GSTR-2B Reconciliation Without the Last-Week Scramble

Input-credit reconciliation is not a hard problem: matching a purchase register against GSTR-2B is mechanical, almost clerical work. It's a timing problem: GSTR-2B arrives on the 14th, GSTR-3B is due the 20th, and in most firms the client's purchase register shows up somewhere inside that six-day window, if it shows up in time at all. Whatever document is in hand when 2B lands decides how good the reconciliation is, not the client's book-keeping, not your team's GST knowledge. What follows is the full cadence, what timing-driven ITC loss looks like on one illustrative client, who owns each kind of mismatch, and the questions every partner eventually asks about QRMP, provisional credit, and how much is really at stake.

The six-day window, honestly mapped

Lay the calendar out day by day and the scramble stops looking like bad luck. It looks like an entirely predictable outcome of when one document arrives.

1. 114th: 2B is available. Firms with purchase registers in hand start matching the same day.
2. 215th–17th: where most firms actually receive the register, after two reminders.
3. 318th: mismatches surface: vendor hasn't filed, invoice number typos, missing e-way bills.
4. 419th–20th: no time to chase vendors, so ITC gets deferred or claimed on hope. Repeat next month.

None of this is exotic. It's the same four days, in the same order, at almost every firm that hasn't deliberately changed when the purchase register arrives. The variable isn't 2B, that date is fixed by law. The variable is the register, and the register is the one thing you can move.

Collect against the 12th, not the 15th

The purchase register exists in the client's system by the 10th: it just doesn't move until someone asks hard enough. A standing monthly ask that fires on the 1st, itemised, with automatic nudges until the register lands by the 12th, means matching starts the hour 2B drops. Same documents, same client, two extra working days, which is the entire difference between a calm afternoon and a last-hour scramble.

• Name the exact file or export you need (purchase register, not "your GST data"), so the client isn't guessing which report to pull.
• Set the collect-by date against the 12th, not against the 3B due date, the client's mental deadline should never be the 20th.
• Chase automatically, not manually: a reminder that fires itself on the 5th and the 9th beats one your team remembers to send on the 14th.
• Flag the previous month's unresolved gaps in the same ask, so old mismatches don't get forgotten while everyone focuses on the new period.

What one client's timing actually costs

Put a number on it, because "a few days late" sounds harmless until you see what sits behind it. Take a client we'll call Anand Fabrics, a mid-sized trading unit with roughly 60 to 70 purchase invoices a month across a dozen regular vendors. In a month where the register arrives on the 17th, the reconciliation window shrinks to two working days before the 3B due date. That's not enough time to chase even the straightforward gaps, let alone the vendor who filed late.

On that kind of month, illustratively, somewhere around ₹35,000–₹45,000 of ITC sits provisionally unmatched purely because there was no runway to resolve it, not because the credit itself was ever in genuine doubt. The invoices are real, the vendor eventually files, the mismatch would have cleared with two more days of chasing. It gets carried forward, or claimed on hope and reversed later if it doesn't hold up. Multiply that across a dozen GST clients on the same pattern, every month, and the number stops being a rounding error.

Nothing about Anand Fabrics' business changed between a good month and a bad one. The only variable was which day the register landed.

The same client, two different months

Watch the contrast on the same client, two consecutive months, register timing as the only difference. In month one, the register comes in by the 12th. Matching starts on the 14th, mismatches are visible by the 15th, and the gap list goes to the client that afternoon: three vendor-side issues, one invoice-number typo on the client's own entry. All four resolve by the 19th, two vendors correct their upload, the typo gets fixed at the client's end, and 3B is filed on the 19th with ITC fully matched.

In month two, the register arrives on the 17th, same client, comparable invoice volume. Matching starts on the 18th. The same four mismatches appear, one vendor filed a day late, one uses a slightly different invoice number, one e-way bill reference is missing, one GSTIN digit is wrong on the client's side. There's no time left to chase any of it. All four get carried as provisional or deferred, and 3B goes out on the 20th with a note to revisit next cycle. Two of the four eventually turn out to be genuine, resolvable issues that simply never got worked because the clock ran out.

Same mismatches, same client, same effort required to fix each one. The only thing that changed was whether there were five working days to fix them or one.

Every mismatch has an owner

Reconciliation gaps get treated like one undifferentiated problem, when they're really four or five distinct problems with four or five distinct people who need to act. Sorting a gap list by owner, not just by mismatch type, is what makes a chase actually land on the right desk.

Vendor mismatches are the client's chase: arm them early

When a vendor hasn't uploaded an invoice, the leverage sits with the client who owes that vendor money, not with you. Your firm can flag the gap; only the client can make a vendor move. Firms that send clients a vendor-wise gap list on the 15th watch mismatches resolve before the 20th. Firms that discover gaps on the 19th write the same email a month later, about the same vendors, because nobody ever gave the client the specific, actionable list early enough to act on it.

The difference is rarely the message's tone. It's whether the client received something specific enough to forward straight to the vendor, on a day when there was still time for the vendor to respond.

Hi, for this month's GST filing, GSTR-2B is showing three invoices from your side as not yet uploaded: Invoice 442 dated the 6th, Invoice 447 dated the 9th, and Invoice 451 dated the 11th. Could you check with your accountant that these have been filed under your GSTR-1? We need this resolved by the 19th to claim the credit this cycle. Happy to share the exact GSTIN and invoice details if useful.

What about clients on QRMP who don't file GSTR-1 monthly?

The 2B-to-3B pressure doesn't disappear for QRMP filers, it just moves. They still file GSTR-3B monthly (or pay via PMT-06 for the first two months of the quarter), and 2B still generates monthly, reflecting whatever their vendors have uploaded so far that quarter. The purchase register discipline matters just as much here, arguably more, since QRMP clients tend to assume the quarterly GSTR-1 cadence means nothing is urgent monthly, which is exactly backwards for ITC tracking. Collect the register on the same monthly rhythm regardless of which scheme the client is on.

How much ITC is typically at risk, purely from timing?

There's no single honest figure here, and any firm claiming a precise industry-wide percentage is guessing. What's consistent across firms that have actually measured it internally is the pattern, not the number: a meaningful share of monthly ITC that eventually reconciles cleanly still spends one cycle sitting provisional or deferred, for no reason other than the reconciliation window closing before anyone had time to chase it. The fix isn't a better matching tool, the matching itself is already close to automatic. The fix is giving the chase enough days to actually happen before the 20th arrives.

Should we claim provisionally and reverse later, or wait until fully matched?

This is a judgment call for each firm and client, not something to standardise on without thinking it through, and it isn't tax advice we're offering here. What's worth naming plainly is the operational tradeoff underneath the decision: claiming provisionally and reversing later keeps cash flow steady but adds a tracking obligation next cycle; waiting until fully matched is cleaner on paper but can mean sitting on credit that was always going to clear. Either choice is more defensible, and easier to execute consistently, when the reconciliation happened with five working days of runway instead of one. The timing fix doesn't answer the judgment question, it just means you're making that call with real information instead of a deadline forcing your hand.

The metric that matters

Track one number monthly, across the GST book: how many clients had their purchase register in hand by the 12th. Not how many returns got filed on time, they usually do, one way or another. The 12th-by number is the leading indicator; the filing deadline is just where the consequences of missing it show up. Move that one number, and the last-week scramble stops being a monthly event and becomes something that happens to other firms.