Top 5 priorities of analytics, data, and business leaders

I used to run the Zulily store team. As a low-margin high-volume business, operational rigor was essential for us. We looked at dashboards every day and made rapid decisions about inventory, pricing, and marketing to capture customer demand. I loved how data-driven the company was. At the same time it was extremely painful to get the kind of operational visibility we needed from our dashboards.

Our solution: Line up an army of analysts to dive for explanations and answers in our data warehouse every day. This always bothered me. The desire for efficiency and self sufficiency with data ultimately became the inspiration for Falkon (more on this later).

To understand the problem better, I spoke with over 100 leaders (covering analytics, product, data engineering, marketing, sales, merchandising, revenue, customer success, and supply chain) across 46 mid-market and enterprise companies (retailers, B2B SaaS companies, productivity applications, financial services, airlines, hospitality, and entertainment). Through these conversations, I realized that there was a universal desire for the combination of operational rigor and visibility around data and metrics.

Here are the key learnings from these conversations, divided into three core groups: analytics teams, data engineering teams, and business teams. As you’ll see, each have their own priorities and challenges in data, but there’s a high degree of overlap and interdependency.

Analytics teams

Gatekeepers of insight

Top 5 priorities

1. Clean data and create BI tables/models from raw warehoused data
2. Build dashboards or create a self-service experience for business teams
3. Perform ad hoc analyses to help business teams answer “why” questions like “Why is churn going up?” or “Why is engagement going down?”
4. Find opportunities and recommendations for growth
5. Create weekly/monthly/quarterly business review reports to help teams understand the key drivers of business health

Top 5 challenges

1. Dealing with messy data, which actively blocks insight generation
2. Getting business teams to use the dashboards that were built for them
3. Managing an infinite backlog of business or stakeholder questions
4. Educating business teams on how to make sense of data
5. Storytelling with data — what’s the truth and what’s made up?

Reading between the lines

These are subjective observations I made from the subtext in our interviews.

1. 60% of the analysts had embarked on 3–6 month long data cleaning projects. They felt it was necessary to do this work upfront instead of making incremental progress. 40% were prioritizing time to insight over a desire for perfection and precision.
2. Analytics teams were weary of business team data illiteracy and felt that data was dangerous and easily weaponized. There was a strong desire to control how much access business teams were provided. This was in sharp contrast to a stated priority: To make business teams self-sufficient. Analytics teams see themselves as gatekeepers of insight, one of the most wanted outcomes for business teams.
3. Lack of desire to standardize insights across teams. This was perceived as “too much governance” and in the way of time-to-insight.
4. A love-hate relationship with interrupt-driven days. When a business question comes in from a senior leader, it immediately goes to the top of the pile. This is frustrating because it disrupts project work. However, paradoxically, it also makes analysts feel special and mission critical.

Data engineering teams 

Gatekeepers of data

Top 5 priorities

1. Enable product, data science, and analytics workflows
2. Build reliable infrastructure to transport, copy and clean data
3. Data quality and governance
4. Extract business logic out of tools and into company owned infrastructure
5. Create a self-service experience for analysts and data scientists so that every request doesn’t require a data engineer

Top 5 challenges

1. Data quality work gets de-prioritized but when something goes wrong, the data engineering team is held responsible, even though they’ve been pushing for data quality and governance the entire time.
2. Analytics and data science teams favor speed to insight over data governance and standardization of data models, metrics, and workflows across teams.
3. Too many copies and staging areas for the same data. This is exacerbated when working with external vendors.
4. Upstream data sources change, breaking ingestion or, worse, data interpretation downstream.
5. Backlog of data access requests is a mile long and continuing to grow

Reading between the lines

1. Data engineering teams are flushed with capital and buying power, and perceive themselves as the gatekeepers of data, one of the biggest assets for any organization.
2. Data engineering teams love trying new tools and are curious to adopt the latest and greatest technology.
3. There’s a choice fallacy between governance and speed to insight, yet this is the choice that data engineering teams feel they have to make every day.
4. There’s a strong desire to own business logic instead of locking it up in any paid tool. This particularly came up in conversations about BI logic and aggregation, with a preference for dbt in favor of Looker.

Business teams

Gatekeepers of business health

Top 5 priorities

1. Achieve business goals — churn, revenue, growth, conversion, ROAS
2. Get credit for achieving business goals by proving that this was the result of our team’s activities
3. Understand the Why behind business health so  success can be replicated and failure avoided
4. Don’t screw up royally with a critical business decision
5. Understand the impact of exogenous factors on their business success

Top 5 challenges

1. Anxiety about meeting business goals
2. Lack of ongoing access to insights from data — many more questions than answers
3. Lack of certainty about which activity drove business performance
4. Competitiveness with other teams over budget and credit
5. Managing team performance and ensuring that team activities are yielding results

Reading between the lines

1. Business teams care about directional correctness instead of accuracy
2. There’s a high level of conditioning that most business questions can’t be answered due to data messiness. Despite this belief, there is heavy reliance on dashboards.
3. Business teams are hungry for data, but don’t feel that they have the time or resources to learn new tools, remember more passwords, etc.
4. Perception of machine learning and AI is that it’s magical, unexplainable, untrustworthy, and yet extremely powerful.

Summing it up

The common ground across these three stakeholders was their shared belief that data can lead to higher quality decisions. The divergence was in their power dynamic relative to each other and who gets to gate-keep what aspects of this incredible asset that’s sitting in the warehouse.

These conversations, combined with our team’s experience being each of these personas, shaped the product and technology that has become Falkon.

While we’ve been heads down building Falkon the past year, we’re excited to begin sharing learnings, and opinions that we’ve formed through this blog.

Want more articles like this? Follow Falkon on LinkedIn.