Customer review · prepared 23/10/2026 from the live server's records
Bharat Enterprise: their first import with Ledger Map
Customer: Bharat Enterprise. Their PC was approved on 23/10/2026 at 17:57. Everything below happened in that one session, on the company BHARAT ENTERPRISE (CMP0001), year 2026–27 (YR29).
What they did, step by step
All times are Indian time on 23/10/2026. The bar shows the 82 minutes from upload to export, to scale.
- Signed in on the new PC. Brought in 5 Miracle companies, starred 3 and opened 3. Then took a short break.
- Dropped 2 password-protected HDFC PDFs: the current account …2994 (55 pages, 646 lines) and the cash credit account …8205 (6 pages, 74 lines). They typed each password once. Both PDFs were read in under 4 seconds, and every running balance agreed.
- Confirmed the bank ledger for each PDF. The app checked Miracle's books and found 222 lines already booked, so it left them out.
- Contra step (78 seconds): the app had already paired 1 transfer by itself. The user accepted 8 more suggested pairs, one after another.
- Choose ledgers (68 minutes): they worked through the list. They opened the ledger search 91 times and chose from it 79 times. They went back to Contra once and accepted 1 more pair.
- Review (10 minutes).
- On Export, they pressed "Accept every suggestion listed", which saved 119 entries in one click. Then they exported 488 vouchers in 2 files, numbered LM-01 to LM-50 and LM-01 to LM-438. Nothing went to Suspense silently.
Small issues during the session: "Miracle is updating its files. Retry shortly." appeared 3 times, because Miracle was open and saving. There was one harmless screen-resize message. Nothing crashed.
Who decided each of the 498 entries
Each square is one exported entry. Hover over or tap a square to see it.
- 271 matched by name or bank account number, and the user left all of them
- 2 entries = 1 contra transfer paired by its bank reference
- 2 filled from a payee the user had just chosen
- 119 in one click on Export
- 18 entries = 9 contra pairs
- 9 one at a time
- 8 picked the 2nd or 3rd suggestion
- 31 where the app had no suggestion (small UPI costs, salaries, insurance, school fees, some parties)
- 5 put in Suspense (₹3.36 lakh)
- 26 where the suggestion was wrong (25 were red, below 70%)
- 7 where the app had filled the ledger by itself, all of them co-op bank transfers (see below)
Show as a table
| Who decided | Entries | Amount |
|---|---|---|
| App alone | 275 | ₹15.52 crore |
| User accepted a suggestion | 154 | ₹3.25 crore |
| User chose themselves | 36 | ₹0.76 crore |
| User changed the app's answer | 33 | ₹0.67 crore |
| Total exported | 498 | ₹20.20 crore |
How good were the suggestions?
Each suggestion carries a coloured mark. This is what the user did with the suggestion shown at the moment they decided. Contra pairs and the app's own matches are not counted here.
✓ Green marks were right 76 of 77 times, and yellow marks were right 25 of 26. ✕ Red marks were right only 23 of 60 times. The colours did their job.
42 more entries had no suggestion at that moment. The user chose 31 of them and sent 4 to Suspense. The other 7 were settled in other ways.
Where the app went wrong, and the user put it right
1. Co-op bank transfers all show the same account number
Payments that come through Veraval Mercantile, Junagadh Commercial and Jamnagar People's co-op banks print only the bank's own pooled account number, not the payer's name. The app treated that number as one party.
Numbers: 25 such lines. The app filled 10 lines from a payee the user had chosen, and the user changed 8 of them. 4 of its account-number matches were also changed. The amounts often repeat for each payer (for example, ₹2,29,478 is always Shri Krishna Gupta Chanawala), so the amount says who paid better than the number does.
2. The company's own name in the narration
Incoming NEFT and RTGS lines name the receiver, which is Bharat Enterprise itself. The app took that as a clue to the ledger "BHARAT ENTERPRISE SALAYA".
Numbers: this was the first suggestion for 37 entries. The user corrected 14 by hand. After their choices, the app changed most of the others to the right sender before the one-click accept, and none was saved to the company's own name.
Contra entries: transfers between their own two accounts
Each transfer is one line in each PDF, so it becomes one voucher, not two. 10 transfers were found, worth ₹15.57 lakh.
✓ The app paired 1 transfer by itself and suggested the other 9. The user accepted all 9, and rejected no suggestion. It took them 78 seconds.
✕ One transfer the app could not pair: on 23/10, ₹1,30,000 came into the current account. Its narration, "IBFUNDS TRANSFERCR-50200025098205", says it came from their own cash credit account. But the cash credit PDF's last line is on 21/10, so that account's side of the transfer is not in this import, and Miracle's books do not have it either. With no line to pair it with, the app never showed it in the Contra step. It stayed in Choose ledgers, where the app suggested "BHARAT ENTERPRIZE DHORAJI" (53%, the company's own-name problem). The user typed "SUS" and chose Suspense.
Better choice: the ledger HDFC CC A/C-50200025098205 itself. Miracle would then hold the whole transfer, and when the next cash credit statement is imported, its side would be found in Miracle and left out. The app could suggest this, because the narration contains the exact account number that is in that ledger's name.
The statements compared with Miracle's books
Before anything was exported, the app read each bank's ledger in Miracle on the user's PC and looked for every statement line by date and amount.
✕ Gap found: Miracle has September for the cash credit account, but 10 lines from 19/09 to 29/09 were never booked. They are ₹3.09 lakh received and ₹2.19 lakh paid, including 3 transfers with BHARAT INTERNATIONAL (₹74,000, ₹60,000 and ₹85,000). The app found them and exported them with the rest.
✓ Everything else agreed: the 222 lines already in Miracle matched exactly on date and amount. Both PDFs passed every balance check, with 0 reading errors.
The app does not send Miracle balances to the server, so this review cannot compare closing balances. It can only compare line by line.
How much they gained
For 720 lines: 222 skipped as already booked, 498 booked, 488 vouchers written. About 10 seconds per entry.
If each of the 498 vouchers takes about 1 minute to type in Miracle, and the 222 already booked are checked by eye. This is an assumed rate, not a measurement.
- The app chose 275 ledgers by itself, and the user left all of them. The user changed only 4 of the app's name and account matches, and all 4 were co-op bank lines.
- The user searched for a ledger for only 69 entries (14%): 31 with no suggestion, 33 corrections and 5 to Suspense. A typical search took 6 seconds.
- No double entries: the 222 lines already in Miracle were left out. No missed entries: 10 unbooked September lines were found.
- Contra: 10 transfers became 10 vouchers, not 20.
Worth a look
- 119 entries were accepted in one click, 17 of them with a red mark. I checked all 119 by eye, and the names match the narrations. ✕ One is wrong: ₹1,05,100 from "HARESHKUMARJAYANTILAL" was saved to MANJIT HARESHBHAI NARVANI (38%), an Unsecured Loans ledger. Their customer ledger HARESH KUMAR JAYNTILAL exists, and their Miracle books show bills to him. One more is uncertain: a ₹68,003 Jamnagar co-op line saved as M.J TRADING CO.
- 6 co-op bank lines kept the app's own choice (Shri Krishna Gupta Chanawala ×2, Nisarg Trading ×1, Thakkar Dhirajlal ×2, Virumal Totamal ×1). Most of them share an amount with the user's own choices, so they are probably right. The app had no name to go on.
- ₹1,30,000 in Suspense is the missing half of a contra transfer (23/10). The other 4 Suspense entries are ₹2,06,075: a GTPL bill of ₹6,075, ₹1,10,000 to "VINOD", and ₹55,000 and ₹35,000 to "ANILGO".
- For the product: (a) on co-op bank lines, do not trust a pooled bank account number; use the amount instead. (b) Ignore the company's own name in a narration. (c) "Miracle is updating its files" appeared 3 times while Miracle was open.
Source: the live server's records for this customer, read without changes. Statement text stays in Ledger Map. This page shows only what is needed to explain it.