Advice for

Data Scientists and

Managers
Karn S ingh
Tailoring Your Data Presentation to Your Audience.
Think About Who Your Audience Is

Identify Your Audience:

• Technical Stakeholders (e.g., Engineering)

• Non-Technical Business Teams
• Executives

Tailor Your Approach:

1. For Technical Stakeholders:

• Use detailed, academic-style analysis

• Include technical details and methodologies
• Present findings in a structured, comprehensive manner

2. For Non-Technical Business Teams/Executives:

• Focus on key insights and implications

• Avoid deep technical jargon
• Connect data to business decisions and outcomes
Tailoring Your Data Presentation to Your Audience.
Effective Communication Framework

The Pyramid Principle (Barbara Minto):

Top-Down Structure:

• Start with the main idea or recommendation

• Follow with supporting arguments and data
• End with detailed evidence and methodology

Benefits of the Pyramid Principle:

1. Clarity and Focus:

• Ensures the main message is clear from the beginning

• Helps audience grasp the significance of data quickly

2. Logical Flow:

• Supports a structured, logical presentation of information

• Makes it easier to follow and understand the narrative
The Importance of Business Acumen for Data Scientists.

Why Business Acumen is Essential

Differentiate Beyond Technical Skills:

• Senior data scientists already possess strong technical

abilities
• Stand out by driving maximum impact for stakeholders

Maximizing Your Impact:

1. Understand Business Priorities:

• Deeply grasp the strategic goals and challenges of your

stakeholders
• Tailor your analytics solutions to directly support these
priorities

2. Scope Relevant Solutions:

• Develop analytics projects that address key business

problems
• Ensure your work aligns with overall business objectives
The Importance of Business Acumen for Data Scientists.
Developing Business Acumen

1. Engage in Company Meetings:

• Pay attention during Company All Hands and cross-team

meetings
• Note strategic priorities and connect them to your work

2. Practice Strategic Thinking:

• During planning cycles, ask how new projects relate to

high-level business goals
• Discuss connections (or lack thereof) with your manager

3. Ask "So What?":

• Always question the relevance and impact of your data
insights
• Ensure every data point has a clear, actionable takeaway

4. Ultimate Goal:
• Transition from Task Executor to Thought Partner:
• Move beyond simply fulfilling requests
• Shape the analytics roadmap collaboratively with your
stakeholders
 Be an Objective Truth Seeker
The Importance of Objectivity in Data Science

Avoid Cherry-Picking Data:

• Many people select data to fit pre-existing narratives

• Organizations often reward goal achievement over objectivity

The Role of Data Scientists:

• Less pressure to meet short-term goals compared to other

teams
• Unique position to maintain objectivity and push back against
biased narratives

Challenges and Benefits:

1. Pressure from Stakeholders:

• Stakeholders may request data that supports their

preferred narratives
• Short-term compliance might yield immediate approval
but harms long-term credibility
 Be an Objective Truth Seeker

2.Commitment to Truth:

• Prioritize data integrity and truth over convenience

• Promoting accurate narratives builds long-term trust and
respect

Long-Term Advantages:

1. Establishing Credibility:

• Stand out by being a reliable source of unbiased data

• Gain trust from executives who seek unfiltered insights

2. Building a Reputation:

• Being a truth seeker can be uncomfortable but positions

you as a trusted advisor
• Executives will rely on you for honest and objective
perspectives
 Data + Primary Research =

Combining Quantitative Data with Qualitative Insights

Complementary Nature:

• Quantitative data provides statistically significant insights

• Qualitative research uncovers signals missed by large datasets

Benefits of Integrating Primary Research:

1. Uncover Hidden Issues:

• Dive into closed-lost deal notes, customer conversations,

and support tickets
• Identify problems and root causes that structured data
may not reveal

2. Example Scenario:

• In a B2B SaaS business, data shows declining win rates for

Enterprise deals
• To understand the decline, engage with Sales reps,
examine their notes, and talk to prospects
 Data + Primary Research =
Process of Discovery:

1. Initial Exploration:

• Gather anecdotal evidence from various sources

• Initial findings may seem random and noisy

2. Pattern Recognition:
• Over time, patterns and common themes will emerge
• These insights often highlight issues not captured by
standard metrics

Practical Application:

1. Identify Trends Early:

• Use qualitative data to spot emerging trends before they
become statistically significant
• Gain a deeper understanding of customer needs and
challenges

2. Enhanced Decision Making:

• Combine data analysis with direct feedback for more
informed decisions
• Develop solutions that address both quantitative results
and qualitative insights
If the Data Looks Too Good to Be True, It Usually Is

Understanding Unexpected Metric Changes

Initial Excitement:

• Steep upticks in metrics often attributed to recent

actions, e.g., feature launches
• Temptation to quickly claim success

Common Causes of Misleading Positive Data:

1. Incomplete Data:

• Recent periods may lack full data points

• Metrics may level out as more data comes in

2. One-Time Effects:
• Temporary boosts, e.g., early January sales spikes from
holiday backlog
• Misinterpreted as sustained improvements
If the Data Looks Too Good to Be True, It Usually Is
Adopt a Skeptical Approach:

1. Healthy Skepticism:

• Question sudden positive changes in metrics

• Investigate thoroughly before drawing conclusions

2. Curiosity and Experience:

• Dig deeper into the data to understand the true cause

• Leverage past experiences to identify potential pitfalls

Practical Steps:

1. Verify Data Completeness:

• Ensure all data points are accounted for

• Re-evaluate metrics as new data is added

2. Consider External Factors:

• Assess if changes are due to temporary external influences

• Distinguish between short-term effects and long-term trends
 Be Open to Changing Your Mind
Evolving Opinions:
• New data can change initial interpretations and
recommendations

Challenges:
Fear of Looking Incompetent:
• Hesitation to change statements to avoid seeming unsure
• Risk of frustrating stakeholders with adjustments
Importance of Adaptability:
• Leaders like Jeff Bezos advocate for changing opinions
with new information
• Clearly explain why recommendations change
Benefits:
1. Demonstrates Intellectual Rigor:
• Openness to new data shows strong analytical skills
• Adapting reflects a commitment to accuracy
2. Builds Trust:
• Stakeholders will trust your recommendations if they are
based on the latest data
 You Need to Be Pragmatic
Challenges in the Analytics Realm:
Perfectionism:
• Training in scientific methods promotes ideal approaches
to analysis
• Desire for thorough, statistically significant results
Real-World Business Constraints:
1. Time Pressure:
• Need answers faster than experiments can provide
• Limited time to achieve statistically significant results
2. Data Limitations:
• Insufficient users for unbiased splits
• Incomplete historical datasets for comprehensive analysis
Pragmatic Approach:
1. Prioritize Actionable Insights:
• Help teams ship products and close deals with timely data
• Balance ideal methods with practical business needs
2. Focus on Impact:
• Deliver good-enough solutions that drive business forward
• Collaborate with stakeholders to meet their immediate
needs
Don't Burn Out Your Data Scientists with Ad-Hoc Requests
The Problem:
Burnout Risk:
• Full-stack data scientists often tasked with building
dashboards and ad-hoc data pulls
• High volume of these tasks leads to burnout and team
churn

Challenges in High-Growth Startups:

1. Limited Headcount:
• Hesitation to hire dedicated Data Analysts or BI specialists
• Preference for flexible, well-rounded Data Scientists

2.Task Overload:
• Ad-hoc requests and dashboarding tasks consume
disproportionate time
• Frequent interruptions hinder progress on strategic
projects
Don't Burn Out Your Data Scientists with Ad-Hoc Requests
Solutions to Mitigate Burnout:

1. Implement AI Chatbots:
• Field straightforward data questions automatically
• Reduce the volume of routine queries for data scientists

2. Basic SQL Training:

• Train team members on basic SQL (1-2 analysts per team)
• Use tools like Snowflake SQL AI Assistant or Gemini in
BigQuery to simplify data access

3. Self-Serve BI Tools:
• Enable autonomy and flexibility for users to obtain insights
• Tools like Omni improve self-serve analytics capabilities
 Common Misconception
Perception of Maturity:
• Companies believe moving data from spreadsheets to BI
tools indicates a strong data culture
• Dashboards are seen as essential for mature data practices

Reality Check:

Efficiency with Simpler Tools:

• Google Sheets can be more efficient for quick, ad-hoc
analyses
• Avoids the time-consuming process of scoping and building
complex dashboards

When to Use Google Sheets:

Speed and Flexibility:

• Ideal for quick insights without extensive development
• Facilitates fast decision-making and problem-solving
 Common Misconception
Strategic Use of BI Tools:
Critical Dashboards:
• Reserve robust BI tools like Tableau for high-impact, widely
used dashboards
• Use for decisions that are critical and hard to reverse

Encouraging Nimble Analytics:

Leverage Existing Tools:
• Use analytics capabilities in day-to-day software (e.g.,
Salesforce)
• Promote decentralized analytics for speed and agility
Resource Preservation:
• Prevent bottlenecks in BI tool usage
• Free up Data Science teams for more strategic tasks
Supporting Teams:
Basic Training:
• Provide basic SQL training, data modeling, and
visualization best practices
• Equip teams with the skills needed for efficient,
independent data analysis
WAS THIS POST USEFUL?

FOLLOW FOR
MORE!