Tally is very good software. It is used by millions of Indian businesses because it does statutory accounting well, your chartered accountant already knows it, and it has never pretended to be something else.
The problem is not Tally. It is that a growing business needs an operations system and is using an accounting one — because that is what was already there.
Eight signals
None of these is fatal alone. Three or more together usually means the business has outgrown it for operations.
- A parallel spreadsheet exists. Production keeps its own workbook for work orders, or despatch runs a separate sheet the transport desk maintains. This is the clearest signal there is: the sheet marks precisely where Tally stopped describing the work.
- Stock never matches the physical count. The gap has been absorbed for years by a monthly adjustment nobody questions.
- Nobody can say who changed a rate. There is no audit trail on the decisions that move money, because the system was not designed to carry one.
- Two people cannot work on the same thing. Work is serialised around who has the file or the screen.
- Month-end has quietly stretched. A close that used to take three days takes nine, and the extra six are reconciliation rather than accounting.
- Commitments are made from memory. Sales promises delivery dates against a picture of capacity that exists in someone's head.
- Reporting is an export cycle. The board sees Tuesday's position on Thursday, and nobody experiences that as a failure because the number arrives and looks authoritative.
- The rules live with a person. Pricing logic, credit policy and exception handling exist in formulas or habits that one individual maintains.
The mistake almost everyone makes
The instinct on recognising these signals is to replace Tally. That is usually wrong, and expensively so.
Tally is doing its actual job well: statutory books, GST returns, and a format your CA and auditor already work in. Ripping it out means changing how you operate and how you file at the same time — two risks with no reason to be taken together.
The shape that works for most Indian businesses is Tally stays, operations move. The new system owns quoting, production, stores, job costing and despatch, and posts summarised financial entries into Tally. Finance keeps a tool it trusts. The filings do not move on day one. And the operational data becomes clean enough that replacing the accounting layer later becomes a real option rather than a leap.
It is also the lower-risk sequence: if the operations system has a bad month, the statutory books are untouched.
Timing is not negotiable
Move at financial year-end, or at absolute minimum a quarter-end.
A mid-year migration splits the year across two systems. Every comparative report needs stitching, the audit gets a reconciliation exercise nobody planned, and the questions arrive exactly when finance has least capacity to answer them.
Work backwards from that date. Discovery, data cleansing and a dry run take the time they take, so the useful planning question is when to start rather than when to finish.
What actually decides the effort
Not the number of vouchers. The condition of the masters.
Tally data has predictable characteristics after a decade of use: the same material entered under several names because search did not find the existing record, ledgers created for one transaction and never reused, units of measure that vary by who entered them, and opening balances carrying an adjustment whose origin nobody remembers.
Profile that before anyone quotes. Duplicate rate on item and ledger masters matched on description rather than name, null rate on fields the new system will require, and whether stock reconciles today. It takes days and it changes the estimate.
Before you build anything
Run the fit-gap honestly. If a packaged product — ERPNext, Odoo, an industry-specific system — covers around ninety per cent of the requirement through configuration, buy it. The test is not a feature comparison; it is counting what would need customising rather than configuring, because customisation on a packaged product is paid once to build and again at every upgrade.
Custom earns its cost where the way you work is why you win — an unusual production flow, a credit policy no template represents, job work with scrap credits that behaves differently from anything a product assumes. Where the process is standard and simply outgrown, the product is the cheaper answer and we will say so.
The one-question version
Look at the spreadsheet that grew up beside Tally, and ask what it is compensating for. If the answer is a handful of reports, you need better reporting. If the answer is how you price, produce or decide credit, you need an operations system — and Tally can keep doing the job it has always been good at.