Written by Sravan Vadigepalli-

AI is the front and center of every tech, data, and product conversation. To realize the full potential of AI in data products, we need to focus on having the right data, right systems, and right organizational structures. In this post, I’m going to touch on the third piece, ‘right organizational structure’, since the setup of your data teams plays a visceral role in driving successful outcomes. “Right organizational structure” is subjective, so consider this as a guide rather than a rule.

Data & Analytics teams are traditionally built to solve problems quickly, prioritizing the speed and quality of an insight over long-term planning. This approach can lead to solutions that address immediate needs but lack sustainability or fail to gain widespread adoption. Process, scale, standardization, and long-term roadmapping often take a backseat. While valuable for small-scale projects or startups, this approach doesn’t help in building impactful solutions that drive behavior change or contribute significantly to the bottom line. Data teams need to evolve to balance responsiveness with long-term planning. We need a different operating model, where “customer” becomes the center of your outcomes, not just the problem at stake.

Depending on who/where you work, data & analytics teams primarily have 3 major functions 

1. Provide Insights (Analyst function) 
2. Build Models (Data Sciences)
3. Automate & Visualize data (Data Engineering) 

Traditional “problem solving” approach can get away with #1, where speed and richness of insight takes priority. These are mostly ad hoc questions teams must grapple with on a regular basis, like ‘why is my sales down?’, ‘are we bringing new customers’, etc. It is better to have a well-structured process and automation, but the reality is it is difficult and most of the time you can get by with poor process and rich insights. However, when it comes to #2 and #3, this is where the challenge lies. Anytime we build something that has a more shelf life, be it data science models or dashboards/scorecards, process, automation and customer centricity takes precedence over just being nimble. 

Let’s do a quick thought exercise: Reflect on the data science and reporting solutions you’ve built. How many of them have seen widespread adoption across the business? How often have you felt that solutions weren’t fully leveraged to their potential? In today’s resource-constrained environments, underutilized data solutions represent a missed opportunity. This is where a product-centric mindset in data & analytics can be a game-changer. By focusing on user needs and driving adoption, we can ensure our data solutions deliver the right results.

This brings me to make a case for Product led Data & Analytics. 

You might be thinking what does the product even have to do with analytics? Analytics is about insights, reporting, models, etc. you are right and none of that is going to change. The central idea behind product led analytics is to put the customer in the front and center. To simply put, we pivot from problem first mindset to customer first mindset. On any given day, there are a lot of problems we can solve with analytics. But a product led approach forces us to have a structure around it and make sure the problem we are trying to solve is the problem the end user/ customer is looking for to be solved. There comes OKRs (Objectives and Key Results) and Prioritization. OKRs enable us to think surgically about the key results we want to achieve and prioritization frameworks (there are a plenty out there) help us focus on the most important and high value projects. However, OKRs have also gained a bad rep these days as it becomes more process and paper exercise than a true intent. That’s for future discussion.

So, what is a Product led Analytics model? 

At the core of this model lies the product organization, which spearheads the work. Product managers interact with various stakeholders to create an overarching vision and roadmap for achieving that vision. They collaborate closely with embedded teams like data engineering, data sciences, and UX. The product organization is responsible for bringing these teams together towards a common purpose. Clarity in communication and influence play a major role in the product team’s success. Depending on the organizational structure, these teams might report to the same leader or different leaders. Regardless of the structure, communication remains vital. Product managers need to secure buy-in from their peers to drive the roadmap forward. This ability to influence without formal authority is perhaps the most challenging aspect of being a product owner in this model, as it requires achieving alignment across teams for a singular purpose.

 

An example of Product led organizational set up

Are there any downsides to this operating model? Of Course! I’d classify them into 3 main buckets

1. Build time – It does take you little longer than just going ahead and building a new dashboard or a report that you are being tasked with. That said, over time, you will avoid lot of duplication, redundancy, and build things that have a greater impact.
2. Coordination – The reality is you’ve multiple players in this model and you need them to operate towards a shared goal. So, it does take a good amount of communication within and across the teams around the purpose/vision of the data product itself. Everyone needs to be onboard and have a similar level of understanding.
3. Organization Bureaucracy – Depending on how the teams are situated, one might have to navigate across multiple leaders, especially when Data Engineering, Data Sciences are not reporting into the same leader. Again, this goes back to #2, if that’s done right, it minimizes the effect.

I always think about this as a flywheel. It takes a lot of energy expenditure to move the wheel at the start, but once you pick up the steam and rotate it, it takes its own course. Once you establish the systems and processes, the operating model requires less day to day oversight.

So, how can you implement?

This might be a cliched response, but I’d start with baby steps. I don’t think one needs to go complete 180 from wherever they are in the operating model. It is important to recognize the philosophy behind it than the model itself. The core behind product led model is to think from the lens of customer, impact, and long-range planning. It has less to do with how you are tactically set up and more with the shift in mindset itself.

Hope this was valuable. Let me know how your organizations are set up and the lessons you’ve learnt.