In short
A builder running several sites usually keeps four or five separate records (site diary, purchase file, hazri book, cheque register, accountant’s sheet) and none of them can total the others, which is why the net position is always a fortnight old.
Every builder we have sat with keeps good records. That is the part outsiders get wrong. The site diary is neat, the hazri book is current to yesterday evening, the cheque counterfoils are all there. Nobody is careless with a notebook that holds thirty lakh rupees of commitments.
The problem is never one book. It is that there are five of them, and none of them can read the others.
The five books, and what each one cannot do
Walk into most offices running two or more projects and you will find roughly this set, in some order:
- A site diary per project, holding deliveries, stage progress and instructions to the contractor.
- A purchase file: supplier bills, some paid, some part-paid, some still to be checked against what actually arrived.
- A hazri book carried by the site manager, usually one page per day, sometimes one page per week.
- A cheque register or a bank app, which knows the amount and the date but not the site.
- A spreadsheet the accountant keeps, updated whenever the paper reaches the office.
Each of these is correct on its own terms. The site diary is the truth about the site. The cheque register is the truth about the bank. But the question an owner actually asks, "across everything I am running, where do I stand today?", cannot be answered from any one of them, and answering it from all five means a person sitting down for two hours with a calculator.
So it gets asked once a month instead of once a day. And a number you look at once a month is not a control; it is a post-mortem.
What the gap actually costs
The losses here are rarely dramatic. Nobody steals a lorry. What happens is smaller and much more expensive over a year:
- A supplier is paid twice, because the bill was cleared at the site and again at the office, three weeks apart.
- A booking amount arrives and nobody chases the balance, because the to-receive figure lives in the broker’s head.
- A site that is quietly losing money keeps getting funded, because the profit on it has never been separated from the profit on the site next door.
- A wage rate is revised verbally in Ramzan and never revised back, and the difference shows up nine months later as a total that nobody can explain.
What a system adds that a book cannot
A system is not a neater notebook. If all it does is move the same entries onto a screen, it has added typing and removed nothing. Three things make the difference:
Notebook & spreadsheet
- Entries are recorded where they happen and totalled somewhere else, later.
- A changed figure leaves no trace beyond the overwriting.
- Per-site profit is a separate exercise somebody does at year end.
- The owner sees the position when the paper reaches the office.
One connected record
- The entry and the total are the same act: recording a payment moves the net position.
- Every create, edit and delete is kept with who did it and when.
- Per-site profit is a view of entries that already exist.
- The owner sees the position while the site manager is still standing at the gate.
The entry has to happen where the work happens
This is the part that decides whether any of it works. If the site manager has to remember the delivery until he reaches the office, the record is already a reconstruction. If he can record it on his phone while the driver waits for a signature, it is a fact.
Which is why the whole thing has to survive a site: one hand on the phone, dust on the screen, a network that comes and goes, and staff who are quick in Gujarati and slow in English. Software that assumes a desk and a good connection gets abandoned in a fortnight, and the notebook comes back. Correctly, because the notebook at least works.
The books were never the problem. Nobody was ever going to add up five of them every morning.
Where to start, if you are starting
Do not begin by moving history. Three years of old entries will take a month to enter and will not tell you anything you can act on. Begin from a date (the first of next month is fine) and enter forward from it. Put in the sites, the open balances as opening figures, and then let the ordinary week fill the system in.
By the end of the first month you will have something the notebook could never give you: one number, current as of this morning, that covers everything you are running.
Sitegride puts sites, sales, purchases, labour and money in one place, in English and Gujarati.