The broadcaster wanted to ensure that any scheduled advertisement is shown to the right audience at the correct time—a complicated process. It entails integrating data from ad servers, content delivery devices, targeting algorithms, and viewer demographics. Our task of integrating 30+ data sources—amounting to petabytes of data—was particularly daunting. Adding to the challenge, ad viewership data is largely unstructured, yet our customer wanted to eliminate viewer fatigue from repetitive commercials while also minimizing irrelevant ads.
The second challenge involved analyzing content viewership across consumption channels to assess show effectiveness, ad placement efficacy, and viewer behavior. Such insights inform decisions related to contract renewal, demographic-based targeting, and ad platform selection.
Prior to implementing our data product, a team of 20+ FTEs manually cleansed and unified our customer’s data from various sources, aggregated it, and prepared Excel spreadsheets. But such aggregated reports were of little help in analyzing granular data, rendering them unfit for decision support. For example, insights such as exact ad impressions within specific timeslots were unavailable.
The result was uninformed program decisions caused by fragmented metrics across teams and businesses. Lack of series- and episode-level insights led to miscalculated ad impressions, revenue leakage, and failed targeting.