Data & Analytics

From Spreadsheet Chaos to One Dashboard: A Practical Guide

How to get from five disagreeing spreadsheets to one number everyone trusts, without buying enterprise software.

Nepal Digital Labs5 min read

Almost every growing business reaches the same point. Sales are in one spreadsheet, stock in another, payments in a third, and the figure the accountant uses does not match the figure in the sales meeting.

Nobody is doing anything wrong. The business simply outgrew the method. Here is how to get out of it.

Start with the decision, not the data

The most common mistake is beginning with "what data do we have?" That produces a dashboard with forty charts that nobody opens.

Start instead with the decisions you make regularly, and work backward:

  • Which products or services actually make money after costs?
  • Which customers are slipping away?
  • Can I afford to hire someone next quarter?
  • Which channel brings customers worth having?
  • What is owed to me, and how overdue is it?

Three questions answered reliably beats forty charts nobody trusts.

Fix the data before building anything

This is the unglamorous phase, and skipping it is why most dashboard projects fail. A beautiful chart built on inconsistent data is worse than no chart, because now people act on it.

Agree what each thing means

Does "sale" mean the order date, the delivery date or the payment date? Does revenue include VAT? Is a cancelled-then-rebooked order one sale or two?

These sound pedantic. They are the entire reason two departments report different numbers.

Write the definitions down. One page is enough, and it will settle more arguments than any software.

Fix the identifiers

The same customer appears as "Ram Traders", "Ram Traders Pvt Ltd" and "ram traders". To a computer those are three customers, and all your per-customer analysis is wrong.

Give every customer, product and location one ID that never changes, and use it everywhere.

Stop typing the same thing twice

Every place the same figure gets entered by hand is a place the numbers will eventually diverge. Identify these and remove them — one source, everything else reads from it.

Clean historical data once

Decide how far back matters. Usually two or three years. Clean that properly and draw a line.

Then build the smallest useful thing

Resist building everything. Pick the one question causing the most pain and answer it properly.

A first dashboard might be a single page with:

  • Revenue this month against the same month last year
  • Top ten customers by value
  • Outstanding receivables by age
  • One trend line over twelve months

That is enough to change decisions. Add to it once people are actually using it.

Choosing tools

You probably do not need expensive software.

Keep using Excel or Google Sheets if your data is small, one or two people use it, and the main problem is organisation rather than volume. Properly structured sheets with validation go further than people expect.

Move to a proper BI tool — Power BI is the usual choice — when several people need the same view, when data comes from more than two sources, when you need scheduled refresh, or when "who edited this?" has become a real question.

Add a database when spreadsheets are slow, when you need multiple people entering data at once, or when access control matters — for instance, when each regional manager should see only their own region.

We compared these in more detail in Power BI vs Excel vs Google Sheets.

Automate the assembly

The value is not the chart. It is never having to build the report again.

  • Data pulls on a schedule, not when someone remembers
  • Cleaning rules run automatically, the same way every time
  • The dashboard refreshes itself
  • The monthly summary generates rather than being typed

A report that takes two days to assemble gets made once a month. A report that builds itself gets looked at every morning, and that changes how a business is run.

Write down what the numbers mean

Alongside the numbers, provide a short written summary: what changed, why it probably changed, and what to look at. Three sentences a month. That is what turns a report into a decision.

A realistic sequence

  1. Week 1: List the decisions. Agree definitions. Write them down.
  2. Weeks 2–3: Clean identifiers, consolidate sources, remove duplicate entry.
  3. Week 4: Build the smallest dashboard that answers your top question.
  4. Month 2: Automate the refresh.
  5. Month 3: Add the second and third questions, now that people trust the first.

Most of the work is in weeks 1–3, and that is the part everyone wants to skip.

Key takeaways

  • Start from the decision you need to make, not the data you happen to have
  • Agreeing what each term means settles most reporting disputes
  • One customer ID used everywhere fixes more than any chart
  • Build the smallest dashboard that answers your biggest question
  • A written summary alongside the numbers is what makes it get used

If you want help getting out of the spreadsheets

This is most of what our data and BI work consists of: cleaning and modelling first, then a dashboard that refreshes itself, then a monthly summary in plain language.

Book a free data review and we will look at how you report today and tell you what could be automated.

Want this done properly, without the guesswork?

Tell us what you are trying to achieve and you will get an itemised quote — no obligation, no follow-up pressure.

Nepal Digital Labs

Websites, hotel presence and business data, from Kathmandu.

Have a question this did not answer?

Ask us directly. We reply in plain language — the same way we write.