Raw material
Rows, columns, metrics, labels, categories, dates, and messy real-world context.
Session 6
AI for Analytics & Decision Making
Turn data into insights. Turn insights into decisions.
Intro
What is analytics?
The class starts with a simple sequence students can reuse on any dataset.
- Data
- Questions
- Patterns
- Decisions
Direction
Who is improving? What is declining? Where are we wasting money? What should change?
Value
A chart matters when it changes a budget, a lineup, a plan, or a management action.
Mini Project 1
Sales Performance Analytics
Use a company sales dataset to teach ranking, trend analysis, product analysis, regional analysis, and management recommendations.
Dataset
Salespeople, products, revenue, units sold, dates, regions, and customers.
Open Sales SheetWho is the best salesperson?
Compare revenue, units sold, customers, average order value, and recent momentum.
Who is improving month-on-month?
Ask AI to group by month, calculate growth, and separate one-time spikes from real improvement.
Which products are selling best?
Analyze revenue, volume, repeat purchases, and product-region combinations.
Which region is strongest?
Compare total performance, growth rate, customer base, and underperforming pockets.
What would you do next month?
Convert the spreadsheet into hiring, coaching, territory, pricing, and product recommendations.
Mini Project 2
Cricket Performance Analytics in Google Sheets
A large batting dataset makes the lesson feel fun, familiar, and decision-driven.
Dataset
50 Indian batters, 12,000 rows, runs, balls, strike rate, bowlers faced, dismissal type, format, opponent, and venue.
Open Cricket SheetAI inside Google Sheets
Show how AI or Codex can inspect a Sheet, generate formulas, build summaries, and explain results.
Best player analysis
Most runs, best strike rate, most consistent batter, and best T20 performer.
Bowler weakness analysis
Which bowlers trouble a batter, which bowling type reduces strike rate, and who dismisses them most.
Team selection dashboard
Select a T20 batting lineup using recent form, strike rate, consistency, and match impact.
Stats to selection
AI can move from descriptive statistics to selection decisions.
Cricket Dashboard
From scorecards to selection committee thinking
The dashboard should not only say who scored runs. It should explain who deserves a place and why.
Decision criteria
- Recent form
- Strike rate
- Consistency
- Match impact
- Role fit
Best T20 batting lineup
Ask AI to rank players, explain trade-offs, flag risk, and produce a final batting order with roles.
Select the best T20 batting lineup based on recent form, strike rate, consistency, and match impact. Explain your assumptions and show the top alternatives.
Mini Project 3
Meta / Facebook Ads Analytics
Use a large ads dataset in Google Sheets, or connect through Meta MCP in Claude if available.
Dataset
Campaign, ad set, audience, spend, impressions, clicks, CTR, CPC, leads, CPL, applications, revenue, and ROAS.
Open Meta Ads SheetWhich campaigns should we stop?
Identify high spend, weak conversion, low ROAS, poor audience fit, and repeated underperformance.
Which campaigns get more budget?
Find campaigns with strong ROAS, low CPL, scalable audiences, and consistent lead quality.
Who is performing best?
Compare audience groups by CTR, CPC, CPL, applications, revenue, and downstream quality.
Where are we wasting money?
Separate bad creative, bad audience, bad landing page, and weak conversion follow-up.
Where should Rs. 10 lakh go?
Build a next-month media plan with budget allocations, risks, and expected outcomes.
AI + Excel / Sheets Tricks
Seven high-impact demos to show quickly
These prompts make spreadsheet analysis feel immediately useful, even for students who are not data specialists.
Executive summary
Explain this spreadsheet to a CEO in 5 bullet points.
Anomaly detection
Find unusual patterns, outliers, or suspicious rows in this data.
Dashboard builder
Design the ideal dashboard for this dataset.
Forecasting
Forecast the next 3 months based on historical data.
Story, audit, manager mode
Turn the dataset into a management story, ask auditor questions, and recommend what to do next.
Student Outcomes
Students leave knowing how to convert raw data into decisions
Analyze CSVs and spreadsheets
Open data, understand columns, clean obvious issues, and ask structured questions.
Ask business questions
Frame analysis around decisions instead of only summaries.
Find top performers
Rank people, products, campaigns, audiences, and players by relevant metrics.
Detect trends and anomalies
Spot growth, decline, outliers, and suspicious data points.
Build dashboard ideas
Define the charts, filters, and views a manager would need.
Write the decision memo
Use AI inside Sheets or Excel to turn marketing, sales, or performance analysis into action.
Data Source Connections
Use MCPs to connect AI with real data sources
MCPs let AI tools read from approved apps and files, so students can analyze live business context instead of copy-pasting everything manually.
Model Context Protocol
A safe connector layer that gives AI access to selected tools, files, databases, and business systems with user permission.
Connect, choose, ask, verify
Connect the app, choose the exact file or source, ask a focused question, then verify the result against rows, records, or source links.
Google Drive / Sheets MCP
Analyze spreadsheets, documents, slide decks, CSV imports, student notes, campaign exports, and dashboard data.
Supabase, Postgres, GitHub
Query databases, inspect app data, review code, read issues, summarize PRs, and connect product analytics with implementation context.
Gmail, Slack, Notion, Linear, Jira
Turn emails, meeting notes, tickets, tasks, and docs into summaries, decisions, follow-ups, and action trackers.
Meta Ads, HubSpot, Stripe, Analytics
Analyze campaigns, audiences, leads, revenue, payments, funnels, and customer behavior when approved connectors are available.
Subscription Reality Check
Common paid tools and what they cost
Use this as a quick cost reality check before students start subscribing to every shiny AI tool.
| Tool | Typical subscription value |
|---|---|
| ChatGPT / Codex | Plus $20/mo; heavier Pro tiers from about $100/mo |
| Claude | Pro $20/mo; Max from $100/mo |
| Perplexity | Pro about $20/mo |
| NotebookLM / Google AI | Free tier; advanced AI features may sit inside paid Google plans |
| Genspark | Paid plans vary; check current credit pricing |
| Supabase | Free tier + Pro about $25/mo per project |
| Vercel | Free tier + Pro $20/mo plus usage |
| Canva | Free tier + Pro about $15/mo |
| Gamma | Paid plans commonly start around $10-$20/mo |
| Loom | Business $18/mo; Business + AI $24/mo |
| HeyGen | Creator $29/mo; Pro $49/mo |
| Lovable | Often around $20-$30/mo to start, higher for more credits |
| Emergent | Credit-based builder pricing; verify before buying |
| Teal | $29 every 30 days |
| Granola | Free tier + Business $14/mo |
| Mem.ai | Free tier + paid knowledge-base plans; verify current pricing |
| Kling AI | Video credit plans vary; can get expensive with heavy use |
| OpenArt | Essential $14/mo; Advanced $29/mo |
| Supermeme | Paid meme-generation plans vary; verify current pricing |
Smart Buying Strategy
Do not subscribe to tools. Subscribe to outcomes.
If a student pays for every tool mentioned in the course, the stack can easily cross $350-$500/month before taxes, credits, hosting overages, and team seats.
One tool for every task
Pay for separate research, writing, design, video, code, resume, meeting notes, database, deployment, and analytics subscriptions.
One strong AI assistant first
Use Claude or Codex well for planning, prompting, writing, analysis, coding support, debugging, resume help, content repurposing, and workflow design.
Only when the output needs it
Pay separately when you need production hosting, avatar video exports, high-volume image/video credits, team collaboration, or commercial-grade templates.
Recommended starting point
One main AI subscription + free tiers of NotebookLM, Google Sheets, GitHub, Supabase, Vercel, Canva, Loom, and occasional paid credits only when needed.
When building seriously
Claude/Codex + GitHub + Supabase + Vercel. Add Lovable only if it clearly speeds up your build workflow.
When publishing seriously
Claude/Codex for scripts, captions, posts, carousels, and strategy. Add Canva, Loom, HeyGen, Kling, or OpenArt only for final asset production.
Course Closing
You now have the AI foundation to learn, build, and stand out
This course started with tools and ends with judgment: use AI to study smarter, build real projects, grow your career, create visible proof, and make better decisions.
Use AI as your study partner
Research faster, understand difficult concepts, summarize notes, and build a personal learning system.
Create proof of skill
Turn ideas into apps, dashboards, automations, content, and portfolio projects people can actually see.
Use AI for career momentum
Improve resumes, prepare interviews, analyze jobs, publish insights, and build professional relationships.
Move from answers to judgment
Use AI to analyze data, compare options, detect patterns, and recommend what to do next.
Choose your stack wisely
One strong AI assistant used well can replace many shallow subscriptions and scattered workflows.
Next Step
Build your project and submit it directly to me
Your final project is the proof that you can use AI to create something useful, clear, and shareable.
Submit by 30-Jun-2026
Complete your project and send the final submission directly to me before the deadline.
piyush@admitkard.com
Email your final project submission to this address.
We will organize a demo
Once students have submitted their projects, we will organize a demo so everyone can showcase what they built.
Completion certificate
On submission, you will receive your course completion certificate.