Conversion data is the lifeblood of the digital advertising ecosystem. Conversion tracking captures website interactions or browser events, such as product views, additions to a cart, checkouts and more. In turn, conversion data not only powers ad reporting but is also the primary engine for follow-up marketing actions, including optimizing future ads, ML audience optimization, nurturing prospects and dynamic ad targeting for potential customers.
Traditionally, conversion tracking relied on client-side web pixels, which have fallen out of favor in response to new consumer privacy laws, ad blockers and other developments. Even more sweeping changes are on the near horizon.
In response, major social media platforms are rolling out conversions APIs as a way to share events server-side. Conversions (or server-side) APIs are much more powerful because the integration is extremely reliable and unencumbered by browser restrictions. The downside is that they’re much more time-intensive to set up than simply embedding a web pixel into your website.
Fortunately, a robust, modern data stack can feed event streams and platform data into your data warehouse, providing central access to conversion events. Combined with reverse ETL to move those data models to a conversions API, you can make full use of these advertising endpoints with modest effort.
What is a conversions API?
In the past, the primary way to share conversion data was through a web pixel that leveraged third-party cookies to directly share data back to a social media platform from browser pixel events. However, recent trends in consumer privacy laws, web browsers, mobile IDs, third-party cookies and ad blockers have dramatically limited the ability to collect conversion data and customer actions from web pixels. In a recent article, Facebook (now Meta) formally acknowledged the extent of this problem, estimating that it underreports iOS web conversions by approximately 15%. In turn, advertisers should turn to server-side tracking and utilize the Meta Conversions API (Meta CAPI) to capture lost event data as a replacement to its tracking pixel.
A conversions API (or server-side API) is similar to a web pixel but it is much more powerful because all of the data is shared server-side and managed directly by you. This conversion tracking method is more flexible because it is not just limited to website interactions captured through a pixel. Conversions APIs can receive data from offline conversions and events or anywhere within the customer journey. Server-side conversion tracking is also much more reliable because it is unencumbered by network issues like ad blockers or tracking restrictions imposed across web browsers. For example, in the case of the Meta Conversions API, you can choose exactly what user activity and user behavior you share with Facebook to optimize your ad spend at an extremely granular level.
Nearly all major advertising platforms have been equally affected by privacy changes and are introducing their own flavor of conversions APIs — Google Enhanced Conversions, SnapChat Conversions API and TikTok Events. Pinterest, LinkedIn, Twitter and Amazon are expected to release conversions APIs sometime in 2022.
Getting started with conversions APIs
The downside of server-side conversion tracking or conversions APIs lies in the time-consuming engineering tasks required to adopt them. Web pixels have an advantage in that you simply have to embed a piece of code on top of your front-end codebase.
Fortunately, the right architecture and implementation choices can drastically reduce the development effort and setup process required to improve your data quality and overall ad performance. In the remainder of this article, we'll cover our two key recommendations:
- The data warehouse should be your source of conversion data
- Reverse ETL is the easiest way to sync this conversion data with conversions APIs
How to approach data integration for conversion events
Conversions consist of more than interactions on your website. There are conversions that a typical web pixel cannot capture, like offline purchases and post-purchase events (product returns, loan approvals, etc.) from other ETL pipelines. Top-of-funnel and anonymous events like page views and add-to-cart require a first-party behavioral tracking tool to be added to your site, store or app.
Event tracking is often distributed across multiple point-to-point solutions used to power your business:
- Lead events in CRM
- Utilization events in digital product
- Engagement events in marketing
- Purchase events in e-commerce
These solutions often track key events like:
- Purchases and donations
- Product trials and subscriptions
- Lead and newsletter sign-ups
Connecting multiple point-to-point solutions to conversions APIs has several disadvantages:
- Development costs - Connecting multiple point-to-point solutions requires many brittle, custom-built data pipelines and increases development and maintenance costs (see below). It is worth emphasizing that this is a tremendous investment of time, money and labor.
- Incomplete data - Some important conversion events may occur in independent or third-party systems that you do not have direct access to.
- Lower match rates - When pushing events directly from your application servers, you will only have access to the data available within the context of a specific user action.
Instead of individually connecting point-to-point solutions to conversions APIs, the key is to first centralize all business data, including from point-to-point solutions, in a data warehouse using ELT.
A data warehouse serves as a collection point for data from all business sources, including event streams, CRMs, e-commerce and other common business applications and operational databases. Aggregating all data results in a unified, 360-degree view of the customer and associated events. As detailed in Facebook’s Best Practices for Conversions API, the more customer information provided via the API the higher the associated Event Match Quality Score – i.e. correctly attributing actions to a customer. A higher score increases the rate of attribution and ad delivery optimization and drives down the cost per action.
An automated data pipeline like Fivetran can move conversion events to your data warehouse with very little development effort and very short time to value. The process takes little more than supplying credentials to each source, choosing which tables and columns to import and pushing a button to sync.
Then, use reverse ETL to send conversion event data models to your conversions API endpoints.
A reverse ETL service like Hightouch can quickly send conversion events to conversions APIs. Reverse ETL makes it easy for you to forward data/events within your warehouse to any destination of your choosing (ex: Facebook Ads and Google Ads). Much like with Fivetran, all you have to do is provide access to the appropriate tables in your data warehouse, match up the columns and identifiers to each respective platform in the Hightouch UI and schedule your data sync. On top of this, you can even build custom audiences directly within Hightouch using standard SQL or your existing data models.
A quick implementation time using automated off-the-shelf tools will minimize marketing downtime and provide an immediate return on investment. Note that it might be tempting to use a customer data platform (CDP) to improvise advertising performance, but there are good reasons to simply construct your own equivalent using your data warehouse.
Want advice on the best way to use your data warehouse to power your martech stack? If you haven’t already, start your free trial of Fivetran to see how the first leg of the process can be automated.