We understand that seeing unattributed data in your reports can raise questions.
This article will help you understand the possible causes.
Unattributed data used to be labelled as Direct. We seperated this to give insight in real direct conversions vs unattributed conversions
In general, the main reason for unattributed data is website visits where (analytics) cookies have not been accepted, preventing our pixel from triggering and measuring orders.
What is unattributed data?
Billy Grace receives data from two sources:
From the pixel that is installed on your website
From your server (e-commerce or CRM platform such as Shopify, Magento, Hubspot, Pipedrive or Custom platform)
Custom events and their server counterparts (like purchases and orders from your e-commerce platform) sent through the pixel will only appear in your dashboard and be credited if they match an ID from your server records. This ensures that events are accurately counted and allows us to use the most up-to-date information for each event. For example, this approach helps us track refunds and calculate important metrics such as profit.
If we have order data from the e-commerce platform, but do not have Pixel Data, we show the order/event in channel Performance, but it will be labeled as "Unattributed".
What causes unattributed data?
Keep in mind that "unattributed data" often results from missing pixel data. However, this doesn't always point to a tracking or setup issue.
There are many types of orders that don't trigger web tracking pixels, so missing pixel data or unattributed data is typically expected.
The most common cause of unattributed events / orders is that the event occurred in a location where your pixel is not running. Common examples include:
event / order imports
app events / orders
manually created events / orders.
Any order that does not trigger the Billy Pixel or Purchase event will be Unattributed.
Are there ways to lower the amount of unattributed orders?
In some e-commerce platforms, there are ways to improve tracking and attribution by sending more than one event to Billy Grace. Below an overview.
E-commerce platform | Improvements available? | Solution |
Shopify | Yes | With the native Shopify integration, we already track as much as possible. |
Magento (Adobe Commerce) | Yes | |
Lightspeed (c-series) | Yes | Add a purchase_started event |
Medusa | Yes | Add a purchase_started event |
Salesforce Commerce Cloud | No | No optimization possible |
WooCommerce | No | No optimization possible |
Attribution on Unattributed orders
By using marketing mix modelling techniques, our Unified Marketing Measurement (UMM) model will try to find effects between paid marketing channels and the unattributed orders. If an effect is found, you will see that Unattributed orders are lower when UMM is selected as attribution model, this means that these orders are attributed to paid channels.
How does this work?
We model based on total events, and then multiply the total effect by the share of untracked events to estimate the impact on untracked data. We also account for the possibility of a longer delay between impressions and events, as the time between views and purchases tends to be longer than between views and clicks.
The requirement for UMM to model on Unattributed orders is at least 30 days of 50 unattributed orders per day. Lastly, we always perform a quality check. If the models don’t meet our standards, you will not see that Unattributed orders are lower when UMM is selected.
If after you have set up the additional event to improve the attribution, you still notice a high number of unattributed events, it might be worthwhile to debug your pixel installation. Learn more about this here.