Why your dashboard cannot answer the next question
Every finance team gets its answers one of 2 ways: an afternoon in Excel, or a dashboard that someone built last year. Both break on the same thing: the question nobody planned for. That is usually the one the owner asks.
Way 1: the afternoon in Excel
It is the 3rd of the month. The owner wants to know how October went. The finance head opens the ERP and downloads the sales register, the purchase register, the stock summary, the ageing, the trial balance and perhaps a cost sheet. 6 files. Then the real work starts: lookups to join them, pivots to cut them, a chart or 2, and a summary sheet that took longer to format than to think about.
In most companies we have worked with, this is still how the month is explained. It works. It is also slow, it depends on one person, and every number in it is a copy of a number in the books, one paste away from being wrong.
Way 2: the dashboard
The more organised companies have moved on. A consultant connects Power BI or Tableau to the ERP, builds a data model, and designs a set of dashboards: sales by region, margin by product, cash, debtors. They refresh by themselves. On the first day, it feels like the problem is solved.
Then the owner looks at the margin page and asks something the dashboard was not built for.
A dashboard answers the questions someone thought of when it was built. The question that matters this month is usually one nobody thought of.
One October, 4 questions
A pump manufacturer. Gross margin fell from 44.9 percent in September to 42.2 percent in October. The company has a good margin dashboard. Here are the 4 questions the owner asked that week, in order.
| The owner asks | The dashboard | What it takes without one |
|---|---|---|
| What was gross margin in October? | Answers at once | The trial balance |
| Why did it fall? | Shows margin by product; the reasons are left to the reader | A margin bridge in Excel: half a day |
| Which customers bought the pumps at the old price? | Not built for it | Invoices joined to the price list by hand |
| Which supplier bills carried the new copper price, and from when? | Not built for it | Purchase register, line by line |
The dashboard answered the first question well. The owner needed the other 3 to decide on a price.
None of these questions is unusual. The data for all 4 is already in the ERP. What is missing is a report designed in advance for each. Adding one means a request to the consultant, a change to the data model and a wait, by which time the decision has been made on instinct.
What each way really costs
| Excel from ERP reports | Power BI or Tableau | Controller in the ERP | |
|---|---|---|---|
| First answer | Hours, after month end | Instant, once built | Instant |
| A new question | Another afternoon | A change request and a wait | Ask it in plain words |
| Are the figures live? | No, copies | As fresh as the last refresh | Yes, from the books |
| Can a figure be traced? | Back to a pasted cell | Back to the data model | Back to the invoice or the bill |
| Who keeps it working? | One person in finance | A consultant or a BI team | The ERP itself |
How Controller answers the next question
Controller is the AI inside Niyantrak ERP. It does not need a report to be designed first, because it does not work from reports. It works from the books.
- The AI reads the question. Which customers bought the pumps at the old price? becomes: sales invoices in October, for the 2 pump products, at a rate below the new price list, grouped by customer.
- The accounting engine does the numbers. Every figure is computed from the posted entries, exactly, to the rupee. The AI never does the arithmetic.
- The answer shows its working. Each figure opens the invoices behind it, so the owner can check any number before acting on it.
The next question, and the one after it, work the same way. That is the difference between analytics that were designed and analytics that answer.
When a dashboard is still the right tool
To be fair to dashboards: they are still the best way to watch the same handful of numbers every week, to put a board pack on a screen, or to combine data from several systems outside the ERP. A good business may well keep a few. What it should not need is a consultant every time the owner is curious.
Niyantrak ERP: analytics without a dashboard project
Niyantrak ERP, the AI native ERP built in India by Niyantrak Consulting LLP, puts Controller inside the books, so the finance team and the owner can ask any question about the business and get an answer from the live accounts, with every figure traced to its entries. Every company deserves a Controller; not every company can afford one. Niyantrak ERP puts one inside the books. The cloud edition is coming soon.
Questions and answers
Why can a Power BI dashboard not answer a new question?
A dashboard shows the measures and cuts that were designed into it. A question outside that design needs a new report or a change to the data model, which takes a consultant or a BI team and time.
Is Excel still the most common way finance teams analyse ERP data?
In most small and mid sized companies, yes. The finance team downloads several reports from the ERP and joins them in Excel, which is slow, depends on one person and leaves every figure as a copy of the books.
Can an AI answer finance questions directly from the ERP?
Yes, if it is built into the ERP. In Niyantrak ERP, Controller reads the question, the accounting engine computes the answer from the posted entries, and every figure opens the invoices and bills behind it.
Does Controller replace Power BI or Tableau?
For questions about the business in the books, it removes the need to design a report first. Dashboards remain useful for a fixed weekly view or for data from systems outside the ERP.
How many questions does your month end leave unanswered?
Niyantrak Consulting LLP has built analysis for finance teams in Excel, in Power BI and inside ERPs, and has built Controller to answer the questions they could not. Tell us the 3 questions your owner asks most, and we will show you how each would be answered.