Leighton Welch is CTO and co-founder of Tracer. Tracer is an AI-powered tool that organizes, manages, and visualizes complex data sets to drive faster, more actionable business intelligence. Prior to becoming the Chief Technology Officer at Tracer, Leighton was the Director of Consumer Insights at SocialCode, and the VP of Engineering at VaynerMedia. He has spent his career pioneering in the ad tech ecosystem, running the first ever Snapchat Ad and consulting on commercial APIs for some of the world’s biggest platforms. Leighton graduated from Harvard in 2013, with a degree in Computer Science and Economics.
Can you tell us more about your background and how your experiences at Harvard, SocialCode, and VaynerMedia inspired you to co-found Tracer?
The original idea came a decade ago. A childhood friend of mine rang me on a Friday night. He was struggling with aggregating data across various social platforms for one of his clients. He figured this could be automated, so he enlisted my help since I had a background in software engineering. That’s how I was first introduced to my now co-founder, Jeff Nicholson.
This was our light bulb moment: The quantity of money being spent on these campaigns was far outpacing the quality of the software tracking those dollars. It was a nascent market with a ton of applications in data science.
We kept building analytics software that could meet the needs of increasingly large and complex media campaigns. As we hacked away at the problem, we developed a process – clear steps from getting the disparate data ingested and contextualized. We realized the process we were building could be applied to any data set – not just advertising – and that’s what Tracer is today: an AI-powered tool that organizes, manages, and visualizes complex data sets to drive faster, more actionable business intelligence.
We’re helping to democratize what it means to be a “data-driven” organization by automating the steps needed to ingest, connect, and organize disparate data sets across functions, providing powerful BI through intuitive reporting and visualizations. This could mean connecting sales data to your marketing CRM, HR analytics to revenue trends, and endless more applications.
Can you explain how Tracer’s platform automates analytics and revolutionizes the modern data stack for its clients?
For simplicity, let’s define analytics as the answering of a business question through software. In today’s landscape, there are really two approaches.
- The first is to buy vertical software. For CFOs, this might be Netsuite. For the CRO, it might be Salesforce. Vertical software is great because it’s end-to-end, it can be hyper specialized, and should just work out of the box. The limitation of vertical software is that it’s vertical: if you want Netsuite to talk to Salesforce, you are back to square one. Vertical software is complete, but it’s not flexible.
- The second approach is to buy horizontal software. This might be one software for data ingestion, another for storage, and a third for analysis. Horizontal software is great because it can handle pretty much anything. You could certainly ingest, store and analyze both your Salesforce and Netsuite data through this pipeline. The limitation is that it needs to be put together, maintained, and nothing works “out of the box.” Horizontal software is flexible, but it’s not complete.
We offer a third approach by creating a platform that combines the technologies necessary to report on anything, made accessible enough to work out of the box without any engineering resources or technical overhead. It’s flexible and complete. Tracer is the most powerful platform on the market that is both application agnostic, and end-to-end.
Tracer processed on the order of 10 petabytes of data last month. How does Tracer handle such a vast amount of data efficiently?
Scale is incredibly important in our world, and it has always been a priority at Tracer even in the beginning days. To process this volume of data, we leverage a lot of best in class technologies and avoid reinventing the wheel where we don’t need to. We’re incredibly proud of the infrastructure we’ve built, but we’re also quite open about it. In fact, our architecture program is outlined on our website.
What we say to partners is this: It’s not that your in-house engineering teams aren’t capable of building what we’ve built; rather, they shouldn’t have to. We’ve assembled the pieces of the modern data stack for you. The framework is efficient, battle-tested, and modular for us to dynamically evolve with the landscape.
A lot of partners will come to us looking to free up engineering resources to focus on bigger strategic initiatives. They use Tracer’s architecture as a means to an end. Having a database doesn’t answer business questions. Having an ETL pipeline doesn’t answer business questions. The thing that really matters is what you’re able to do with that infrastructure once it’s been put together. That’s why we built Tracer – we’re your shortcut to getting answers.
Why do you believe structured data is critical for AI, and what advantages does it provide over unstructured data?
Structured data is critical for AI because it allows for manual human interaction, which we believe is an essential component to effective outputs. That being said, in today’s ecosystem, we are actually better equipped than ever before to leverage the insights in unstructured data and previously hard to access formats (documents, images, videos, etc.).
So for us, it’s about providing a platform through which additional context can be incorporated from the people who are most familiar with the underlying datasets once that data has been made accessible. In other words, it’s unstructured data → structured data → Tracer’s context engine → AI-driven outputs. We sit in between and allow for a more effective feedback loop, and for manual intervention where necessary.
What challenges do companies face with unstructured data, and how does Tracer help overcome these challenges to improve data quality?
Without a platform like Tracer, the challenge with unstructured data is all about control. You feed data into the model, the model spits out answers, and you have very little opportunity to optimize what’s happening inside the black box.
Say for example you want to determine the most impactful content in a media campaign. Tracer might use AI to help provide metadata on all the content that was run in the ads. It also might use AI to provide last mile analytics for getting from a highly structured dataset to that answer.
But in between, our platform allows users to draw the connections between the media data and the dataset where the outcomes live, more granularly define “impactful,” and clean up the categorizations done by the AI. Essentially, we’ve abstracted and productized the steps, in order to remove the black box. Without AI, there is a lot more work that has to be done by the human in Tracer. But without Tracer, AI can’t get to the same quality of answer.
What are some of the key AI-based technologies Tracer uses to enhance its data intelligence platform?
You can think of Tracer across three core product categories: Sources, Content, and Outputs.
- Sources is a tool used to automate the ingestion, monitoring and QA of disparate data.
- Context is a drag and drop semantic layer for the organization of data after it’s been ingested.
- Outputs is where you can answer business questions on top of contextualized data.
At Tracer we don’t see AI as a replacement for any of these steps; instead, we see AI as another form of tech that all three categories can leverage to expand what can be automated.
For example:
- Sources: Leveraging AI to help build new API connectors to long tail data sources not available through our partner catalog.
- Context: Leveraging AI to clean up metadata prior to running tag rules. For example, cleaning up variations of publication names in every language.
- Outputs: Leveraging AI as a drop-in replacement for dashboards where the business use case is exploratory, rather than a fixed set of KPIs that need to be reported on repeatedly.
- AI allows us to achieve these types of applications in ways that are both simple and accessible.
What are Tracer’s plans for future development and innovation in the data intelligence space?
Tracer is an aggregator of aggregators. Our partners will lean on us for specific applications within teams and functions, or for use in cross-functional business intelligence. The beauty of Tracer is that whether you’re leveraging us for making better decisions with your media spend and creative, or building dashboards to link disparate metrics from supply chain to sales and everything in between, the building blocks are consistent.
We’re seeing organizations who formally relied on us within one area of the business (e.g., media and marketing), expand applications to elsewhere in the business. So where our primary customers were formally senior media executives, or agency partners, these days we work across the org, partnering with CIOs, CTOs, data scientists, and business analysts. We’re continuing to build out our tools to accommodate for more and more applications and personas, all while ensuring the core tech is scalable, flexible, and accessible for non-technical users.
Thank you for the great interview, readers who wish to learn more should visit Tracer.
Credit: Source link
