Skip to Content

Bringing In-Store Sales into Google Ads: What Changed When We Asked "Why?"

An automotive retailer wanted to connect its physical-store sales to Google Ads. The solution began by questioning the proposed approach and looking more closely at what the business actually needed.
November 16, 2026 by
Bringing In-Store Sales into Google Ads: What Changed When We Asked "Why?"
Ignacio Sbaraglia


An automotive retail chain with physical stores invests consistently in Google Ads campaigns. Yet a significant share of its sales happens offline: a customer sees an ad, compares options, and eventually makes a purchase at a store. Those sales were missing from Google Ads measurement. The platform could track what happened on the website, but it didn't have the information needed to connect physical-store purchases with advertising activity. The goal was to close that gap: send in-store sales to Google Ads as offline conversions, improve campaign measurement, and provide additional information to support optimization.

A request that already came with a solution

The client records its sales in Odoo, its ERP system. Long before this project, a middleware developed and maintained by menze was already retrieving that information and sending it to Zendesk, the client's CRM, where customer information is managed. The initial approach involved connecting Zendesk to Google Ads. Google had suggested using Zapier, a no-code automation platform, as part of the integration. The proposed flow was: Odoo → Zendesk → Zapier → Google Ads.

By the time the request reached menze, the task seemed relatively straightforward: retrieve contact information associated with in-store sales from Zendesk, hash the personal identifiers used for conversion matching, and automatically send the resulting data to Google Ads. This would allow Google Ads to match conversions without receiving those identifiers in plain text. It was a reasonable starting point. Zendesk was already part of the client's operation, Google's recommendation pointed in that direction, and using an established automation platform appeared to offer a faster way forward. We also started with an assumption of our own: that the main challenge was technical and that, once the connection to Google was resolved, the rest would be relatively simple.

The question that changed the direction

While waiting for additional technical guidance from Google, we began reviewing the documentation ourselves, with the support of AI. We discovered that the originally considered integration method was no longer available for new implementations of this type. The alternative required periodically renewing authentication credentials, something Zendesk couldn't handle on its own. That meant our existing middleware would be needed anyway.

And that's when a question changed the direction of the project: If the sale originates in Odoo and our middleware already has access to that information, what does Zendesk add to this particular process?

For this specific objective, the answer was: nothing essential. Instead, keeping Zendesk in the flow introduced another dependency, another component to configure and maintain, and another potential point of failure. The integration could follow a much simpler path: Odoo → existing middleware → Google Ads. This wasn't about replacing Zendesk or questioning its role in the client's operation. The CRM still served its intended purpose. It simply didn't need to act as an intermediary for data that was already available at the source.

We prepared a comparison of both approaches and discussed the options with the client. Once the implications were clear, the client approved the more direct integration from Odoo.








Sometimes, the most valuable technical decision doesn't begin with asking how to connect more tools. It begins with asking which tools we actually need — and why.

When no-code tools help — and when they can become a limitation

Platforms such as Zapier and Make make it possible to connect applications and automate workflows without developing every integration from scratch. In many situations, they're an excellent choice. They can accelerate implementation, reduce development effort, and make automation accessible to teams without extensive technical resources. But they're not necessarily the best fit for every integration.

When a process involves specific authentication requirements, sensitive customer information, data transformations, or particular API constraints, it's important to assess whether a no-code platform can handle those requirements effectively. In this case, the key discovery was that Zendesk couldn't independently manage the authentication required by the available Google integration method. The middleware would still be necessary.

That raised a broader question: why introduce another platform and keep the CRM in the middle when the information was already available in Odoo? Our existing middleware could access the source data and be adapted to meet Google Ads' requirements. Adding more tools would have increased the number of dependencies without providing a clear advantage for the objective we were trying to achieve.

The lesson wasn't that no-code tools are inherently limited or that custom development is always better. It was that the right solution depends on the problem, the requirements, and the architecture already in place. Sometimes, improving an integration means adding new capabilities. Other times, it means recognizing which components aren't needed.

What we built

With the simpler architecture, the work focused on adapting the middleware that was already part of the client's operation. The integration selects only sales from physical stores, since online conversions are already tracked through other Google Ads measurement mechanisms. Before sending the information, the middleware protects personal data, prepares records according to the integration requirements, and filters out records containing errors that could affect processing.

We also implemented a test mode to verify the configuration without sending actual sales data. In addition, we prepared documentation so the client could generate the necessary credentials through its own accounts, maintaining control over its data and permissions. The result was an integration built around existing infrastructure, without introducing additional systems that weren't necessary.

What took longer than expected

The development itself wasn't the most difficult part of the project. Coordinating accounts, permissions, and access across the different systems required considerably more time than anticipated. We also discovered that the server running the middleware needed an upgrade that hadn't been identified during the initial assessment.

Before the first uploads could be accepted, additional configuration was required in the client's Google Ads account. We worked through those adjustments together with the client's team. These challenges reinforced another practical lesson: in integration projects, reviewing infrastructure, access, and dependencies from the beginning is just as important as defining the technical solution.

These challenges also reminded us that there's always room to improve how we approach a project. Looking back, some of these dependencies could have been identified earlier. Acknowledging that helps us refine our processes, ask better questions from the start, and make more informed decisions in future implementations.

Results: what's verified and what's still to be measured

The integration is working. In-store sales are automatically sent from Odoo to Google Ads through the middleware. During one of the initial data runs, 942 in-store sales were successfully processed without errors. That initial volume helps illustrate how much sales information had previously been missing from Google Ads measurement. The integration continues to operate and process new sales. Google also confirmed that the implementation was correctly configured. The client didn't need to modify Zendesk or introduce additional manual tasks.

What this enables. Google Ads can use protected customer contact information to match physical-store sales with advertising activity. This extends measurement beyond online conversions and provides additional signals that can support campaign optimization. However, Google Ads doesn't necessarily provide a sale-by-sale breakdown identifying exactly which advertisement led to each purchase.

What we still can't claim. Successfully transferring data doesn't automatically mean that advertising campaigns are already performing better. Google needs sufficient information to match sales with advertising interactions, and the actual impact on attribution and campaign performance still needs to be evaluated. That distinction matters. A technically successful integration creates new opportunities for measurement and optimization, but it doesn't, by itself, prove a commercial improvement.

What comes next: bringing the physical and digital worlds closer together

The integration achieved its initial objective, but that doesn't mean the challenge of omnichannel measurement is completely solved. There are still limitations in the information Google Ads makes available, particularly when it comes to connecting individual in-store transactions with specific advertising interactions.

After the initial implementation, another opportunity emerged: identifying sales associated with Google Ads within Zendesk itself. That possibility is still being explored and could extend the value of the existing integration without requiring a complete redesign.

This is also how we understand our role as an extended part of the client's team. Our work doesn't end when an initial request has been delivered. It involves understanding the business, questioning assumptions, collaborating on decisions, and identifying opportunities to keep improving.

Humility means recognizing that even a successful solution has limitations. Continuous improvement means using what we've learned to explore what could come next. We'll continue working alongside the client to find ways to narrow the gap between physical-store activity and digital measurement, building on what's already working and reconsidering what can still be improved.

Because sometimes, the most valuable technical decision doesn't begin with asking how to connect more tools. It begins with asking which tools we actually need — and why.

The Real Value of a Digital Experience Analysis Isn't the List of Problems
Why prioritization, not thoroughness, is what turns a technical audit into a decision leaders can act on.