Podknife’s content exploded to 3 million+ podcasts. Audiences could no longer find relevant, useful material, while brands had no way to jump in on the growth.
Ekohe’s custom NLP & Machine Learning engines analyzed all content quickly, powering features like semantic (conversational) search and hyper-targeted insights.
Podknife’s content model now flows from automatic deep-learning ‒ driving exceptional user engagement and new ad revenue streams at the same time.
Born in 2016, Podknife’s trajectory soon shifted ‒ from curated information/review provider to open-aggregator, helping audiences bypass the closed walls of App stores. Today, the firm holds more than 20,000 podcasts from over 50 countries, or 3 million+ episodes in its database. And with the platform’s explosive growth, audiences struggled to find relevant podcasts. Platform usefulness started wavering, threatening future scale. Podknife needed to answer 4 questions ASAP: (1) Could we quickly analyze our giant set of audio data ‒ finding actionable insights to increase engagement? (2) How fast could these insights make it to market? (3) Could a solution automate workflow while also serving custom audience/admin needs in the future? (4) Would this enable us to differentiate ourselves with brands?
Check what Podknife founder shared their journey
Podknife sought hidden treasure in its 3mm+ podcast episodes. Could they harness the power of data to drive audience/brand engagement, complement manual work, set up for scale, and make their mark against big players?
Ekohe was up to this challenge. Partnered since Day 1, Podknife trusted our experts to offer the strategy, Data Science/ML knowledge, and rapid technical delivery to meet its critical business needs. Understanding that exceptional, useful, and automated engagement will drive all wins, we tagged our Natural Language Processing (NLP) Pipeline as the optimal way to solve core questions. Our custom solution included: 1. **NLP Transcribe**: Quickly analyzing millions of assets ‒ by titles, description, and actual audio ‒ uncovering hidden themes to drive relevant platform content & navigation; serving as a powerful tool to complement manual tagging, via 24/7 automated content classification. 2. **Neural Search**: Enabling truly conversational, semantic Q&A in the future ‒ maximizing content relevancy via deep-learning search; while driving the most accurate, useful results by weighting exact words and frequency in each podcast episode.
Word use frequency visualization and filtering for each episode
Podknife now automatically creates highly personalized feeds from searches, and can even export them as newsletters; recommending podcasts based on deep interests/segmentation. The firm wins on multiple fronts, also powering new hyper-targeted ad revenue channels. Brand engagement flows from invaluable, continuously evolving insights ‒ at their foundation, a deeply engaging, relevant, and useful audience experience. Innovation in action? That’s Ekohe. By tapping into the power of Podknife's existing audio data, we’re helping the firm make its mark as a one-of-a-kind platform ‒ driving an open, scalable model for podcast fans, creators, and brands/advertisers.