Building out a comprehensive marketing strategy across all channels and platforms, each with varied and complex user journeys, is difficult. Evaluating the effectiveness of your marketing spend and calculating it’s ROI however should not be.
Understanding how your marketing spend is contributing to conversions and sales is crucial in enabling you to continually adapt your strategy to changing customer preferences and the market environment. Many commercial solutions exist to help you understand how your marketing campaigns are performing. However, there are two main issues with many of these solutions:
- They are silo-ed. For example, Facebook gives you an overview of your performance on their platform, but it is difficult to compare that to performance of other channels.
- They are one-size-fits-all black boxes that remove flexibility and control. For example, Google Analytics attribution can only attribute one channel to each session.
Snowplow puts you in control of your attributing marketing spend to outcomes. You decide what data goes into your attribution model, you decide what counts as a marketing touch or a conversion across all your channels and platforms, you decide what attribution mechanism you want to use, and you decide how this information is displayed to the various stakeholders in your business.
To get you started, we’ve put together a quick introduction to marketing attribution on web using Snowplow, focusing on:
- Capturing where users come from
- Modeling sessions
- Adding marketing costs (with a Google ads example)
Capturing where users come from
utm_ fields, but other marketing parameters can be specified as well. The referrer parser enrichment will also classify the referrer. Together, these two enrichments populate the following fields in atomic.events:
|Marketing fields||Referred fields|
More information on these fields can be found in Snowplow’s canonical event model.
domain_sessionid, as well as a session index. The session cookie is set against the same domain as the
domain_userid cookie (a first-party cookie set against the domain the tracking is on). By default, it expires after 30 minutes of inactivity, but a different interval can be picked in the tracker initialization (i.e.
sessionCookieTimeout: 3600 ).
The session ID is used to model sessions in Snowplow’s web data model. This SQL model aggregates out-of-the-box page views and page pings into a set of derived tables: page_views, sessions and users. These tables have one row per page view ID (as captured in the web page context), session ID (i.e. the
domain_sessionid) or user ID (i.e. the
In the session table, the marketing and referrer information of the first page view in that session is saved. This can be used to attribute conversions in a given session to a specific marketing channel.
Adding marketing costs (Google Ads example)
If marketing costs are pulled into the data warehouse (using an ETL tool such as Stitch), they can be added to the sessions table based on the marketing parameters. For example, if the Google click and keyword performance reports are available, the average cost per click can be added to sessions that originated from a paid Google search using the marketing click ID:
Once you have developed an understanding of what channels drive customers to your digital products, you can proceed with defining what activities you want to attribute - whether its newsletter signups, pdf downloads, product purchases, subscriptions, etc. This information can also be added to sessions as additional metrics, or flags. The resulting table can then be used as the basis for your various attribution models.
Marketing attribution with Snowplow
Data is driving more high-stakes decisions across companies and industries, and marketing strategies are no exception. As your channel mix and user journeys grow more complex, it becomes less likely that silo-ed or one-size-fits-all commercial tools will deliver what you need to attribute and optimize your marketing spend accurately. Attributing credit to different events in the journey provides evidence of what is and isn’t working, but without being able to take charge of your data to choose the attribution logic that reflects your customers’ journeys (and their touchpoints), you cannot truly understand the real return on your investment. With Snowplow, you have that flexibility and control.