- Karina Datascientist's Newsletter
- Posts
- The Tools Trap: What You Actually Need to Learn (And What You Don't)
The Tools Trap: What You Actually Need to Learn (And What You Don't)
Karina Samsonova
October 17, 2025
When I landed my first data analytics role in 2014, my interview consisted of two questions:
Can you create pivot tables?
Looking at this report, which marketing campaign performed best?
That was it. I got the job.
Fast forward 10 years. To land a junior data analyst role today, you're expected to be fluent in Excel, SQL, at least one visualisation tool, preferably one programming language AND have 3 years of experience.
For a junior role. Let that sink in.
It's frustrating, I know. But it's also reality—supply has increased dramatically in this field.
So today, let's cut through the noise. I'll share what tools I've actually used across different roles and what you should focus on if you're just starting out.
My Tool Journey: A Reality Check
Quick context for those who don't know my story: I transitioned to data from finance, where I'd spent 5 years and worked my way up to Head of FP&A.
Role 1: In-house Data Analyst
My job description said "strong Excel knowledge required." Great—I had that covered.
Week one reality check: I also needed VBA. Which I'd never touched before. So I learned it. On the job. While doing the job.
Then came SQL. Then SSRS. Then R. Then Power BI (which had literally just been released to the market).
Welcome to data analytics, where the required tools magically appear after you're hired.
Role 2: Consulting
Consulting taught me the golden rule: you use whatever tools your client uses.
My toolkit expanded to include:
Databases: SQL, BigQuery, Mixpanel, PostgreSQL, MongoDB
Spreadsheets: Excel, VBA, Google Sheets
Visualization: SSRS, Power BI, Tableau, DOMO
Programming: R, Python
Marketing/Analytics Platforms: Google Analytics, Meta Ads Manager, AppsFlyer, Firebase, Mailchimp, Klavio and many more niche tools (even Lotus Notes which is a grandfather of Outlook)…
Sounds impressive on a CV, right? Here's the truth: I learned most of these on the job, Googling frantically, checking StackOverflow and hoping nobody noticed.
Role 3: Current Position
Today I work with Databricks, Oracle, Cognos, Excel, Tableau, Power BI and a LOT of PowerPoint.
Notice what changed? Notice what stayed the same?
The 3 Core Skills That Actually Matter
After 12 years and countless tools, here's what you genuinely need:
1. Excel/Google Sheets
No matter where you end up, you'll use spreadsheets. Why? Because when you're working with non-technical stakeholders, they'll ask you to "just send it in Excel."
I've worked at massive corporations with cutting-edge analytics platforms. Know what executives still ask for? Excel files.
2. SQL (Some Flavour of It)
You'll work with some type of SQL. Whether it's BigQuery, MySQL, PostgreSQL, or Oracle—the core syntax is remarkably similar.
Learn the fundamentals: SELECT, WHERE, JOIN, GROUP BY, basic aggregations. The rest? Google it when you need it. Or ask ChatGPT. Nobody expects you to memorise every function. I still double check ROW_NUMBER and NTILE syntax every time I need to use it.
3. A Visualisation Tool
The specific tool varies by company:
Large organisations: Usually Power BI (because they're married to the Microsoft ecosystem)
Startups: Often Tableau (many startups prefer Macs, and Power BI doesn't play nice with Mac)
Big corporates: Sometimes both (like my current role, which is equal parts confusing and redundant)
Pick ONE to start with. Get comfortable with the concepts: connecting to data, creating charts, building dashboards, telling stories. These skills transfer between tools.
What About Python and R?
Here's my honest take:
R: Used less and less outside academia. If you're planning a career in research or universities, learn it. Otherwise, it's optional.
Python: Increasingly expected for senior roles. As a junior or mid-level analyst? Nice to have. As a senior? Almost mandatory.
But here's the key: focus on the fundamentals first. Don't try to learn everything at once.
The Skill Nobody Talks About (But Everyone Needs)
You know what matters more than knowing 10 different tools?
Communication.
Learn how to:
Present findings clearly in PowerPoint
Explain technical concepts to non-technical people
Tell a story with your data
Tailor your message to your audience
But here's something crucial that nobody teaches you: creating a data story in PowerPoint and presenting those findings to an audience are two completely different skills.
Your slides should be visual and clear—people can read them. Your job during the presentation isn't to read every bullet point or explain every single chart. That's painful for everyone involved.
Your task is to:
Give a compelling summary
Highlight the most important insights
Read the room and adjust based on audience interest
Go deeper on points that spark questions
Skip details that don't matter to THIS specific audience
You can't put every finding into PowerPoint anyway. And you shouldn't try. Your slides are the highlights reel. Your presentation is where you bring those highlights to life and adapt to what the room actually cares about.
I've watched too many analysts create beautiful 50-slide decks and then proceed to read every word on every slide. The audience checks out by slide 5. Don't be that person. (I was that person, learnt my lesson).
I've seen brilliant analysts who can write complex Python scripts but can't explain their findings to stakeholders. They struggle.
I've seen average technical analysts who can communicate brilliantly. They thrive.
Guess which one gets promoted?
The Certification Question
I get asked this constantly: "Do I need certifications on my CV?"
My honest answer after 12 years in data: probably not as much as you think.
In my entire career, I've seen exactly ONE job description (in Saudi Arabia) that required specific certifications. One. Out of hundreds.
Most companies care more about:
Can you do the work?
Can you show examples of your work?
Can you explain your thought process?
Are certifications beneficial? Sure, they can't hurt. If you have the time, money, and desire—go for it. They provide structure for learning and look nice on LinkedIn.
Should the absence of certifications stop you from applying? Absolutely not.
I don't have a single data analytics certification. Not one. Yet I've worked across three countries, built data teams, and consulted for major organisations.
Your portfolio projects, your ability to communicate, and your problem-solving skills matter infinitely more than a certificate that says you completed an online course.
My Advice for Beginners
If you're just starting out:
Master the fundamentals: Excel, SQL, one visualisation tool
Learn by doing: Build projects that solve real problems
Focus on storytelling: Practice explaining your findings clearly
Don't chase every shiny tool: Learn deeply, not widely
Apply anyway: Missing one "required" tool? Apply. You can learn it.
The job descriptions asking for 7 tools and 3 years of experience for junior roles? They're wish lists, not requirements. Apply anyway.
Remember: I learned VBA in my first week because I had to. You'll do the same with whatever tool your job requires.
The key isn't knowing everything before you start. It's being comfortable learning on the go.
Keep pushing 💪
Karina
AI News: When Hallucinations Hit the Big Leagues
In last week's newsletter, we talked about AI and the critical importance of fact-checking results because LLMs can hallucinate and make things up. link to last week’s newsletter
Well, here's a perfect cautionary tale:
Deloitte Australia delivered a 237-page report to the Australian government. The report was later found to contain numerous AI-generated errors, including fabricated references and a fake quote attributed to a federal court judgment. - article
Ouch.
Let me say it again for the people in the back: AI will not take your job. But if you ask AI to write a report for you without fact-checking—you might lose your job.
Python Tip
Everyone knows value_counts() — it's probably one of the first pandas methods you learned.
Power feature 1: normalize
Skip the manual percentage calculations by adding normalize = True. Perfect for quick distribution analysis without reaching for plotting libraries.
Power feature 2: dropna=False
Missing data tells a story. This instantly reveals data quality issues and missing data patterns that might affect your analysis.
Power feature 3: Instant Binning
Combine pd.cut() with value_counts() for quick distribution profiling.
Two lines give you a population profile that would normally require plotting
import pandas as pd
import numpy as np
# Basic usage
data = pd.Series(['A', 'B', 'A', 'C', 'A', 'B'])
basic_counts = data.value_counts()
basic_counts
ages = pd.Series([23, 25, 31, 31, 40, None, 25, 25, 57])
# Get proportions instead of raw counts
proportions = ages.value_counts(normalize=True).round(3)
proportions
# Include NaN in the analysis
full_distribution = ages.value_counts(dropna=False, normalize=True).round(3)
full_distribution
ages = pd.Series([23, 25, 31, 31, 40, None, 25, 25, 57])
# Quick bins + counts
bins = pd.cut(ages, bins=[0, 25, 35, 50, 80], right=False)
binned_counts = bins.value_counts().sort_index()
binned_counts
Grab your freebies if you haven’t done already:
Data Playbook (CV template, Books on Data Analytics and Data Science, Examples of portfolio projects)
Need more help?
Just starting with Python? Wondering if programming is for you?
Master key data analysis tasks like cleaning, filtering, pivot and grouping data using Pandas, and learn how to present your insights visually with Matplotlib with ‘Data Analysis with Python’ masterclass.
Grab your Pandas CheatSheet here. Everything you need to know about Pandas - from file operations to visualisations in one place.
 | Data Analyst & Data Scientist |