Digital content consumption analysis

A system for estimating user engagement with a content item, the system comprising:

• a memory;
• a processor operatively coupled to, and configured to execute instructions stored in, the memory;
• a consumption estimation module comprising instructions, stored in the memory, that when executed using the processor are configured to receive content consumption data extracted from a tracking log, determine a plurality of content consumption session progress values based on the received content consumption data, and store the content consumption session progress values in an incomplete session progress matrix, wherein a specific session progress value associated with an interaction between a specific content consumer and a specific content item is based on a ratio of (d−1) to D, wherein d equals a quantity of content item segments of the specific content item that were initiated at a content consumption device, and D equals a total quantity of content item segments comprising the specific content item;
• a partial matrix construction module comprising instructions, stored in the memory, that when executed using the processor are configured to refine the incomplete session progress matrix by removing content consumption session progress values associated with (a) a content item that comprises fewer than a threshold quantity of segments, and (b) a content consumer who has interacted with fewer than a threshold quantity of unique content items;
• a matrix completion module comprising instructions, stored in the memory, that when executed using the processor are configured to generate a completed session progress matrix based on (a) the content consumption session progress values determined by the consumption estimation module and (b) the incomplete session progress matrix as refined by the partial matrix construction module;
• and
• a user interface module comprising instructions, stored in the memory, that when executed using the processor are configured to respond to a received analytical query that identifies a particular content item and a particular content consumer, wherein responding to the received analytical query includes (a) extracting, from the completed session progress matrix, a particular content consumption session progress value that is associated with the particular content item and the particular content consumer, and (b) generating a recommendation to insert targeted content into the particular content item at a time point that depends on the particular content consumption session progress value.

A non-transitory computer readable medium having instructions encoded thereon that, when executed by one or more processors, causes a content consumption session progress prediction process to be carried out, the process comprising:

• receiving content consumption data from a content provider, wherein the content consumption data characterizes a plurality of interactions between a first group of content consumers and a first repository of content items;
• calculating a plurality of session progress values based on the content consumption data, wherein each of the calculated session progress values is associated with one of the interactions between one of the content consumers and one of the content items;
• populating an incomplete session progress matrix with the plurality of calculated session progress values;
• refining the incomplete session progress matrix by removing calculated session progress values associated with (a) content items that comprise fewer than a threshold quantity of segments, and (b) content consumers who have interacted with fewer than a threshold quantity of unique content items;
• after refining the incomplete session progress matrix, completing the incomplete session progress matrix using a collaborative filtering matrix completion technique that results in a completed session progress matrix that includes a plurality of estimated session progress values;
• receiving a query that identifies a particular content item and a particular content consumer;
• in response to receiving the query, generating a content consumption session progress prediction for the particular content item and the particular content consumer identified in the query using a corresponding particular estimated session progress value extracted from the completed session progress matrix;
• and
• generating a recommendation to insert targeted content into the particular content item at a time point that depends on the content consumption session progress prediction.