System1, a leading omnichannel customer acquisition platform, revolutionizes how advertisers connect with targeted audiences using advanced machine learning and Generative AI.
By optimizing internet traffic monetization, System1 ensures ads reach the right people at the right time, boosting customer acquisition, engagement, and conversion rates. With a commitment to driving ROI across industries, System1 helps advertisers maximize their impact and stay competitive in a rapidly evolving digital landscape.
Bridging the Gap Between Business and Data Science
Before adopting Dataiku, System1 faced a common challenge in data science: aligning technical work with business outcomes. The team often spent considerable time solving complex technical problems. As Graham Yennie, Senior Manager of Machine Learning Engineering at System1, noted, “data science as a cost center directly competes for budget with its internal customers. You want business leaders defending your budget, not you [because you’re laser-focused on their P&L]”.
Meanwhile, media buyers — System1’s clients — were responsible for ideating and planning marketing campaigns, from defining target audiences to selecting channels and crafting messaging to meet their goals. Their role also required leveraging data insights provided by the data science team to optimize campaigns and scale them with data science and engineering rigor.
To bring these ideas to life, media buyers would provide detailed requirements to the data science team, who would then run experiments to find the most effective approach. However, with the data science team’s limited involvement in the early stages of campaign planning, disconnects emerged, making it difficult to fully align with business objectives throughout the development process. This resulted in a trial-and-error process, with results being passed back and forth for approval, ultimately slowing down the development timeline.
As System1’s business scaled and its tech stack grew more complex, this disconnect between teams became a significant barrier to scaling innovation at the required speed.
The more time you spend reinventing the same wheels that drive data science pipelines, the less time you spend on the work that is of high value to your business. If you say yes to building the same framework, infrastructure, and tools that smart people have already solved for you, what are you saying no to? The whole point of being a software engineer is to automate the boring stuff away and build your value story on top of the tools that enable you.
Graham Yennie Senior Manager, Machine Learning Engineering, System1
Transforming Collaboration Between Data Science and Business Teams
To overcome operational bottlenecks, System1 turned to Dataiku, the Universal AI Platform designed to unify teams and foster true collaboration between technical and non-technical teams while rapidly accelerating data science projects and maintenance. By implementing Dataiku, System1 brought its data scientists and media buyers into a shared workspace, allowing them to collaborate in real time from the early stages of campaign planning. This ensured clear communication, alignment on business objectives, and ultimately transformed how teams worked together.
With Dataiku, data scientists and business stakeholders could ideate, develop, and execute ideas side by side. The platform enabled rapid experimentation and the ability to kill bad ideas faster, thanks to the tight feedback loop between business and technical teams. This transition from feature-focused, engineering-driven workflows to a collaborative, product-focused model not only accelerated strategy iteration and ensured teams stayed aligned at every step, but also kept them agile in adapting to evolving business needs and developments.
Being able to actually sit with a business stakeholder and show them exactly what we are implementing and walk them through the steps without it being hidden and abstracted away into some code base that value can't be overstated.
Kenan Yates Group Product Manager, System1
Dataiku also fostered a “one team” mentality, uniting data science, engineering, and business teams to collaborate closely. By working around the same workbench, they make real-time decisions and solve challenges as they arise. This alignment not only empowered business users to take on more analytics tasks, freeing data scientists to focus on high-impact projects, but it also accelerated innovation and drove better business outcomes, all while keeping teams engaged in continuous upskilling.
Now we have a much more connected development process. [Now], we are all sitting in the same room to understand what is the actual problem that we're trying to solve? What are the metrics that we're trying to drive? Dataiku as an "open development forum" enables that.
Kenan Yates Group Product Manager, System1
Leveraging Visual Recipes and the LLM Mesh for Team Alignment
A key driver of this transformation was Dataiku’s visual recipes. These intuitive tools allowed data scientists to visually demonstrate data processes, giving business stakeholders clear visibility into outcomes without needing to understand the technical details. This transparency broke down communication barriers, enabling technical and non-technical teams to collaborate seamlessly and move faster. It also empowered business users to independently trace the origin of data and understand how it was transformed, reducing reliance on other teams. By minimizing the back-and-forth, it maintained context and improved overall efficiency.
The amount of quick and rapid experiments that Dataiku enabled allowed us to move exponentially quicker. We were able to identify, kill off, or graduate good or bad ideas at a much more rapid pace than we were previously because of the very excellent abstractions that Dataiku provides through the visual recipes and through the LLM Mesh.
Kenan Yates Group Product Manager, System1
Additionally, the Dataiku LLM Mesh played a pivotal role in System1’s success, particularly in their Generative AI initiatives. The LLM Mesh seamlessly integrated large language models (LLMs) into System1’s workflows, providing business users with end-to-end visibility into the data transformation process — from raw data inputs to the prompts used and, ultimately, to the final AI-generated outputs. This level of transparency helped business users fully understand how each transformation impacted critical business metrics, such as campaign performance and customer engagement. And by bridging the gap between data scientists and business stakeholders, the LLM Mesh ensured that every initiative stayed aligned with business objectives.
We generated 107 campaigns in 10 minutes which would normally take a team of three people an entire week to produce in our existing infrastructure. [After a few days] they’re already profitable.
Kenan Yates Group Product Manager, System1
Accelerating Asset Generation
One transformation brought by Dataiku was in System1’s performance marketing campaign asset generation process. With development timelines stretching up to six months just to reach proof of concept. With Dataiku, System1 unified these processes, drastically reducing prototype development time to just 10 days. By eliminating complex glue code and minimizing maintenance burdens, the team could focus on quickly testing ideas and delivering solutions that were better aligned with business goals.
Streamlined Operations With Snowflake and Dataiku
In addition to leveraging Dataiku, System1 integrated Snowflake as its data warehouse to streamline data operations. Snowflake’s ability to decouple storage and compute enabled rapid data access and processing, even for less-optimized queries. This freed data scientists from the burden of managing query optimization, allowing them to focus on higher-value tasks like building and refining models.
With Snowflake acting as the data warehouse and Dataiku serving as the “remote control,” System1 was able to quickly build and iterate on data pipelines. Snowflake tables were used as intermediate stages, and Dataiku simplified the orchestration of data transformations that would have traditionally required complex environments like Jenkins or AWS. This integration helped System1 address multiple challenges simultaneously, accelerating their data workflows and improving operational efficiency.
By combining Snowflake’s scalability with Dataiku’s ease of use, System1 optimized its data science processes, allowing teams to focus on tasks that added the most value to the business. This streamlined setup significantly reduced time spent on data engineering tasks, empowering the team to deliver insights and improvements more rapidly, driving faster and more effective decision-making.
We use Dataiku like a remote control for our Snowflake server, allowing us to transform data with easy building blocks. This approach helps us solve multiple problems at once and keeps accelerating our progress.
Graham Yennie Senior Manager, Machine Learning Engineering, System1
Driving Future Agility With Dataiku
Dataiku has helped transform System1 into a more agile, product-driven organization. By fostering a collaborative, hands-on environment, Dataiku has empowered both technical teams and business users to work closely together, aligning their efforts with business goals and driving measurable results.
Looking ahead, System1 plans to extend the use of Dataiku to more business users. By giving them direct access to the Dataiku platform, System1 aims to enable these users to ask better questions and gain deeper insights from the data they work with every day. This will enhance their ability to contribute directly to business operations and decision-making while upskilling critical analytics skills.
As Dataiku becomes more integrated into daily workflows, System1 will continue to streamline its operations, making faster, data-driven decisions and reinforcing its leadership in the digital marketing space.
I think Dataiku has really been a central force for how we're operating now and how we want to see the team continue to operate.
Kenan Yates Group Product Manager, System1