipping,
  • returns,
  • agency and production cost,
  • discount,
  • payment commission,
  • Difference between new and past customers.
08

When is Platform ROAS useful?

Platform ROAS is not worthless. It is a strong operating signal for:

  • campaign comparison,
  • creative analysis,
  • mass or product group analysis,
  • initial signal for budget distribution,
  • in-platform optimization.

However, this sentence is risky:

ROAS is high, increase budget.

The following should be considered first:

  • Is the stock of the product sufficient?
  • Is the margin still healthy after the campaign discount?
  • Is the revenue coming from new customers or past customers who will buy anyway?
  • Is the number of conversions sufficient for the decision?
  • Does the result consist of several large orders?
  • Is there a sudden change in measurement?
09

How do you read the customer journey more accurately?

01

Determine single source financial truth

Identify the primary source for orders, net revenue, returns, and customer information. This is usually the commerce platform.

02

Show source for every metric

Instead of typing “ROAS 3.2” in the Dashboard:

  • Source: Meta Ads
  • Attribution window
  • Date range
  • Currency
  • Last sync

should be shown together.

03

Separate new and returning customer

The same ROAS can have different growth quality.

  • Is the campaign gaining new customers?
  • Is he reactivating the past customer?
  • Does discounting create addiction?
04

Track brand search and direct traffic

A campaign can increase brand search rather than directly generate clicks. Google brand search or direct traffic may also drop when the channel is closed.

05

Find the point where funnel first broke

Run the analysis from front to back:

  1. 01
    Are there any screenings?
  2. 02
    Is there interaction?
  3. 03
    Are there any clicks?
  4. 04
    Is the product page working?
  5. 05
    How is the cart and checkout transition?
  6. 06
    Is the sale taking place?
  7. 07
    Does the customer come again?

If the problem is on the product page, turning off creative may be the wrong intervention. If the problem is with creative, increasing the discount will hide the problem.

06

Use controlled testing

When possible:

  • holdout group,
  • regional testing,
  • time based comparison,
  • campaign on/off test,
  • customer segment control group

Try to measure the actual additional contribution using

10

Example: ROAS dropped, what should you do?

In one Meta campaign, ROAS dropped from 4.1 to 2.9.

Just before closing:

  • Has CPM changed?
  • Has CTR fallen?
  • Has the frequency increased?
  • Has product page conversion dropped?
  • Has AOV changed?
  • Is the discount over?
  • Is there a stock or variant issue?
  • Is there a secondary effect on Google or direct sales?
  • Has the share of new customers changed?

For example, if CTR has decreased, frequency has increased, and landing page conversion has remained constant, creative fatigue is more likely. If landing page conversions have decreased but advertising metrics are stable, the problem may be on the site side.

11

How does Optimisium help?

Rather than forcing platform numbers to a single “correct” number, Optimisium Decide shows sources in context.

System:

  • Collects meta, Google, store and analytics data in a common view,
  • keeps the attribution source and window visible,
  • shows platform performance separately with global ROAS/MER,
  • adds new customer, stock, margin and repurchase signals to the decision,
  • creates tasks that detect the first breakpoint,
  • submits the proposed action for approval within the limits of evidence and policy.

So the team just asked “which number dropped?” not the question "at what point should we intervene?" answers the question.

Set Meeting

SOURCE NOTES
READ NEXT

Continue from here.

FAQ

Questions, answered.

01Is global ROAS the same as platform ROAS?+

No. Global ROAS/MER shows the relationship between total store revenue and total ad spend. Platform ROAS, on the other hand, uses the revenue attributed by the platform according to its own attribution model.

02Which platform's revenue is correct?+

The store and payment/ordering system should be the main reference for the financial total. Marketing platforms provide contact and optimization signals.

03Should the campaign always scale if ROAS is high?+

No. Inventory, margin, conversion volume, new customer share and measurement quality should be controlled.

04Can Attribution be completely solved?+

Perfect and complete attribution is not possible in most businesses. But better decisions can be made by clearly showing sources, models and limitations.

05If MER drops, does that mean advertising is bad?+

Not necessary. Changes in season, price, organic traffic, repeat purchases, promotions and metrics can also have an impact. The breakpoint should be examined along the chain.

Book a Call