Insights · Data
What a business dashboard really costs a Nepali SME
The short version
A one-page dashboard for a Nepali SME usually costs more in setup hours than in software. Tool licences run from zero to about US$90 a month for a five-person team. The real cost is two to six weeks of connecting, cleaning and agreeing on numbers. It pays back when someone is spending hours every week building reports by hand, and not otherwise.
Where does a Nepali SME's data already live?
Almost every business we talk to already has more data than it uses. It just sits in five places that don't talk to each other.
- Billing or POS software, which records every sale. Since April 2026, businesses above Rs 200 million in turnover must bill through IRD-certified software linked to the government's CBMS system, so that data is already structured.
- Accounting books in Tally or a similar package, holding receivables, payables and expenses.
- Bank statements, including QR settlements. Merchant QR codes in Nepal went from 282,000 in mid-2021 to 2.34 million by January 2024, so a lot of sales now land in the bank as digital records.
- Google Sheets and Excel files for stock counts, targets and anything the software doesn't cover.
- The owner's head, which is the most up-to-date and least shareable source of all.
None of this needs new software to collect. It needs a way to put it side by side.
Why can't owners see it, then?
Because turning it into a picture is manual, slow and usually one person's job. The accountant exports sales, copies bank lines into a sheet, fixes the names that don't match, and emails a summary. That happens at month-end. So by the time you see that a product stopped selling or a customer stopped paying, it has been happening for three weeks.
This isn't only a Nepal problem. McKinsey's research on knowledge workers found they spend about 19% of the working week searching for and gathering information. Anaconda's 2020 survey of data professionals put loading and cleaning data at about 45% of their time. If trained analysts lose that much to preparation, a busy accountant doing it on the side is unlikely to do better.
There is also a quality problem. Raymond Panko's review of fifteen years of spreadsheet research concluded that spreadsheet errors are common and non-trivial. A monthly report rebuilt by hand gets fresh chances to be wrong every month. And when the one person who understands the file goes on leave, the report simply doesn't happen.
What does a dashboard project actually involve?
Five steps, and only the fourth is about charts. First you connect: get a copy of each source out on a schedule. Then you clean: make names, dates and codes consistent. Then you model: write down, once, what each number means. Then you visualise: put six to ten of those numbers on one page. Finally you set a refresh schedule so it updates without anyone touching it.
The step people underestimate is modelling. It sounds technical, but it's really a conversation. Does a sale count when the bill is raised or when the money arrives? Are returns taken off on the day they happen or against the original sale? Two partners with different answers will get different numbers from the same data. Settling that is most of the value.
How does the pipeline fit together?
Here is the shape we'd use for a typical trading or retail business with a billing system, Tally, a bank account taking QR payments and a few working sheets.
From billing software to one page, refreshed every morning
Tap any box. The chart at the end is the smallest part of the job.
Tap any component above for its role and the real tech.
Cleaning figure: Anaconda, 2020 State of Data Science. IRD billing threshold: ICT Frame, Apr 2026.
- Billing / POS software (External, IRD-certified billing, CSV or database export): Every sale is already recorded here, often with item, branch and time. Businesses above the IRD turnover threshold must bill through certified software that reports to CBMS, so the data is structured. The job is getting a copy out on a schedule, not typing it in again.
- Accounting books (External, Tally or similar, ledger export): Receivables, payables and expenses. Usually exported as Excel or XML. Ledger names rarely match the billing system, which is why the cleaning step exists.
- Bank and QR statements (External, Statement downloads (CSV / Excel)): The cash truth. QR settlements arrive here, often in batches, so matching them to sales takes rules, not eyeballing. A scheduled statement download is the realistic starting point for most SMEs.
- Google Sheets and Excel (External, Stock counts, targets, manual logs): Whatever the business tracks by hand: stock, targets, staff logs. Keep what works, but give each sheet a fixed layout so the pipeline can read it without a person fixing columns first.
- Scheduled import (Service, Connector or small script, runs nightly): Pulls each source on a timetable and lands a raw copy. Nothing is edited here, so if a number looks wrong later you can trace it back to the file it came from.
- Cleaning rules (Service, SQL or script: names, dates, duplicates): Merges 'Ram Traders' and 'Ram Traders Pvt Ltd', converts Bikram Sambat dates, drops duplicate rows and flags anything that fails a check. Anaconda's 2020 survey found data professionals spend about 45% of their time loading and cleaning data, and small-business data is no tidier.
- Business model tables (Data, A small database (PostgreSQL, SQLite or the BI tool's own store)): One agreed definition per number: what counts as a sale, when revenue is recognised, which branch owns which customer. This is where most of the arguing happens, and it should.
- One-page dashboard (Client, Power BI, Data Studio or Metabase): Six to ten numbers, each with a trend and a comparison. The tool matters less than people think; licence fees are the smallest line in the budget.
- Owner and managers (External, Phone or laptop, plus a morning email): The people who act on it. A scheduled email or snapshot often gets more use than a login nobody remembers.
The four boxes on the left already exist in most businesses. A dashboard project is mostly the middle column.
A small business doesn't need a data warehouse for this. A scheduled import, a set of cleaning rules and a small database are enough. The same pattern scales up, as in our executive KPI dashboard blueprint, but most SMEs never need the bigger version.
What does it cost in hours and licences?
Build effort is where the money goes. For three or four sources and one page of numbers, our planning range is roughly 54 to 174 hours. The wide range is almost entirely down to how messy the data is and whether the sources can be exported automatically.
A one-page SME dashboard, line by line
Planning ranges for 3-4 data sources and one page of 6-10 numbers. Sort by hours to see where the time really goes.
| What pushes it up | |||
|---|---|---|---|
| Agree the numbers and their definitions | 6 h | 16 h | Partners who define 'sale' or 'margin' differently |
| Connect sources (billing, books, bank, sheets) | 8 h | 30 h | No export option; manual downloads; many branches |
| Clean and match data | 16 h | 60 h | Inconsistent customer and item names; BS/AD dates; QR settlement matching |
| Model tables (one definition per number) | 8 h | 24 h | Several price lists, returns, credit sales |
| Build the one-page view | 8 h | 20 h | Mobile layout; per-branch or per-role views |
| Scheduled refresh and checks | 4 h | 12 h | Sources that cannot be pulled automatically |
| Training and first-month fixes | 4 h | 12 h | More than one person using it |
Totals: roughly 54 to 174 hours, which is two to six calendar weeks once waiting for exports and sign-offs is included. Hours are NeuralYug planning ranges for scoping, not a quote or a measured benchmark. The size of the cleaning line reflects Anaconda's 2020 State of Data Science finding that about 45% of data professionals' time goes on loading and cleaning data.
'Build the one-page view' is one of the smallest lines. The chart is not where the money goes.
Licences are the smaller line. As of August 2026, Power BI Pro is US$14 per user per month, up from $10 before April 2025. Google's free report builder, renamed from Looker Studio back to Data Studio in April 2026, costs nothing, and its Pro tier is $9 per user per project. Metabase is free to self-host, with a hosted Starter plan from about $90 a month for five users on yearly billing.
Monthly tool licence for a 5-person team
US$ list prices before VAT, card fees and exchange margin. Hosting for self-run options is extra.
Metabase Starter is shown at its yearly-billing rate with 5 users included; monthly billing costs more. Power BI Pro has been $14 per user since April 2025. Data Studio Pro is $9 per user per project. Microsoft Power BI Blog (Nov 2024) and The Register (Apr 2025); Google Cloud Data Studio documentation and release notes (Apr 2026); Metabase pricing page.
| Category | US$ per month, 5 users |
|---|---|
| Data Studio (free) | 0 USD |
| Metabase open source | 0 USD |
| Data Studio Pro | 45 USD |
| Power BI Pro | 70 USD |
| Metabase Starter (cloud) | 90 USD |
Even the priciest option here is small next to the 54-174 build hours above. Pick the tool last.
Those are dollar list prices. A Nepali business also pays VAT, a card or bank fee and the exchange margin, and needs a dollar-capable card with enough limit for a recurring charge. That's one reason free and self-hosted options are popular here. Choosing between the tools deserves its own comparison, so we'll leave it there.
Where does the payback come from?
Three places, in order of how easy they are to count.
- Hours of manual reporting you stop spending. This is the one you can measure before you start.
- Faster decisions. A slow-moving product or a branch having a bad week shows up in days, not at month-end.
- Problems caught earlier. Unpaid credit sales, stock that isn't moving and a tightening cash position all become visible sooner, when they're cheaper to fix.
Do the first one as arithmetic. Say your accountant spends six hours a week putting together sales and cash reports. That's roughly 300 hours a year. If a dashboard build takes 100 hours and cuts the manual work to one hour a week, the build hours come back in about five months, before counting anything the faster numbers help you catch. Plug in your own hours. If the answer comes out in years rather than months, that's telling you something. Our post on how automation ROI actually works walks through the same method in more detail.
When is a dashboard not worth it?
More often than vendors admit. Skip it, or wait, if any of these are true:
- Nobody spends more than an hour or two a week on reports now. There's nothing to save.
- Your sales and stock aren't recorded in software yet. Fix the recording first; a dashboard over guesses is just a prettier guess.
- You can't name the decisions it would change. If the numbers won't change what you do on Monday, they're decoration.
- The business is changing shape fast, with a new product line or a merger. Definitions will move, so wait until they settle.
In all of these cases a cheaper first step is a fixed-layout spreadsheet that one person updates weekly. It shows you which numbers people actually look at, which is exactly the brief a proper dashboard needs later.
What should you do this week?
List every report someone in the business builds by hand, how often, and how long it takes. Write down the three decisions you most wish you could make sooner. Check whether your billing software and bank can export a file on a schedule. With those three answers you can tell in an afternoon whether a dashboard would pay for itself, and roughly how big the job is.
If you'd like a second opinion on those answers, talk to us. Our data visualisation service starts with exactly that scoping, and we'll tell you honestly if a spreadsheet will do.
Sources
Frequently asked
How much does a business dashboard cost for a small company in Nepal?
Tool licences are the small part: from zero for Data Studio or self-hosted Metabase to about US$70 a month for five Power BI Pro users, as of August 2026, before VAT and card fees. The bigger cost is build effort, typically 54 to 174 hours for three or four data sources and one page of numbers, depending mostly on how messy the data is.
How long does it take to build an SME dashboard?
Usually two to six calendar weeks for a one-page dashboard fed by billing software, accounting books, bank statements and a few spreadsheets. Much of that time is waiting for exports and agreeing on what each number means. The chart itself is often done in a few days once the data is clean and modelled.
When is a dashboard not worth building?
When nobody currently spends more than an hour or two a week on reports, when sales and stock aren't yet recorded in software, when you can't name the decisions it would change, or when the business is changing shape too fast for definitions to settle. In those cases a fixed-layout weekly spreadsheet is the cheaper and wiser first step.
Want this run on your numbers?
We'll do the same analysis on one of your workflows in the two-week Automation Sprint.
Related service · Data & Analytics