Sujan Shrestha

Data analysis case studies

Each one follows the same structure: the question, the data, how I worked with it, and what it made possible. Switch between them with the sheet tabs.

One workforce dataset across three countries

Deurali HRM: attendance, leave, payroll and performance for a company with offices in the UK, Dubai and Nepal.

3countries
6access roles
2salary currencies, plus UK contractor invoices
1source of truth

Question	How do you report on people, hours and pay consistently when staff are spread across three countries, two currencies and different contract types?
Data	Shift-level attendance, leave records, salaried pay in NPR and AED, UK contractor invoice summaries, and performance reviews.
Method	Designed the data model in Google Sheets with an Apps Script web app on top. Every shift has a defined lifecycle, so attendance figures are complete rather than estimated. Role-based access across Admin, CEO, COO, CTO, Lead and Staff controls who sees what. Reviews require a self-assessment before the lead review, so both sides are always captured.
Outcome	Leadership gets attendance, leave and payroll summaries per country and per currency from one dataset instead of separate local spreadsheets.
Analyst lesson	The questions you can answer later are decided when the data model is designed. Clean structure at entry beats cleaning at analysis time.

Google Sheets, Apps Script, data modelling, access control, payroll reporting

Why weren't people taking their leave?

Team Hub: an HR and leave tracker built after the old employee records went missing.

12company leave days going unused
5departments
0reliable employee records at the start

Question	The team had 12 outstanding company leave days and wasn't using them. Nobody could say who was affected, because the employee information sheet had gone missing.
Data	Employee details, documents, key dates, and every leave request with its status.
Method	Rebuilt the dataset from scratch with a Google Form feeding a structured sheet. Leave requests go through a form too, with templated emails sent automatically whenever a request changes status. Alerts go out two days before birthdays and company events.
Outcome	Every leave request is now a record with a date, department and status, so leave uptake can be measured by person and department instead of guessed. The live demo below shows the kind of view this data supports.
Analyst lesson	Before you can explain a problem, you have to make it measurable. Here, the first deliverable was the data itself.

Google Forms, Google Sheets, Apps Script, process automation, HR reporting

Fixing the data before trusting the chart

Family Budget Tracker: income and expense tracking with a monthly comparison dashboard.

1silent data-type bug found
Monthlyincome vs expenses view

Question	Where does a household's money go each month, and is income keeping up with spending?
Data	Every income and expense entry with date, category and amount.
Method	Built a dashboard with a category breakdown and a filterable transactions table. When adding the monthly Income vs Expenses chart, the monthly totals came out wrong: Google Sheets was silently converting month labels such as "2026-07" into full dates, which broke the grouping. I wrote a normaliser that reads genuine dates first, falls back through several parsing strategies, and rewrites values that had already been converted.
Outcome	Accurate month-by-month totals, and a reusable cleaning pattern I now apply to every Sheets project.
Analyst lesson	A chart that looks plausible can still be wrong. Check data types before checking conclusions.

Data cleaning, data types, Google Sheets, Chart.js, dashboard design

An inbox that reports on itself

Gmail automation: triage, filing and a daily digest, with every run logged.

Dailydigest with calendar agenda
Everyrun logged to a sheet

Question	How do you stay on top of a busy inbox and know the automation is actually working?
Data	Incoming email, attachments, calendar events, and a run log.
Method	An Apps Script pipeline that triages the inbox, files attachments into Drive, and emails a daily HTML digest with the day's calendar. Each run writes a row to a Google Sheet, so there's a record of what was processed and when.
Outcome	A morning summary instead of manual sorting, and a log that shows whether each run succeeded.
Analyst lesson	Any recurring data process needs its own audit trail. If you can't see that it ran, you can't trust its output.

Apps Script, Gmail, Google Drive, Google Calendar, logging

Reading campaign data to steer marketing

GrowYourBNB, 2023–2025: as CTO and Operations Lead.

Weband campaign performance metrics
Landing pagesoptimised for conversion

Question	Which campaigns and pages bring in leads, and where should the next round of effort go?
Data	Website and campaign performance metrics, client records, and sales activity.
Method	Analysed site and campaign performance to inform marketing decisions, and used the findings to prioritise landing page and UX changes. Kept client, sales and marketing records organised so the numbers could be trusted and compared over time.
Outcome	Marketing and website decisions based on measured performance rather than instinct.
Analyst lesson	Analysis only matters when someone acts on it. I was both the analyst and the person making the change.

Web analytics, campaign reporting, conversion, CRM data

Live demo: leave uptake by department

The kind of view Team Hub data makes possible. Switch the measure and the chart, ranking and summary all update.

Sample data for illustration, not real employee records.

View the underlying data

Department	Staff	Allowance (days)	Taken	Still to take	% used

Data analysis skills and tools

Each skill links back to work above, so you can see it in use.

Spreadsheets	Google Sheets and Excel as working databases: structured data entry, validation, cleaning, and summaries.Seen in: Deurali HRM, Team Hub, Budget tracker
Data cleaning	Finding and fixing type errors, inconsistent formats, and silent conversions before analysis.Seen in: Budget tracker
Visualisation	Dashboards and charts that answer one question each, built with Chart.js, SVG and spreadsheet tools.Seen in: Budget tracker, live demo
Python	Back-end and analytics code, and generating Excel workbooks with openpyxl.Seen in: Code Himalaya role
Automation	Google Apps Script pipelines across Sheets, Forms, Gmail, Drive and Calendar, with logging.Seen in: Inbox automation, Team Hub
Web and marketing analytics	Campaign and site performance analysis, SEO, conversion-focused page changes.Seen in: Campaign analytics, Online Notes Nepal role
Communication	Turning findings into plain-language updates for clients and leadership.Seen in: every role below

Work experience

From building websites to running operations: each role put me closer to the data.

2023 – 2025

CTO / Operations Lead, GrowYourBNB
- Analysed website and campaign performance metrics to inform marketing decisions.
- Managed landing pages and UX improvements to support conversion goals.
- Kept client data, sales activity and marketing assets organised and reliable.
- Coordinated digital campaigns for brand visibility and lead generation.
2022 – 2023

Python Developer, Code Himalaya
- Wrote clean, maintainable Python for digital solutions and analytics.
- Built back-end components to improve application responsiveness and performance.
- Connected user-facing features to back-end functionality.
2018 – 2022

SEO Analyst & WordPress Developer, Online Notes Nepal
- Applied SEO practices to improve search visibility and enquiries.
- Designed and maintained business websites focused on usability and engagement.
- Gathered client requirements and delivered on deadlines.

Bachelor of Computer Information Systems (BCIS)

Medhavi College, Kathmandu, 2018 – 2023

Let's talk

I'm looking for a full-time data analyst role in the UK. If you have a team that needs someone who understands both how data is collected and what it means, I'd like to hear from you.

Email me