How to triple AC sales in the U.S.
We will tell how to set up GTM and Google Tag for Shopify. Which audiences to use as signals. And how to start Search Shopping from scratch.
Simple goal
We started working with this project at the end of January. The goal was clear and simple to increase sales three to four times while working with the same ROI.
Initially we analyzed the account and saw great prospects for growth.
At the very least, it was immediately obvious that the account was not set up for regular search advertising.
Only companies like Performance Max were set up.
We put together an action plan that included testing and creating new types of ad campaigns.
We also took note of what could be done now without launching new campaigns.
We launched a campaign with video advertising, which is generally not bad and works well for brand awareness, but clearly it is not the final one in the purchase chain.
More than 1000 dollars were spent on it in January and there were no conversions, so we immediately turned it off.
There was also a banner campaign working with creative for the new year, so we also turned it off, because the new year has already passed.
We started to understand the difference between the two Performance Max campaigns.
It turned out that one works purely with feeds, and the second one works with both feeds and regular text and media ads.
In general, this approach has a place, but companies compete with each other and work with approximately the same placement and strategies.
We decided to make the campaign that works with feeds work on the max value strategy, and the second campaign of the top 50 work on max conversion, to separate them by strategy.
In February we have already started to launch new types of campaigns.
As you can see in the first screenshot, we launched DemandGen, updated the English ads, added Spanish ads, launched DemGen product type with feeds for the top categories.
Speaking of types, categories for ads, when we started there was a little bit of a mishmash of those types because there were too many of them, and not all of them were very different from each other. It was just necessary to put in order after we together with the client put in order categories remained 7-8 of which 5-6 categories are launched in advertising, as we then removed from advertising accessories and all categories or goods with a low check, which just does not make sense to advertise, as these categories of related products when buying the main.
We tested another approach, we ran separate companies with higher bids on categories with a higher average check. In general, the results were of this approach, but the payback was quite low - after all, we returned to general campaigns, which contain expensive cheap goods.
We also tested search advertising by brand, we also tested the exclusion of brand queries from performance max, by the way this exclusion worked very badly, on the one hand we saw that there are not so many brand queries and it is logical to exclude them from p-max, but on the other hand pmax lost part of conversions and began to work much worse. At the same time we saw that the costs for branded queries are small. So we removed exceptions from pmax to make it work better. After that its performance went back to the previous level.
Also this company top50 it had a separate page on which were collected the best selling products on the site, which also allowed to increase conversion. Later we made more such pages with top products by category.
We also launched keyword campaigns for competitors, campaigns with dynamic ads without keywords with and without feeds, category ads, video ads like demgen.
All new texts and ads that we tested in regular search companies or in media companies we synchronized on the basis of results with other companies, that is, if somewhere we saw that a good text we copied to other companies and in the main company top 50 performance-max we also copied new ads, created new groups with Spanish language and so on.
February, low season
February is one of the worst months for sales due to seasonality, so many of the campaigns we launched didn't have much effect. However, these tests were useful for understanding the audience and testing and ads - what works and what doesn't, at least in comparison with those campaigns that have already been launched.
A definite plus was the launch of a regular search shopping campaign, because we saw that because of the specifics of pmax, if you look at the analytics on competitors and visibility and compare visibility with them in the merchant center of Google, the ads that go through pmax are not taken into account, and accordingly the merchant center shows that visibility is almost zero, that we are not advertising, and so on. We wanted to fix this, because it is not quite correct and a good signal for the Merchant Center, so that we practically do not advertise products, and it constantly wrote notifications that you have good products, you need to run advertising and so on. Merchant center only takes standard search campaigns into account and when we launched search shopping, we immediately saw that we had increased visibility within the merchant center.
Also the search campaign a week after two also gave good results in general and we decided to develop it as the most correct strategy for an online store.
In March we stopped some of the campaigns, but part of the campaign that was launched at the end of February was still running. Mostly we tested search advertising of the usual type by brand and competitors search worked better and better, although it clearly did not reach pmax top 50 in terms of effectiveness.
Analytics, conversions, signals, audiences
Another significant area of work involved synchronizing analytics. Shopify defaults to using the Last Click attribution model, which usually differs from Google Analytics and Google Ads data. Our goal was to ensure consistent and unified data across all analytics platforms.
To achieve this, we adjusted Google Analytics to use the Last Click attribution model exclusively for paid channels. We made similar adjustments in Google Ads conversion settings, switching from a Data-driven attribution to Last Click to avoid significant data delays. We also observed that most conversions occurred within 30 days after a click, prompting us to change the attribution window from 90 to 30 days.
Despite these changes, discrepancies persisted, prompting us to undertake more in-depth technical adjustments. It turned out that Shopify does not support native integration with Google Tag Manager (GTM). Shopify provides direct integration only with Google Analytics, but without GTM, it is impossible to correctly configure the Conversion Linker. We assumed this was causing the data discrepancies.
We found instructions for installing GTM on Shopify through custom code rather than native integration. This task was technically challenging: we had to thoroughly study manuals and configure necessary variables. Eventually, we successfully installed GTM and configured the Conversion Linker tag. Additionally, we noted that Consent Mode was unnecessary for our store as it operates exclusively in the United States, where such consent isn't mandatory.
Even after these configurations, discrepancies between Shopify and Google Analytics persisted since Shopify directly integrates with Analytics through an app, limiting deeper adjustments.
Consequently, we focused our efforts on Google Ads, meticulously setting up the transmission of conversions and user data. Recently, GTM introduced Enhanced Conversions for user data transmission, which we also implemented. Thus, the GTM tag triggers for all conversions, providing Google Ads with comprehensive user data.
A few weeks post-configuration, we noticed a significant reduction in discrepancies between Google Ads and Shopify data. This development is crucial for business decision-making, as previously we had to compare data across different systems to optimize campaigns. Currently, statistics are highly transparent and update almost instantly, enabling quicker campaign optimization and improved performance.
Another key focus was audience management. Initially, we set up a simple retargeting audience based on interests such as home and garden. We enhanced this by adding audiences of existing customers (email lists), YouTube channel subscribers, and customers who contacted us for servicing. Additionally, we configured dynamic audiences based on Google Analytics customer data.
Using this data, we created look-alike audiences at various levels of granularity: 2.5%, 5%, and 10%. Subsequently, we built a comprehensive audience encompassing look-alike groups, existing customers, email lists, and built-in Google segments ("air conditioner buyers"). We integrated this audience into Performance Max campaigns and partially into display and search campaigns.
Evaluating the immediate impact of adding these audiences was challenging since campaigns already had established data and algorithms. Nevertheless, over several weeks and months, we observed gradual improvements in performance.
In June, we decided to cease active testing and focus on the most effective campaigns: search and Performance Max Top-50. Regularly, we add new ad groups to the Top-50 campaign. For instance, after learning that competitors were out of stock, we created a group targeting competitor site visitors and users searching for similar products. We set up a dedicated page showcasing filtered products and crafted ads highlighting product availability.
We noticed that launching new groups as separate campaigns led to internal competition and fragmented traffic. Hence, we decided to integrate new groups exclusively into already successful campaigns.
Additionally, we analyzed performance metrics and removed poorly performing product categories from the Top-50 campaign. In the search campaign, we excluded several unprofitable audiences to enhance overall effectiveness.
A notable complexity arises as the search campaign and Performance Max frequently compete for the same placements, especially in product ads. Google previously prioritized Performance Max, but recently shifted its approach to favor the campaign with the higher CTR and better signals.
Nevertheless, the regular search campaign still includes a priority setting whose precise impact on traffic distribution alongside Performance Max remains unclear. Therefore, this issue requires further analysis and remains open-ended for us.
At the moment we have a low priority for a regular search campaign. We plan to test changing the priority to medium or high, but it is important not to degrade the Performance Max campaign.
300% Revenue Increase
ROAS 13.00
Avg. check
+22%
Stable CPC $0.91
In general, the main goal was achieved - we tripled our revenue, while maintaining a return on investment in advertising. And this goal was accomplished ahead of schedule - in 5 months, without waiting for the end of June.
Ключові фактори успіху
1. Correctly setting up and communicating conversions to Google Ads, which made the data as consistent as possible with Shopify's metrics.
2. proactive work with audiences and signals that allowed Google to better identify our target audience.
3. Switching to Target ROAS and Maximize Conversion Value bid assignment strategies, which noticeably increased our average check.
4. Switching from manual bid management to automated strategies, which became a necessity in the current market conditions.
All these changes in combination allowed us to achieve the desired result in just 5 months. Good luck to everyone! If you have any questions, please do not hesitate to contact us.
