March 2016 Summaries

Snowplow enables comprehensive tracking of both ad impressions and ad clicks, providing valuable insights for users engaging in display advertising. Ad impression tracking can be implemented using either the Javascript tracker or a pixel tracker, with both methods allowing for the collection of detailed data points such as user IDs, ad identifiers, and campaign details. The Javascript tracker offers ease of integration and supports multiple ad units on a single page through namespacing, while the pixel tracker provides a lightweight alternative by composing a GET request with essential data points. For ad clicks, Snowplow utilizes a redirect mechanism that captures click events by rerouting the user through its collector, which records the event before redirecting to the intended target URL. This approach allows the inclusion of additional data points through the querystring to enhance the detail of tracked click events. Both methods support the creation of custom schemas to tailor the tracking to the specific data availability of different ad servers, and encourage contributions to Snowplow's Iglu Central schema registry.

Snowplow distinguishes itself by reporting not only successful data but also bad data that fails to process, allowing users to quickly identify and address data tracking issues. Recently, Snowplow enhanced its capabilities by enabling the loading of bad data into Elasticsearch, facilitating the use of Kibana for monitoring and diagnosing data issues. The Kibana Discover interface provides tools for viewing and filtering bad rows, helping to differentiate between genuine data tracking errors and irrelevant data such as those generated by malicious bots or necessary browser processes like OPTIONS requests. By filtering out these unimportant errors, users can focus on genuine tracking issues, diagnose their causes, and implement solutions, such as schema updates, to resolve them. This proactive approach to managing bad data ensures a high degree of confidence in the data trends, reflecting true business trends rather than data anomalies.