Avoid Conversion Tracking Issues In Shopping Campaigns

Avoid Conversion Tracking Issues In Shopping Campaigns

For eCommerce Shopping ad campaigns, an accurate conversion tracking setup is critical because it generates the metrics used to calculate two of the most important measures of paid search performance: Net Advertising Contribution (NAC) and return on ad spend (ROAS).

Said another way, accurate measurement of conversions and conversion value provides the data that Google Ads automated bidding systems use to make bidding and traffic-allocation decisions.

In our work with client accounts, we’ve found that the majority of incoming eCommerce accounts have some kind of problem with conversion tracking.

Sometimes conversions are missing. Sometimes transactions are counted more than once. Sometimes sales are recorded but the wrong values are being passed. In other cases, several different conversion actions are feeding conflicting information into automated bidding. In other cases, the wrong products are being advertised – for example, the bundle pack issue described below.

Because conversion tracking accuracy is fundamental to performance, we identify and correct these problems early on.

Why Accurate Conversion Tracking Matters

It’s important for advertisers to comprehend that conversion tracking isn’t simply a reporting function! Conversion metadata is the foundation for automated bidding. When that information accurately represents actual business results, automated bidding can make increasingly informed decisions about which searches, users, products and auctions deserve higher or lower bids.

If conversion values are inaccurate, transactions are duplicated, important purchases are missing, or the wrong conversion actions are included in bidding, the algorithm is learning from distorted information.

This can create a damaging feedback loop:

Bad conversion data → poor bidding decisions → lower-quality traffic allocation → poorer business results.

The advertiser may conclude that Shopping, Performance Max or automated bidding isn’t working when the underlying problem is actually measurement.

Conversion Tracking Is More Than Counting Orders

One of the first questions we ask when auditing an eCommerce account is whether purchases are being tracked correctly.

But simply seeing conversions in the Google Ads interface doesn’t answer that question.

A thorough conversion tracking review should also determine:

• Are all legitimate purchases being recorded?
• Is each transaction being counted only once?
• Is the correct transaction value being passed?
• Are transaction IDs being passed correctly?
• Are the appropriate conversion actions designated as Primary?
• Are older or duplicate conversion actions also influencing bidding?
• Does Google Ads conversion data reasonably reconcile with the eCommerce platform and analytics data?
• Are product identifiers being passed consistently between the store, Merchant Center and Google Ads?

The objective isn’t merely to make the Conversions column contain numbers. The objective is to give the bidding system an accurate representation of the business result we want it to optimize.

Common Conversion Tracking Problems We Find

Conversion tracking problems in established eCommerce accounts can be surprisingly common.

Some of the problems we encounter include duplicate purchase tracking, missing transactions, incorrect revenue values, obsolete conversion actions that were never removed, and multiple tracking implementations operating simultaneously.

This last problem deserves particular attention.

Over time, an account may have had conversion tracking installed manually, through Google Tag Manager, through Google Analytics, and later through an eCommerce platform integration. Unless the old implementation is carefully removed or changed, several systems can end up recording the same purchase.

A conversion tracking setup should therefore be audited as a complete system rather than simply checking whether an individual tag fires. This is also why PPC campaign performance tracking needs to be considered together with conversion tracking rather than as a separate reporting exercise.

The Google Tag, Google Ads Tags and Google Tag Manager

These terms are easy to confuse because they describe related but different parts of Google’s measurement infrastructure.

The Google tag is Google’s sitewide tagging framework. It evolved from what was previously called the global site tag, or gtag.js. The Google tag can support Google Ads and Google Analytics measurement and is generally loaded across the website.

A specific conversion event, such as a completed purchase, then provides the additional information needed to record that particular action.

Google Tag Manager (GTM) serves a different purpose. GTM is a tag-management system that allows marketers and developers to deploy and control Google tags as well as many third-party measurement tags without repeatedly modifying the website’s source code.

For many conventional websites, GTM remains extremely useful. It gives an experienced practitioner centralized control over measurement and makes it easier to implement and troubleshoot sophisticated tracking configurations.

But eCommerce platforms can introduce another consideration.

Shopify Requires Special Consideration

Shopify has changed the way tracking can be implemented, particularly around checkout, Thank You and Order Status pages.

For Shopify stores, Google currently recommends using the official Google & YouTube app for Google Ads conversion measurement rather than attempting to reproduce a traditional Google tagging implementation through Shopify’s custom-pixel environment.

This is an important distinction because a setup that worked properly on Shopify several years ago may no longer be the implementation we would recommend today.

Google has specifically cautioned that running a Google tag or Google Tag Manager container through Shopify’s custom-pixel feature can create measurement limitations. Shopify’s newer checkout architecture has also replaced some of the mechanisms previously used to install tracking code.

That does not mean that Google Tag Manager itself is obsolete or undesirable. It means the appropriate implementation depends upon the platform. For a Shopify store, we generally want to understand exactly what the Google & YouTube app, Shopify, GA4 and any existing GTM implementation are already doing before adding another tracking mechanism.

Layering one more tag onto an already complicated Shopify installation can make the problem worse instead of better.

Primary and Secondary Conversion Actions

Not every useful conversion action necessarily belongs in an automated bidding strategy.

An eCommerce account might track:

• Purchases
• Add-to-cart actions
• Begin-checkout actions
• Newsletter registrations
• Phone calls
• Other engagement events

These can all provide useful information, but they don’t all represent the same business value.

For most Shopping campaigns, the completed purchase is the primary business objective. If lower-value actions are inadvertently included as Primary conversion actions and used for bidding, Google may begin optimizing toward activity rather than actual sales.

This is one reason we review not only whether conversion actions exist, but also which actions are actually included in campaign bidding objectives.

Using Conversion Values in Shopping Campaigns

For an eCommerce purchase, the conversion value will often correspond to the value of the transaction.

This enables Google Ads to distinguish between, for example, a $40 order and a $400 order. That distinction is essential for value-based bidding strategies.

Revenue, however, isn’t always the same thing as economic value.

Two products selling for $200 may have dramatically different product costs and gross margins. An advertiser that optimizes entirely around revenue or ROAS may therefore spend heavily promoting the product producing less actual profit.

For this reason, sophisticated eCommerce measurement may incorporate cost of goods sold, gross margin or other business-value information into the broader optimization process.

We are cautious, however, about simply replacing transaction revenue with gross margin as the standard purchase value. Preserving actual sales revenue makes it easier to reconcile Google Ads against Shopify, GA4 and backend financial reporting. Margin and profitability data can then be incorporated where appropriate through more advanced value-based bidding and reporting strategies.

ROAS and Net Advertising Contribution

ROAS is useful, but it doesn’t tell the whole story.

The basic calculation is:

ROAS = Revenue ÷ Advertising Cost

That tells us how efficiently advertising generated revenue.

Net Advertising Contribution looks at the problem from another direction: how many actual dollars did advertising contribute after the relevant costs are considered?

This distinction becomes especially important when comparing products with different margins or when deciding whether increasing advertising spend at a somewhat lower ROAS might actually produce more total profit.

Accurate conversion values are therefore important not only for Google Ads automation, but also for making sound business decisions about the advertising program itself.

Product IDs Are Part of the Measurement System

One area that is frequently overlooked is product identification.

An eCommerce product may have an identifier in Shopify, an ID submitted to Google Merchant Center, and product-identification information passed through website events to Google Ads.

These identifiers need to be mapped carefully and consistently.

The Merchant Center id attribute uniquely identifies a product within the product data source. Product IDs are also used throughout the advertising and measurement ecosystem to associate activity and performance with the appropriate item.

Changing an ID can therefore have consequences that go far beyond cosmetic feed organization. This is one reason product feed optimization needs to consider the stability and consistency of the underlying product data, not simply titles and descriptions.

What Happens When Product IDs Change

We recently encountered an account in which the same underlying product had inadvertently been submitted and advertised under as many as four different product IDs.

From a business perspective, it was still the same product.

From Google’s perspective, however, those identifiers represented different items.

The result was fragmentation of the product’s performance history. Instead of continuing to build upon the conversion history associated with an established product, traffic and conversion data became distributed among several different identifiers.

At least three versions were effectively being treated as new products.

This coincided with a major deterioration in sales performance. The automated bidding system had lost much of the continuity of the historical product-level data it had previously been able to use.

This illustrates why feed work and conversion tracking should not be treated as completely separate disciplines. The product feed, product identifiers, website event data, Merchant Center and Google Ads measurement system all have to agree about what product is being advertised.

Consistency matters.

Shopify Product IDs, Merchant Center IDs and Google Ads Item IDs

In a typical Shopify implementation, the store has its own underlying product and variant identifiers. The product feed then submits an id for each offer to Merchant Center.

For measurement applications that send an item_id, the identifier needs to correspond correctly with the relevant Merchant Center product data.

A mismatch can prevent Google from correctly associating website behavior with the corresponding catalog item.

This is particularly important when feeds are rebuilt, feed-management software is changed, a Shopify integration is replaced, or products are migrated from one Merchant Center data source to another.

An apparently harmless change in the way IDs are generated can cause Google to see an established catalog as a collection of new products.

Before changing product IDs in an established, successful Shopping account, we want to know exactly why the change is necessary and what historical continuity may be affected.

Multipacks Can Create a Different Kind of Shopping Problem

Google uses the term multipack for a merchant-created grouping of several identical products sold together as one product.

For example, if a merchant combines six identical bottles and sells them as a single six-pack, the Merchant Center multipack attribute can identify that offer appropriately.

A bundle is different. In Google terminology, a bundle generally combines a main product with different complementary products.

These distinctions are important for feed accuracy, but there is another practical advertising consideration that is easy to overlook.

Just because a multipack can be advertised does not necessarily mean that it should be.

The Price-Comparison Problem With Multipacks

Shopping ads are highly visual and highly comparative.

Imagine that several competing advertisers are selling the same general type of product:

Competitor A — $29
Competitor B — $32
Your 3-Pack — $79
Competitor C — $27

Your price per unit may actually be excellent. Your three-pack may even be the best economic value on the page.

But that isn’t necessarily what the shopper sees.

Someone scanning a Shopping carousel may simply see $79 versus $29.

The result can be a dramatically lower clickthrough or conversion rate even though the underlying offer is competitively priced on a per-unit basis.

We have seen this occur in real client accounts.

This is why feed optimization cannot consist simply of asking whether a product is eligible to advertise. We also need to consider how the offer will appear alongside competing products and how a prospective buyer is likely to interpret it in the few seconds available.

In some situations, advertising the single-unit product while allowing the website to upsell the customer into a three-pack or six-pack can be more effective than advertising every available pack size directly through Shopping.

More products in the feed does not automatically mean more sales.

This same principle becomes increasingly important in highly automated campaigns. PMax product feed optimization, for example, requires looking at individual SKU performance rather than assuming every eligible product deserves equal exposure.

Transaction IDs and Duplicate Conversions

Transaction IDs provide another important safeguard in eCommerce conversion tracking.

Each completed order should have a unique transaction identifier. This helps measurement systems distinguish one legitimate transaction from repeated instances of the same purchase event.

Without proper transaction identification and deduplication, actions such as reloading an order confirmation page can potentially create inaccurate measurement depending upon the implementation.

When reviewing an existing account, we therefore want to confirm not only that purchase events fire but also that transaction IDs and transaction values are being passed correctly.

Google Ads, GA4 and Shopify Will Not Always Match Exactly

Another common source of concern occurs when an advertiser compares Google Ads, Google Analytics and Shopify and discovers that the numbers aren’t identical.

Some differences are normal. These two systems answer somewhat different questions and can use different attribution methods, reporting windows, transaction timing, consent and modeling logic.

The objective is not necessarily to force every system to display precisely the same number.

Instead, we want differences to be understood, reasonably consistent and explainable.

Large unexplained discrepancies deserve investigation. Small and predictable differences do not necessarily indicate broken tracking.

Google Analytics is particularly useful when testing and diagnosing paid traffic and conversion behavior. We use Google Analytics 4 not only for conversion tracking review, but also for understanding engagement and paid traffic flow through the site.

Why Fractional Conversions Appear in eCommerce Reporting

Advertisers are sometimes concerned when Google Ads reports conversion counts such as 3.4 or 7.6. That doesn’t mean Google recorded 3.4 orders.

A sale can receive attributed conversion credit across multiple advertising interactions. Depending upon the attribution model and reporting context, portions of conversion credit can therefore appear as fractional conversions.

Fractional conversions are not, by themselves, evidence of a tracking problem.

They may simply reflect how Google Ads assigned credit among the interactions that contributed to the eventual sale.

Auditing Tracking Before Optimizing the Campaign

There is little value in aggressively optimizing a Shopping campaign against data we don’t trust.

Before making major bidding or structural changes to an inherited eCommerce account, we prefer to understand:

• What constitutes a conversion?
• Where is that conversion being generated?
• Which system is sending it to Google Ads?
• Is the transaction being counted once?
• Is the correct value being passed?
• Are product IDs consistent?
• Which conversions are actually being used by automated bidding?
• Does the resulting data reasonably reconcile with actual store sales?

Once those questions have good answers, campaign optimization becomes much more meaningful.

The measurement system also needs to support the way the campaign itself is organized. A well-designed Shopping campaign structure can separate products into useful product groups and ad groups, but those structural decisions become much less valuable if the underlying conversion and product data cannot be trusted.

Key Takeaways

Accurate conversion tracking is a crucial foundation for high-performing Shopping campaigns.

But modern eCommerce measurement involves much more than installing a conversion tag.

The eCommerce platform, Google tag, Google Tag Manager where appropriate, Google Analytics, Merchant Center product feed, product identifiers, conversion actions and Google Ads bidding objectives all form parts of an interconnected measurement system.

Problems in any one of these areas can contaminate the data being supplied to automated bidding.

And because automated bidding increasingly determines where advertising dollars are spent, bad measurement can eventually become bad media buying.

Our experience auditing existing eCommerce accounts has made us increasingly reluctant to treat tracking and feed configuration as background technical details.

They are part of campaign optimization itself.

Related Shopping Campaign and Tracking Resources

Related topics include Shopping campaigns, PPC campaign management, web analytics, product feeds, and eCommerce advertising results.

For more context, see our related blog articles on automated bidding, conversion tracking, and conversion tracking value models.