flowsam: Helping brands create eye-catching video content with Google Cloud
About flowsam
Copenhagen-based flowsam offers an AI-based video brand performance pre-testing tool used by marketing departments and creative agencies. Companies upload their video content and tag their brand cues, flowsam then delivers detailed and actionable eye-tracking insights so they can optimize their videos, before spending money on media.
Tell us your challenge. We're here to help.
Contact usUsing Google Cloud solutions, flowsam has disrupted the market research industry by creating a tool that uses AI and ML to show brands how visually effective their video content is with audiences.
Google Cloud results
- Video AI improves the flowsam tool’s precision of object detection and tracking
- Custom scaler, via Compute Engine API, allows the flowsam team to set an upper limit to operations costs
- Scalability improves sustainability, helping flowsam to only use the resources it needs without running servers 24/7
A tool that analyzes and helps optimize video content
Advertising has changed radically in the last 20 years. The halcyon days of Mad Men dreaming up big and expensive national advertising campaigns are fast becoming history. The old media landscape of print and TV advertising has fractured across social media, creating big opportunities, and challenges, for brands. Digital ad spend has risen significantly during 2021, growing its share of total ad spend in the U.S., for example, from 40% in 2019 to 51% in 2021. And one of the biggest gainers in the digital field over the two-year span has been video. But how can brands understand the effectiveness of their video content in an always-on world, while maximizing their ad spend?
“The cost of researching the effectiveness of video content is very expensive. That’s why we created flowsam using Google Cloud. When you run a video through flowsam’s artificial intelligence (AI) software you can quickly analyze its strengths and weaknesses.”
—Caroline Deichmann-Bendixen, co-founder, flowsamTech start-up flowsam believes it has the answer. Founded in Copenhagen, Denmark, in 2018 by four colleagues from marketing agency Adnuvo, the company has created a tool to help brands test a video’s visual effectiveness before it’s launched. Caroline Deichmann-Bendixen, one of flowsam’s co-founders, has a successful career in marketing, and she has seen how brands can struggle to get effective analysis of their video content, particularly in the age of “always-on.”
“The cost of researching the effectiveness of each video is very expensive,” says Deichmann-Bendixen. “That’s why we created flowsam using Google Cloud. When you run a video through flowsam’s artificial intelligence (AI) software you can quickly analyze its strengths and weaknesses.”
One of flowsam’s largest clients, global fruit drink producer CO-RO, has already reaped the benefits of flowsam’s innovative technology. “We have a long history of working with eye-tracking technologies. Flowsam’s tool belongs to the new generation of eye tracking technology combining neuroscience with AI, which allows us to have a faster version of traditional eye-tracking without sacrificing accuracy of results,” says Jesper Kvisgaard, Director of Global Insight at CO-RO. “Working with this kind of tool has been a huge breakthrough for us, allowing us to do fast, efficient, and fact-based iterations on our visual material, creating much better results at a higher speed and lower cost. The impact on the emotional response from consumers to our visuals is clear and significant, and using technology like this will create a significant difference for us in terms of meeting our brand KPIs via our visual communication.”
The flowsam tool combines multiple layers of AI and machine learning (ML), classical algorithms, and a web-based front end to create a unique platform, one that has been powered by Google Cloud from the very beginning.
“Video AI Is used for object classification and tracking brand cues. We explored other object detection models, but found that Video AI had the best overall precision.”
—Jon Voetmann, co-founder, flowsamReinventing video advertising with Video AI
As a start-up without rich backers, flowsam had to bootstrap its way through the development process. This didn’t stop its global ambition; the flowsam tool was created as a SaaS solution, to be used anytime, anywhere. The founders knew from day one that they wanted to build the flowsam tool in the cloud, and after testing a few cloud providers, they opted for Google Cloud due to its ease of use, features, flexibility, and cost.
Of those features, Video AI is perhaps the most important in bringing to life flowsam’s vision. Using Video AI, the flowsam tool can predict where the viewers’ eyes are looking at on the screen during a video, and how much attention the various brand cues such as the logo, the product, and brand messages will receive at any given time in the video. “Video AI Is used for object classification and tracking brand cues,” says Jon Voetmann, co-Founder of flowsam. “We explored other object detection models, but found that Video AI had the best overall precision.”
The results of the video tests are then processed and stored in the database. This data is used to calculate the attention the objects get throughout the video. “We’ve added tons of data into the algorithm and trained it to be able to predict where people are actually looking in a video,” explains Voetmann.
The flowsam team needed powerful GPUs for its custom ML training and inference, and they also wanted to be able to dynamically scale CPU and GPU processing power by a custom algorithm. They employed Compute Engine and Compute Engine API to help them achieve the results they wanted.
“All developed components are running in docker containers on Compute Engine instances,” says Voetmann. “The CPU/GPU workers for video processing and ML inference run in preemptible instances. This helps keep costs down. We use the Compute Engine API to monitor, spawn, and shut down worker-instances as needed by demand.”
The flowsam tool also uses Cloud Storage for the large amount of web-accessible data storage it requires to run effectively. All of these Google Cloud features have enabled the team to really push the capabilities of the flowsam tool, ensuring that it runs faster and more efficiently.
“flowsam will help marketers get the insights that they need without paying so much to research companies. This allows marketing to fulfill its potential as predictive, rather than simply being reactive. And we want to keep on innovating in this space with Google Cloud.”
—Caroline Deichmann-Bendixen, co-founder, flowsamBuilding sustainable success with Google Cloud
The relationship between Google Cloud and flowsam has been instrumental in creating a more effective, nimble start-up. The flowsam team is already seeing the benefits it brings to the company and its clients. “We’ve set up uptime checks, error monitoring, and cloud alerting, which enable us to respond to any problems very quickly,” says Voetmann. “And our custom scaler, via the Compute Engine API, also allows us to define an upper limit to our operations cost.”
The cost-effectiveness of the flowsam tool also really helps brands compete on a level playing field with much larger companies, adds Kvisgaard of CO-RO. “There are brands out there with huge marketing budgets, but they still don’t have the insight that we have now because of flowsam. flowsam helps us to communicate our brand message with as few cues as possible. It’s told us more than we have ever known before about what wins the attention of the eye. It has given so much more insight into our visual assets.”
With so many brands and companies bringing their marketing in-house and needing valuable insights into how their video content is performing, flowsam can see the potential to disrupt the model offered by so many traditional research companies. It has a growing number of clients and aims to make the flowsam tool an integral part of marketing departments worldwide.
This scalability also helps flowsam be a more sustainable start-up. “Google Cloud has helped us to build a ‘green’ powerful system, only using the actual resources needed at a much lower cost than using traditional servers running 24-7,” says Voetmann.
“flowsam will help marketers get the insights that they need without paying expensive fees to research companies,” says Deichmann-Bendixen. “This allows marketing to fulfill its potential as predictive. And we want to keep on innovating in this space with Google Cloud.”
Tell us your challenge. We're here to help.
Contact us
About flowsam
Copenhagen-based flowsam offers an AI-based video brand performance pre-testing tool used by marketing departments and creative agencies. Companies upload their video content and tag their brand cues, flowsam then delivers detailed and actionable eye-tracking insights so they can optimize their videos, before spending money on media.