Datastream can be used by any team who finds value in real-time, user-level web and mobile engagement data, including:
Data Analysts and Business Intelligence Teams
Data Scientists and Personalization Teams
Product, Marketing, and Revenue Teams
To help guide other departments, like editorial, marketing, product, or audience development, many data analysis and business intelligence teams run internal analytics products to surface data from multiple data sources in ways that third-party analytics products do not.
To do so, customers many times integrate our data pipeline with their own ETL before dumping the data in Amazon Redshift or BigQuery. Then, they use a tool like Tableau, Looker, or an internal solution to provide data exploration and dashboard interfaces for their teams.
Only raw, user-level data allows BI teams to understand their users at a granular level. By integrating site interaction data with first-party data, subscriptions data, CMS data, Google/Adobe data, and virtually any other data source, BI teams can create custom analyses that help inform company-wide goals. Here are just a few of the metrics included in Datastream:
Rich Time based data: Eg, Engagement data, scroll depth
Multi-platform data: For teams looking to visualize your traffic data with data stream you can pull in data from a variety of platforms
Facebook Instant Articles
Your own native app
Datastream provides unique engagement and high quality data that is easily linkable to other data used in user-level analysis.
Data science use cases:
Understand engagement at a granular level
Conduct user journey analysis
Conversion and subscription attribution
With Datastream, you can use interactive data exploration environments such as:
Python and Pandas, with Jupyter Notebooks
R Studio, for R users
Scikit learn for machine learning
Datastream’s data formats have been specifically optimized to be cleanly integrated by their bulk loading and stream into Amazon Redshift and Google BigQuery. For SQL experts on your team, you can use a tool like Periscope to query your raw data.
Data Science teams working on personalizing their site, or building recommendation engines to better drive engagement for loyal users, Datastream provides unique engagement data that is easily linkable to other data used in personalization algorithms and models.
Datastream provides a real-time pipeline of clean, easy-to-use traffic that can be easily imported into your data infrastructure without any additional extraction or transformation.
Raw data can be a vital tool to improve your product and more accurately target users. Datastream extends your existing customer analytics with user-level attributes such as:
Recency & Frequency
Because the Datastream feed is at the user level, you can analyze and act upon every interaction on your site, allowing your team to do things like:
As a marketer, you might like to provide targeted promotions to guest readers to entice them to subscribe. For instance, readers who frequent cooking and dining-related articles and have higher than average engaged times on those articles may be offered a free cookbook with their subscription, where the cookbook is a Kindle book if the reader most often reads on their tablet or smartphone.
As Head of Customer Engagement, you might like to show readers links to articles similar to those they have engaged with in the past, in order to entice them to continue reading and thus increase the time that readers spend on the site. In order to do this, you need to know which articles (author/section/content type) a user has had the highest engagement with in a recent time period, and on what device, as you may want to construct different recommendation sets dependent on which device the reader is on at the time and what geographic location s/he is coming from. Additionally, the recommendations provided may differ based on the initial page view source, e.g. Search vs Social, as this indicates ‘intent’.
As an Advertising Optimization analyst, you might be charged with determining which ads will be shown where within the context of an article. In order to do this, you’ll want to know the average scroll depth for different reader segments, where the segments might be based on engaged time per section, per author, recency, frequency, geo, device, subscriber status, etc.