Day end at a clinic: counting cash, checking UPI

A clinic day that took only cash and UPI does not have one figure to check at closing. It has two, and they fail for unrelated reasons. Cash is the only money a person can physically verify, so that half is a count against an expected number. The UPI total cannot be counted at all: it is a claim your own records make, testable only against a second system, the bank account or the collection app. Run them as one total and you get a single unexplained number.

The countable half: expected cash, then a blind count

Expected cash is opening float plus cash collected, less refunds paid out in cash, less anything paid out of the drawer. Count what is physically there before you look at that number. Retail guidance calls this the blind count: the counter enters the drawer figure first and only then sees the expected balance, which is given as opening float plus cash sales less cash paid out.

1. Count by denomination and subtotal, the way Northwestern's sample drawer count form does it, rather than eyeballing a bundle.
2. Subtract the float. That form states plainly that the drawer total should always be static, so the float you open with tomorrow is the float you opened with today.
3. Take the expected figure from somebody other than the counter. Northwestern's instructions say the register total is obtained from the approver of deposits.
4. Put the difference on its own over and short line, with a shortage added to and an overage subtracted from the deposit so the result equals the register total.
5. Close out the system and print the total, count the cash, deduct petty cash if you keep any, and the remainder should balance to the day's grand total. That is Utah's closing sequence, without the card machine step a cash and UPI clinic does not have.

The uncountable half: what a UPI total is claiming

When your own record says the UPI side of the day is done, you are asserting three separate things: that the sender's bank moved the money, that it reached the clinic's account, and that none of it will come back. All three can be false tonight and true next week. The RBI circular on harmonisation of turn around time, dated 20 September 2019, sets the windows: a UPI payment to a merchant that receives no confirmation is auto reversed within T plus 5 days, and a fund transfer where the beneficiary account is not credited is auto reversed latest on T plus 1 day, where T is the transaction date. Compensation of Rs 100 per day applies beyond that window, automatically, with no complaint needed. PhonePe's gateway guide restates the same timelines and notes that most refunds in practice land within hours.

The second trap is settlement. If the clinic collects through a payment aggregator rather than straight into its own bank account, today's money is not in the account today. Razorpay documents the standard domestic cycle as T plus 2 working days from capture, worked through as a payment captured on 02 February 2019 at 9:00 a.m. settling on 04 February 2019 at 9:00 a.m., shifted if that date is a bank holiday, and sells instant settlement as the paid way around it.

Causes, sorted by which half they break

PhonePe groups UPI failures the way a desk meets them. Human: wrong UPI PIN, insufficient balance, crossing the daily limit it states as Rs 1 lakh, and a mistyped virtual payment address. Technical: bank server downtime at month end peaks, weak connectivity mid transaction, scheduled NPCI maintenance, and gateway timeouts. Its escalation path is the one to hand the patient: check status through UDIR inside the UPI app, wait 60 minutes before retrying, then take the transaction ID to the bank.

What to do with the difference

1. Recount by denomination before theorising. Most cash gaps are the count.
2. Look for the unrecorded transaction: a payout, a refund, a void, a bill reissued.
3. Have a second person verify. Utah wants a supervisor recount before any entry is made, and FSU wants the cashier and the supervisor both counting at close.
4. Write it down the same day. Northwestern requires shortages documented in full and overages deposited rather than left in the fund, and money handed between staff recorded with name, date, time and amount.
5. Bank on a schedule, not when it feels large. FSU expects deposits daily and nothing held beyond five business days, and Utah says within 3 business days. Both are American university rules, so treat them as a shape to copy rather than a number that transfers.
6. Track the pattern. A variance once a quarter is a different problem from a small shortage every Friday, and a steady run of small shortages is the stronger signal.

What Elaior's day end sheet shows, and what it does not

Verified in the code, because the honest version is narrower than the pitch. Elaior does not record how money arrived. There is a payment mode column in the database, but every payment row and every expense row is written with the same fixed value, and no screen asks or displays a method, so the day end sheet carries no cash figure, no UPI figure and no drawer count. What it does report for one day: collections net of refunds, refunds shown as a positive number, a category split across consult, pharmacy, procedure and other, the expense rows, the net, and counts of invoices issued, invoices voided, patients registered and consults done. Splitting the two tenders and counting the drawer is still paper and a second pair of eyes. Two product rules do remove causes from the table above: billing is pay first, with no khata and no partial payments, and no field on an issued invoice can be edited, so a correction is a void and a fresh bill. Expenses can only be entered for today or yesterday, so a payout cannot be quietly backdated into a day that was already closed.

What we do not know

• Whether a static QR crediting the clinic's own bank account arrives in real time rather than on an aggregator cycle. Unverified, and it is the point that would matter most here.
• Current UPI limits. The Rs 1 lakh daily figure above is vendor stated, through PhonePe. NPCI's own pages did not render for us, so there is no primary citation.
• What share of collections at a small Indian clinic arrives over UPI. We found no figure worth quoting.
• A widely repeated India settlement table, UPI at T plus 1 and cards at T plus 2 to T plus 3, credits the Reserve Bank of India but links to no document, so we have not used it. The gateway's own pages are the citable version.