Ahmed Elsamadisi is the Founder and CEO at Narrator, a data modeling platform that allows companies to answer any and all questions about their business’ marketing data in minutes. By taking minutes to get answers about data that would normally take days or weeks, Narrator allows companies to better understand their business and make better marketing decisions. Before Narrator, Ahmed was the first member of WeWork’s data team. Since founding Narrator, Ahmed graduated from startup accelerator Y Combinator and was named a member of the Forbes 30 Under 30 list for 2021.
Our no holds-barred conversation with Ahmed begins with reflections on his experience building the data team behind WeWork, which at the time was in hyper-growth, which is good for a start-up, but can wreak havoc on the way data is collected, distributed, and utilized within the organization. When asked what his advice was to marketing data teams in hyper growth, his advice is simple.
“Don’t go too quickly,” says Ahmed. “Data scales as more team members are added and this causes data management problems. The job of data engineers is thought to be ‘maintaining the system,’ but the real work is in making sure the design of the system is efficacious from the very beginning.”
Ahmed admits that data can be messy, and one of the main issues isn’t finding the right team or engineers, but how labels and tables interact to generate performance results.
“If the naming hierarchy being used is off and misaligned from the reportable metrics the C-suite is looking for, the whole thing is a house of cards waiting to collapse. You need to make sure you build a system designed for hyper-scale. That means definitions are clarified beforehand, and everyone can easily see what progress looks like.”
Pivoting into the concept of democratization of data, Ahmed outlines the importance of data storytelling, something his company Narrator.ai is aiming to address.
“You should democratize not just the data, but the insights and the analysis, the questions that inspired the collection of the data in the first place. What happens is that folks have thousands of dashboards filled with metrics they can’t interpret, so there’s a lot of backpedaling, redefining, and misinterpretation.”
Ahmed’s advice on how to fold narration into reporting is simple. “Keep data clean, organized, and useful. Reorienting marketing data performance to be narration based may seem like a novel concept, but that’s how humans interpret information best. Define concepts, create storylines that can be explored, and keep your analysis based on context, not just a bunch of content.”
Another frustration marketing data teams face that Ahmed spoke freely on, is channel based metrics, and attributing ROI to campaigns.
“Channel-based metrics can be off from internal performance metrics,” says Ahmed, “And the time delay in figuring all that out can be 30 days or more. If conversion metrics are misaligned with marketing tactics, then you blindly have to trust what external platforms are telling you, and you shouldn’t be doing that.”
“SQL queries and traditional data don’t work in real-time. Google and Facebook will show a crazy good ROI for ads served on their platform, hours after a campaign goes live, and that’s just not reality.”
Advertisers and clients also have to contend with platforms lumping a bunch of unrelated activity into their engagement metrics.
“One successful conversion on your end may be attributed to 16 different engagement points,” explains Ahmed, “and their job as a business is to hand over metrics that look good and keep you spending money on the platform. You have to collect your own data and study the difference between conversions and attribution.”
The rest of the interview dives into attribution, retargeting, why obfuscation and complexity are the enemies of good marketing data management, and two horror stories about data interpretation and blindly believing platform metrics. We end with a few blue-sky moments Ahmed and his team at Narrator were able to provide to clients, and some simple advice on how to better manage your data.
“Choose predictable and simple, over big data tables that can’t pivot and grow with your team. Choose things you can understand. Be aware the people selling you on these platforms are biased. Before you jump into complexity ask yourself, what is the incentive to gather this data, what questions does it answer? Lean towards extreme constraints and insights and you’ll be better prepared to deliver metrics that matter.”