LightTip NSF FInal Presentation
•Download as KEY, PDF•
2 likes•3,475 views
The document discusses developing an AI system to optimize thumbnail images for online videos. It notes that online video advertising generated $2.3 billion in 2011. The system would use neuroscience methods like MRI to analyze which thumbnails are most effective at getting users to click. The goal is to increase click-through rates and video views, helping companies that make money from online advertising. Key partners could include neuromarketing firms and companies with large video libraries. Revenue would come from a SaaS model charging companies per click or view generated.
Recommended
Recommended
More Related Content
Similar to LightTip NSF FInal Presentation
Similar to LightTip NSF FInal Presentation (20)
More from Stanford University
More from Stanford University (20)
Recently uploaded
Recently uploaded (20)
LightTip NSF FInal Presentation
- 1. labs Choices
- 2. labs Sophie Lebrecht Michael J. Tarr Entrepreneurial Lead Principle Investigator Postdoctoral Researcher Co-director of Center for the Center for the Neural Basis of Cognition Neural Basis of Cognition Tepper School of Business Professor of Psychology Tom Kubilius Babs Carryer Mentor Mentor Founder & President Embedded Entrepreneur Bright Innovation Project Olympus, CMU
- 3. At the beginning $58 Billion $2 trillion
- 4. Day Month Year No.1 No. Who are our Key Partners? What Key Activities do our Value Propositions require? What value do we deliver to the customer? What type of relationship does each of our Customer For whom are we creating value? Who are our key suppliers? Our Distribution Channels? Which one of our customer’s problems are we helping to solve? Segments expect us to establish and maintain with them? Who are our most important customers? Which Key Resources are we acquiring from partners? Customer Relationships? What bundles of products and services are we offering to each Customer Segment? Which ones have we established? Which Key Activities do partners perform? Revenue streams? Apps Which customer needs are we satisfying? How are they integrated with the rest of our business model? How costly are they? Production of Market Market software Researchers Improve User community researchers predictably Platform for data Co-creation Neuromarketing interpretation by adding data Researchers Neuro-marketing Improve speed of companies service Design Design What Key Resources do our Value Propositions require? Our Distribution Channels? Customer Relationships? Revenue Streams? Improve Through which Channels do our Customer Segments want to be reached? How are we reaching them now? consultancies How are our Channels integrated? consultancies interpretation Which ones work best? Which ones are most cost-efficient? IP How are we integrating them with customer routines? Ad Consultancies MRI machine Market Ad consultancies Improve researchers convenience of Brands Software data acquisition Discount Retail Design Stores platform consultancies Value driven What are the most important costs inherent in our business model? Software For what value are our customers really willing to pay? Bait and hook Which Key Resources are most expensive? Which Key Activities are most expensive? development For what do they currently pay? How are they currently paying? How would they prefer to pay? apps to brain software How much does each Revenue Stream contribute to overall revenues? R&D lab Data servers MRI data collection Freemium data Access to data interpretation costs collection platform software This work is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-sa/3.0/ or send a letter to Creative Commons, 171 Second Street, Suite 300, San Francisco, California, 94105, USA.
- 5. What we learned... Market Research Neuromarketing Ad Agencies In house Brand R&D
- 6. $58 Billion $2 trillion
- 7. Day Month Year No.3 No. Apps Who are our Key Partners? What Key Activities do our Value Propositions require? What value do we deliver to the customer? What type of relationship does each of our Customer For whom are we creating value? Who are our key suppliers? Our Distribution Channels? Which one of our customer’s problems are we helping to solve? Segments expect us to establish and maintain with them? Who are our most important customers? Use MRI to Which Key Resources are we acquiring from partners? Which Key Activities do partners perform? Customer Relationships? Revenue streams? What bundles of products and services are we offering to each Customer Segment? Which customer needs are we satisfying? Which ones have we established? How are they integrated with the rest of our business model? User How costly are they? Market/neuro expand simple community researchers methods Improve Market predictably Design Production of consultancies researchers software Co-creation Platform for data Improve speed by adding data Snack & Food Neuro-marketers interpretation of service Brands Improve Ad consultancies Design What Key Resources do our Value Propositions require? Through which Channels do our Customer Segments Personal Software Our Distribution Channels? Customer Relationships? want to be reached? consultancies Revenue Streams? interpretation How are we reaching them now? platform How are our Channels integrated? Which ones work best? Which ones are most cost-efficient? Hygiene Market How are we integrating them with customer routines? Brands researchers IP Improve the Ad consultancies MRI machine convenience of data Discount Apps analyzed Samplers Retail acquisition with MRI data Stores Software What are the most important costs inherent in our business model? For what value are our customers really willing to pay? Bait and hook Which Key Resources are most expensive? development For what do they currently pay? Which Key Activities are most expensive? How are they currently paying? Freemium How would they prefer to pay? How much does each Revenue Stream contribute to overall revenues? apps to brain software Value driven data R&D lab MRI data collection Access to data Data servers collection costs software interpretation platform This work is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-sa/3.0/ or send a letter to Creative Commons, 171 Second Street, Suite 300, San Francisco, California, 94105, USA.
- 12. The thumbnail is the gateway to the ad
- 13. People watched 110B online content videos 25B online ads
- 14. 1st 2nd 3rd labs Thumbnail optimizer
- 16. How companies pick thumbnails now Automatic Random Thoughtful By hand
- 17. How companies pick thumbnails now Automatic Random Thoughtful By hand
- 18. How companies pick thumbnails now Automatic Random Thoughtful By hand
- 19. How companies pick thumbnails now Automatic labs Random Thoughtful By hand
- 20. Our technology increases clicks
- 21. Our technology increases clicks “If you get us more clicks, I can make money from clicks” Bryon Schafer, Director, Research & Sales Strategy at Hulu
- 22. 650M videos per month 400M in revenue per year
- 23. Target market size Online video advertising generated $2.3B in ad revenue, 2011
- 24. Next steps MVPs
- 26. labs Using brain science to extract thumbnails that make users click sophielebrecht@cmu.edu
Editor's Notes
- \n
- We worked together as a team to solve this problem. The science underlying this venture was developed between Brown University and Carnegie Mellon led by me and in collaboration with the PI michael Tarr and others. The business development was supported by my mentors Tom Kubilius and Babs Carryer.\n
- In principle our science can take a piece of visual information and using our understanding of particular brain based mechanisms we can predict how people will respond to it\nInitial Idea: was to put people into MRI machines and measure their response to consumer products.\nThis actually is a big market, but given our initial business model it would have been almost impossible for us to have gotten a piece of this\nConsumer goods is a 2 trillion dollar market but havimg me scan each of these packages we were never going to get a piece of that market.\n
- \n
- There are lots of people working to improve package design. And visual ad media. Many people are working in package testing. Other neuromarketing running a service are not making money. This space is crowded. They don’t understand brain data. No time and money for it. Ad agencies as a sales pitch to their customers rather than something meaningful. \n
- In principle our science can take a piece of visual information and using our understanding of particular brain based mechanisms we can predict how people will respond to it\nInitial Idea: was to put people into MRI machines and measure their response to consumer products.\nThis actually is a big market, but given our initial business model it would have been almost impossible for us to have gotten a piece of this\nConsumer goods is a 2 trillion dollar market but havimg me scan each of these packages we were never going to get a piece of that market.\nWe learnt that customers were happy to have a meaningful prediction that came from brain science but they certaintly had no need or interest in the raw MRI data. Which meant we could provide them a valuable prediction score without having to sell them the raw MRI data/\nWe always knew our science was capable of this, but before talking to customers we thought that they would want or need the raw data. \n
- \n
- In the spirit of I-corps we got out of the building and spoke to lots of customers - here is just a subset of some of the companies. What was most helpful were companies like Hulu and comscore where we were able to form a relationship and have multiple meetings to get a deep understanding of their problem and solutions that they need.\n
- \n
- \n
- So then we spoke to other customers that put video content online and realized not only to they share this problem. But no body is solving it in a systematic way. \n
- \n
- Since I-Corps started roughly 181 million US internet users watched nearly 110 online content videos - Comscore video metrix and 8 billion online ad videos, march, 9.5 billion april, 25 billion online ads\nVideo marketing, news, sport, TV music\n \n
- Thought thumbnails, then thought upload the whole video, then realized that we needed an API to OVP like Youtube. Realized youtube channels are customers.\n
- \n
- \n
- \n
- \n
- \n
- \n
- \n
- \n
- \n
- \n
- \n
- We are building a web based software that takes a stream of video and extracts the best thumbnail that will increase the likelihood that will increase users click through rate\n
- \n