Marketing ops · Production

Netflix uses transfer learning and knowledge graphs to support content commissioning decisions at global scale

The problem

Content decision makers at Netflix needed to identify comparable titles and predict audience sizes for new content at global scale, but conventional methods relied on a limited set of comparables and performance metrics that could not handle Netflix's increasingly vast and diverse catalog.

Workflow diagram · grounded in source
1
Executive requests comparables
trigger
“Content, marketing, and studio production executives make the key decisions that aspire to maximize each series' or film's potential to bring joy to our subscribers as it progresses from pitch to play on our service”
2
Supervised transfer learning produces embeddings
ai_action
“we learn a model on a large set of historical titles, leveraging information such as title metadata (e.g., genre, runtime, series or film) as well as tags or text summaries curated by domain experts describing thematic/plot elements. Onc…”
3
Knowledge graph self-supervised embedding
ai_action
“we can craft a self-supervised learning task where we randomly select edges in the graph to form a test set, and condition on the rest of the graph to predict these missing edges. This task, also known as link prediction, allows us to le…”
4
Similarity map surfaces comparables
ai_action
“we "embed" titles in a high-dimensional space or "similarity map," wherein more similar titles appear closer together with respect to a spatial distance metric such as Euclidean distance. We can then use this similarity map to identify c…”
5
Neural network predicts audience size
ai_action
“for audience size prediction, we use a single hidden-layer feedforward neural network to minimize the mean squared error for a given title-country pair”
6
Insights delivered to decision makers
output
“We identified two ways to support content decision makers: surfacing similar titles and predicting audience size, drawing from various areas such as transfer learning, embedding representations, natural language processing, and supervise…”
Reported outcome

Machine learning embeddings enabled scalable surfacing of comparable titles and significantly improved audience size prediction accuracy, supporting content, marketing, and studio executives in commissioning decisions globally.

Reported metrics
Audience size prediction accuracysignificantly improve prediction accuracy
Reported stack
MetaflowBERTLSTMGRUTransE
Source
https://netflixtechblog.com/supporting-content-decision-makers-with-machine-learning-995b7b76006f
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Machine learning embeddings enabled scalable surfacing of comparable titles and significantly improved audience size prediction accuracy, supporting content, marketing, and studio executives in commissioning decisions…

What tools did this team use?

Metaflow, BERT, LSTM, GRU, TransE.

What results were reported?

Audience size prediction accuracy: significantly improve prediction accuracy (source-reported, not independently verified).

How is this marketing ops AI workflow structured?

Executive requests comparables → Supervised transfer learning produces embeddings → Knowledge graph self-supervised embedding → Similarity map surfaces comparables → Neural network predicts audience size → Insights delivered to decision makers.

Related marketing ops cases