Bharat Enterprise · import of 23/10/2026 · tested against their own Miracle books (April 2022 – September 2024)
Same narration, different party
Some customers pay Bharat Enterprise through a co-operative bank. The co-op bank sends the money from its own pooled account, so the bank statement shows only the co-op bank's name and number. The narration is the same for every payer, and the user has to know who paid.
Every line where the narration was the same but the party was different
I grouped all 498 exported entries by narration, ignoring reference numbers. Only these 3 groups had the same narration but more than one ledger chosen. Each colour stands for one party.
Above: two lines from two different parties. Only the bank's own transaction number changes, and it says nothing about the payer.
Veraval Mercantile Co-op Bank · account 0230…0018 · 16 lines · 9 different parties
| Date | Received | Party the user chose | What the app did |
|---|---|---|---|
| 12/10/2026 | ₹1,14,739 | SHREENATHJI TREDERS-NANA MUVHA | No suggestion |
| 12/10/2026 | ₹2,29,478 | SHRI KRISHNA GUPTA CHANAWALA | ✕ Filled another party |
| 12/10/2026 | ₹2,32,155 | NISARG TRADING CO | ✕ Suggested another party |
| 12/10/2026 | ₹44,116 | MADHAV TRADING - (KESHOD) | ✕ Suggested another party |
| 13/10/2026 | ₹1,21,328 | THAKKAR DHIRAJLAL VANMALIDAS | Filled it, kept |
| 13/10/2026 | ₹57,369 | SAGAR SALES AGENCY | ✕ Filled another party |
| 14/10/2026 | ₹1,16,078 same amount, other party | NISARG TRADING CO | ✕ Suggested another party |
| 14/10/2026 | ₹2,42,560 | JAGANNATH PROTIEN. | ✕ Suggested another party |
| 15/10/2026 | ₹2,32,680 | SANJAY ENTERPRISE-UNA | ✕ Filled another party |
| 15/10/2026 | ₹2,29,478 | SHRI KRISHNA GUPTA CHANAWALA | Filled it, kept |
| 16/10/2026 | ₹2,32,155 | NISARG TRADING CO | Filled it, kept |
| 17/10/2026 | ₹1,16,078 same amount, other party | THAKKAR DHIRAJLAL VANMALIDAS | ✕ Filled another party |
| 22/10/2026 | ₹1,21,328 | THAKKAR DHIRAJLAL VANMALIDAS | ✕ Filled another party |
| 23/10/2026 | ₹1,16,078 same amount, other party | THAKKAR DHIRAJLAL VANMALIDAS | Filled it, kept |
| 23/10/2026 | ₹2,28,480 | MALAM RAMJI MAVJI - VERAVAL | ✕ Suggested another party |
| 23/10/2026 | ₹2,29,478 | SHRI KRISHNA GUPTA CHANAWALA | Filled it, kept |
Junagadh Commercial Co-op Bank · account 0101…0955 · 6 lines · 4 different parties
| Date | Received | Party the user chose | What the app did |
|---|---|---|---|
| 13/10/2026 | ₹63,026 | RAMESCHANDRA MAGANLAL AND SONS | No suggestion |
| 13/10/2026 | ₹59,509 | GOPAL ENTERPRISE | ✕ Filled another party |
| 16/10/2026 | ₹61,976 | N Kamlesh And Co. | ✕ Suggested another party |
| 17/10/2026 | ₹2,00,000 | VIRUMAL TOTAMAL AND BROTHERS | ✕ Suggested another party |
| 17/10/2026 | ₹50,000 | VIRUMAL TOTAMAL AND BROTHERS | Filled it, kept |
| 19/10/2026 | ₹35,453 | GOPAL ENTERPRISE | User took its 2nd suggestion |
Jamnagar People's Co-op Bank · account 0101…0516 · 3 lines · 2 different parties
| Date | Received | Party the user chose | What the app did |
|---|---|---|---|
| 14/10/2026 | ₹1,79,130 | SHIVAM ENTERPRISE - JAMNAGAR | No suggestion |
| 15/10/2026 | ₹68,250 | M.J TRADING CO. | ✕ Filled another party |
| 19/10/2026 | ₹68,003 | M.J TRADING CO. | Suggested, accepted in one click |
Left out of this table: 2 groups of transfers between their own HDFC accounts ("IBFUNDS TRANSFER…"). They look alike, but each carries the other account's number, so the app already tells them apart.
How I tested the ideas
This PC has a copy of Bharat Enterprise's Miracle books for April 2022 to September 2024: 13,927 sales bills and 12,948 bank receipts from customers. I replayed every receipt from 01/10/2022 in date order. For each one, I asked each idea to name the payer using only what the books held before that day, and then compared the answer with the party Miracle actually holds.
The numbers below are for the 622 receipts from the 15 parties who paid through co-op banks in this import. Results for all 11,834 receipts are at the end.
1. "Unpaid" is not a Miracle field I read. I settled each payment against that party's bill of the same amount, or else their oldest bills first. Miracle's own record of which payment settled which bill was not used.
2. Which receipts came through a co-op bank is not known. The books keep no bank narration. The 622 receipts are all payments from the 15 co-op parties, however they paid at the time. The examples below are shown as co-op lines only to make the idea clear.
Idea 1: Remember who paid each amount your idea
Good
- Simple, and it uses only what the user chose before.
Bad
- Wrong 5 times for every time it is right. Prices are fixed, so many buyers pay the same amounts.
- You were right that this cannot work alone.
Idea 2: Match the payment to an unpaid sales bill
Good
- It uses what really decides the payer: who owes this money.
- It works even when the amount was never seen before.
- A payment that clears the 2 or 3 oldest unpaid bills together is found too.
- Bills are read on the user's PC, like Miracle history is today. Nothing new goes to the server.
Bad
- The app must read sales bills, not only the bank ledger. That is new reading work.
- It needs Miracle to be up to date with sales. Bills typed after the import cannot help.
- Ties: when 2 buyers owe the same price on the same day, it can only list them.
Idea 3: Idea 2, keeping only parties seen on this co-op bank recommended
Good
- Most ties disappear: 14% becomes 3%.
- The co-op party list is learned from the user's own choices, so it grows on its own.
Bad
- The 80% is an upper estimate. I used the 15 parties known from this import. In real use, a new customer's first co-op payment falls back to Idea 2.
- A party who starts paying through the co-op bank is missed until the user chooses them once.
Recommendation
Build Idea 3: suggest the party with a matching unpaid bill, and prefer parties already seen on that co-op bank.
On these 15 parties' payments it was right 4 times in 5 and wrong about 1 time in 170. Your amount-memory idea was wrong 1 time in 3. When there is no match (part payments and round sums), the user chooses, as today.
The same check for every receipt, not only co-op lines
| Idea | Right | Short list | Nothing | Wrong |
|---|---|---|---|---|
| 1. Who paid this amount before | 7.4% | – | 53.3% | 39.3% |
| 2. Unpaid bill of this amount | 47.7% | 11.2% | 39.0% | 2.1% |
11,834 receipts. Named payers (NEFT with a name) are already found by name, so for them a bill match is a check: "₹79,750 matches the bill of 06/04". It could raise confidence, or warn when a name match has no such bill.
Also found while checking
✕ The ₹1,05,100 from "HARESHKUMARJAYANTILAL" (23/10/2026), accepted in the one-click Accept all, went to MANJIT HARESHBHAI NARVANI, which is an Unsecured Loans ledger. Their ledger list has a customer named HARESH KUMAR JAYNTILAL, and the books show bills to him. A bill check like Idea 2 would have caught this.
Source: Ledger Map's records of the 23/10/2026 import, and a read-only copy of the company's Miracle books on this PC. The books were not uploaded anywhere. Only the counts and the few examples above are shown.