This is a leading home decor brand in the United States, using Shopify Plus. In discussions with data analysts, we found that this brand has tried almost all multi-touch attribution analytics software available on the market. The brand's marketing channels are quite complex, mainly including search ads, Klaviyo, Impact, ShareASale, and social ads.
Order processing is equally complex, as the target audience includes those aged 40-50, with many visitors preferring to establish trust by phone and place orders via phone after visiting the website. This makes tracking phone orders difficult, and the orders often include numerous modifications. Currently, they use CallTrackingMetrics to track the traffic sources of phone orders.
Due to these complexities, analysts find it extremely difficult to analyze the source of each order and ROAS. The data analyst says:
The data on GA4 has never matched the order amounts on Shopify.😱
Furthermore, since the brand's average order value is over $1,000, meaning the sales cycle takes more than a month, only multi-touch attribution models can effectively evaluate the performance of each channel. So it took us nearly 3 months to successfully deploy Attribuly, which has become an essential daily tool for the entire marketing team.
The data analyst says:"
I'm on Attribuly all the time now.😘😀
We mainly help the brand grow through attribution analysis and server-side tracking. Some milestones we have achieved.
1. 99% order accuracy in order quantity and value.
2. Sync data from various platforms every 15 minutes.
3. Including custom channel groups, created over 30 channels.
4. Integrated and attributed all channels including Google, Facebook, Microsoft ads, Criteo, Impact Radius, ShareASale, Klaviyo, establishing real-time ROAS tracking capabilities.
4. Enhanced conversions were enabled for Google and Bing ads, providing Google with more accurate conversion data. The brand has abandoned Pixel tracking and relies entirely on stable server-side tracking. And the cost per conversion has dropped 17%.
5. Enabled Klaviyo server-side tracking, doubling the revenue from corresponding flows.
6. Server-side tracking for all other touchpoints.
Now let's analyze step by step how to achieve all this.
Getting accurate attribution reports requires effective tracking of the customer journey. We have done several things
1. Using Shopify Web Pixels. As one of the earliest companies to use this innovative technology, Attribuly has grown together with the Shopify team to build the best user behavior collection tool in the Shopify ecosystem.
2. Extensive use of API integration and "Auto Tracking" technology. Traffic source tracking mainly relies on UTM parameters, but manual addition is error-prone. Our innovative auto-tracking can automatically add tracking templates to ads, greatly improving the user experience
3. Attribuly supports the checkout extensibility, so it can track the completion of user processes within the checkout page.
Reseller orders are often repetitive and of a larger amount. Therefore, when analyzing ROAS, these orders will be included in the attribution analysis, which will lead to an inflation of ROAS across various channels. It is very difficult to exclude these orders on platforms like GA4, but it will be very easy through customer exclusion.
Since customers can place orders by phone or modify their orders via phone after placing them online, Shopify creates a new order for modified orders, resulting in 2 orders in the analytics tools. Based on Shopify's sales channel data, we excluded edit orders created by Cleverific, thereby improving attribution accuracy.
For Meta and Bing, we've upgraded the ad platform synchronization from the basic once every 24 hours to once every 15 minutes, and for Google Ads, we've improved it to once every 30 minutes.
This ensures that the cost data seen by the team is almost consistent with the ads manager, enabling same-day data analysis. For order attribution, Attribuly can provide attribution reports within 5 minutes after an order, giving brands near real-time ROAS analysis capabilities.
When sales increase, marketing channels will increase significantly at the same time, and different team members are responsible for different channels.
Attribuly's traffic source has its own set of rules, but it may not be suitable for every brand. Like Google's channel group, Attribuly also supports brands to customize their own channel group, which takes the entire order analysis to a new level.
Currently, Attribuly supports creating your own custom channel based on three dimensions: source, medium, and campaign.
We send daily store performance emails every morning, including total sales, sales breakdown by channel, and ROAS.
To help the marketing team better analyze, we also include the average ROAS for the past 7 days. The daily email ensures the team shares an accurate analysis report to plan upcoming work.
Given the high average order value of this brand, we made the following optimizations to the position-based attribution model:
When we completed all of attribution setup, the entire team, from the CEO to every marketer, for the first time had a complete single source of truth. We knew where every budget was spent, how effective it was. The first thing we do at work is have coffee, the second is to open the daily email and think about how to adjust next.
Now we have sufficiently accurate first-party data, we need to feed the data back to different marketing platforms to optimize the targeting and performance. The data analyst says:
We are increasingly using ads like Google performance max, and accurate first-party data is crucial for ad optimization.
Pixel tracking is about to be replaced by server-side tracking. Fortunately, most advertising platforms already support server-side postback. Good server-side tracking software generally needs to achieve:
1. Attribute orders to a specific channel.
2. Get as many identifiers as possible, because most advertising platforms need these identifiers to complete deduplication, such as click id or email.
This brand previously used a well-known server-side tracking software in the Shopify ecosystem, unfortunately, this software could only postback data related to orders and did not have the complete customer journey for each order.
So the brand could only manually configure various parameters to filter out some orders, but because too many orders unrelated to the destination advertising platform were posted back, it led to a decrease in quality score and reduced conversion matching rate.
Therefore, the brand decided to migrate all server-side feedback to Attribuly
The Enhanced Conversion uses first-party data such as customer emails and phone numbers to improve conversion tracking accuracy and bidding performance.
After enabling this feature, we were delighted to see that enhanced conversion in Google goals became fully optimized. During recent promotions, Cost per conversion decreased by 17%.
In Bing, the client no longer needs to manage complex UET tags, as conversion tracking is completed through server-side tracking alone. Events that were difficult to track with UET, such as form submissions, can now display data accurately.
The Facebook Conversion Feed replaced the previous Facebook channel's postback function with enriched customer data, using Facebook Conversion API (CAPI) + custom pixel method. This resulted in a high-quality score of 9.1/10 in Facebook events manager.
We migrated both Browse abandonment and cart abandonment server-side flows to Attribuly. Compared to the Klaviyo original pixel, Attribuly reported three times more unique profiles triggering events for "Viewed product"
and reported around 2 times unique profiles for "Added to Cart" event.
The "Viewed product" flow generated twice the revenue compared to existing flows.
Supplement orders that ShareASale browser pixel fails to track and post them back through server-side tracking.
Supplement orders that Impact browser pixel fails to track and post them back through server-side tracking.
After we use server-side tracking, we can completely stop relying on the unstable browser pixels.