User-Centric Analytics

User-Centric Analytics

(Co-authored by Saksham Agrawal and Vikram Nayak )

The search for a North Star

Analytics teams have an identity issue. Are they engineers wielding a set of tools? Are they partners to the business teams? Or are they inventors of mathematical solutions??

This confusion stems from focusing too much on the solutions, and not enough on the problems. The fact is that analytics can’t exist in a vacuum. Unlike climbing Everest, analytics is never done just “because it’s there”. The most open secret in analytics is to use data to enable better business decisions and products. That is the only way that the analytics org can justify its existence.?

These decision-makers are your users. It is a process of iteration, of discovery, of negotiation with these users, that helps the modeler come up with a useful version of reality.

Users need to trust the data, understand what it’s saying and incorporate it into their workflow. Unless that happens, you feel like you're just spinning your wheels and no one really cares.?

“User-centric analytics” to the rescue

User-centric analytics is a way to combine the “hard science” of analytics (data engineering, statistical modeling, dashboards and reports), with the “soft science” of design thinking, product thinking, lean mindset, UX and data storytelling.

The outcome: analytics products that actually get used to help users make better decisions and take better actions.?When user adoption is the primary focus, then it makes sense to organize all your analytics efforts around delivering value to your users.??

A state of flow: The user-centric analytics organization?

In a user-centric analytics org, users are involved throughout the lifecycle - right from discovery to design and development, to deployment and testing. They feel like their inputs and feedback are taken seriously. They trust the data and understand how the insights have been arrived at. They feel like the data team understands their actual business problem.

The outputs are discoverable, useful, usable, understandable and actionable. Users know that they can rely on the analytics team to be agile and respond to the changing business environment. In other words, analytics drives the business forward without driving the business team crazy.

And perhaps most importantly, the user-centric analyst himself feels engaged and empowered, when users adopt their solutions and come back asking for tweaks.??

Helping the helpers: Does user-centric analytics benefit the analyst?

An emphatic yes, because analysts today are a troubled lot.?

There are more tools to learn than ever before. The modern data stack is only its latest incarnation. Analytics teams have to be great plumbers to move around all the data.

There is a wide variety of deliverables that an analytics team can produce. There are reports, and metrics, there are graphs and dashboards, there are filters and aggregations. Analytics teams have to be mindful about what shape the solution needs to take.?

There are judgments to be made about the data itself. How accurate, how comprehensive, and how detailed do you need to be? What assumptions can you make? What can be postponed and what must be done right away??

We believe that it is easier to navigate this complexity and come out on top, if we keep users at the center of our work.?

The elephant in the room: Generative AI

The immediate future will undoubtedly see an onslaught of Gen AI experiments, POCs and tools. There will be new considerations like observability (of business, systems and data), explainability (of insights, forecasts and predictions), bias and data security / governance.?

Will Generative AI impact how the analytics org functions? Definitely yes, but in a good way. The mundane tasks will be automated, and we’ll focus on the high-value parts of the analytics value chain that involve people, processes and solution architecture.

The Gen AI world is inevitably headed towards the "agentic" approach, where you have specialized agents doing specialized tasks. Think about specialized agents for researching personas, translating requirements into technical specifications, performing data preparation and statistical modeling, generating dashboards, finding insights, visualizing data, summarizing data and creating data stories containing the key insights.

But the real power will come from these agents being part of a mesh, communicating with one another. Reasoning, planning, executing and reflecting, while working within the constraints that you define.

We are entering the age of "Gen-AI-assisted Analytics"!

Achieving user-centric analytics

We don’t want to understate the “hard science” of extracting objective reality out of data. In fact, analytics teams are nothing but the custodians of this mathematical surety that their users come to depend upon. However, we wish to discuss simple and actionable ways in which the ideas of product management can help data teams get the most impact out of their efforts.?

In the coming articles, we’ll delve deeper into the analytics workflow, and break down how to achieve user-centricity in each part of it:

  • Incorporate a strong user discovery process
  • Make each analytics deliverable more user-centric - right from dashboards and reports to models and data warehouses
  • Implement last-mile strategies to boost user adoption and realize actual business value on all our analytics efforts


The next article in this series is coming soon. But in the meanwhile, if you'd like to speak with us about our vision for User-Centric Analytics, feel free to reach out to Vikram Nayak or Saksham Agrawal .

要查看或添加评论,请登录

社区洞察

其他会员也浏览了