CA working papers · 2,506 words

Section 44AB Tax Audit: Bank Statement Reconciliation with Tally Prime and Zoho Books

44AB still begins with bank books. This CA guide shows how to reconcile twelve-month PDFs to Tally Prime / Zoho Books without dropping wraps, inventing amounts, or treating bounced cheques as sales.

Section 44AB tax audit working papers are still, in too many firms, a junior’s Excel dump of a PDF. That dump drops Canara wraps, misses inward returns, and books contra as sales. The statutory auditor then “reconciles” Tally to a lie. This guide is how a CA firm should ingest twelve-month statements: CBS-accurate extract, bounce taxonomy, Law-6 contra, then Tally Prime / Zoho Books mapping.

What 12 months must prove

• Opening and closing balances match the PDF (per account).
• Credits/debits foot to the engine totals, not to a typed sum.
• Bounces identified and reversed in books if they were booked as income.
• Owner contras excluded from turnover analytics.
• GST vs bank: receipts that never hit GST and GST that never hit bank are explainable.

Tally Prime mapping

Import or rec against FinAnalyser All Txns (date, narration, debit, credit, balance, account). Match on amount+date+narration tokens, not on Excel row numbers. Cheque numbers from bounce extract should match Tally bill-wise pending. If Tally shows a receipt the PDF never had, you have a booking error or a second account not uploaded.

Zoho Books mapping

Bank feeds in Zoho may be incomplete versus PDF (especially password statements). Use the PDF pack as the twelve-month spine. Match Zoho bank transactions to extract; unexplained Zoho rows are usually duplicates or cash modules mis-posted to bank.

Bounces in the audit file

Outward returns: sales not realised. Inward returns: payables not honoured — going-concern and related-party clues. ECS/NACH fails: EMI and statutory standing instructions. Do not present a single bounce integer in Annexure working papers.

Self-transfers and partners

Law 6: operational cashflow excludes own-account transfers. For audit, contras still exist in the ledger — they just must not inflate “receipts from operations.” Match account numbers / UPI handles / NEFT to related entities.

Opening/closing and NaN totals

Per-account opening and closing must be numbers, not NaN from a failed extract. Report total rows should not show openingBal=0 when the PDF had an opening line. Continuity across the year is the CA’s friend: if Layer-4 breaks, stop and fix the parse before you sign.

The working-paper pack

28-sheet Universal BSA: All Txns, Cashflow, bounce ledgers, contra-aware biz cashflow, monthly summary, daily balance. Attach PDF. Note exceptions. That is a 44AB bank file a reviewer can re-perform.

What to sample vs what to fully extract

Material current accounts: full twelve-month PDF extract. Dormant savings: maybe closing confirmations. The failure mode is sampling the current account by three months and missing a bounce wash in month ten. 44AB turnover is annual. Your bank spine should be annual for the operating accounts.

Multi-bank: proprietors with five accounts. Missing one HDFC is a recon hole that looks like suppressed receipts. List IFSC/account from the client and tick PDFs. FinAnalyser multi-bank All Txns with Account column is the tick sheet.

GST vs bank — a practical recon

Not every GST invoice hits bank (cash, other entity). Not every bank credit is GST supply (loans, contras, capital). The BSA cashflow sheet after Law-6 is the starting “bank receipts from operations” number. Investigate the gap; do not force-tie by booking ghosts.

File documentation for peer review

PDF, extract workbook, exception list (continuity breaks, password files, scans), bounce summary unmerged, contra list, Tally/Zoho rec screenshot. That is reviewable. A junior’s unmarked xlsx named final_final2 is not.

March crunch staffing

44AB season is when firms hire typists who create the worst extracts. Freeze the method in January: engine on all operating accounts, juniors only tick PDFs received. Do not invent a March-only Excel process. Reviewers in May will not reconstruct it.

Password PDFs pile up. Unlock in-memory per file. Do not keep a master password workbook for the whole client list on a shared drive named audit2026.

Related-party receipts in the bank book

Director loans, sister concerns, and cash-like UPI from family distort both turnover and FOIR if this were a PL file. For 44AB, identify them, disclose, and keep them out of operational cashflow analytics. Law 6 is an audit friend. Party-wise sheets from BSA give you the first cut; judgement still names the related party.

Who signs the bank recon

The article assistant prepares. The CA signs. The sign-off checklist is: continuity OK or exceptions listed; bounces unmerged; all operating accounts present; Tally/Zoho rec performed against extract not against a retype. If any box is empty, do not date the audit report against that bank book.

What 44AB bank recon is actually proving

Section 44AB tax audit still begins with bank books whether the firm lives in Tally Prime, Zoho Books, or a messy hybrid. The objective is not a pretty Excel. The objective is: every material business receipt and payment in the books should be explainable in the bank ledger (and vice versa), after timing differences, with bounces identified, contras named, and opening/closing balances that are numbers rather than NaN. Tally’s own bank rec only recs what was typed. If the clerk omitted a bounce or merged wraps, the rec is circular — assertion reconciled to assertion. The CBS PDF is independent evidence. Start there. FinAnalyser’s job is a Layer-4 extract with Law-3 wraps kept, Law-4 bounces unmerged, Law-6 contras identifiable, and ExcelJS working papers a reviewer can re-perform. The CA still signs. Software does not replace the auditor. It stops the auditor starting from a lie.

Twelve months means all operating accounts, not the one current account the client remembered. Proprietor SB used as a till is in scope if business money went through it — see the current-versus-savings guide. Password PDFs pile up in March; unlock in-memory per file. Do not keep a master password workbook for the whole client list on a shared drive named audit2026. Freeze the method in January: engine on all operating accounts, juniors only tick PDFs received. Do not invent a March-only Excel process. Reviewers in May will not reconstruct it.

Tally Prime: match tokens, not row numbers

Import or rec against FinAnalyser All Txns: date, narration, debit, credit, balance, account. Match on amount + date + narration tokens, not on Excel row numbers that will shift when a wrap merge is fixed. Cheque numbers from the bounce extract should match Tally bill-wise pending. If Tally shows a receipt the PDF never had, you have a booking error or a second account not uploaded. If the PDF shows a CHQ DEP RET that Tally never reversed, turnover is overstated — that is the finding. Contra and related-party receipts (director loans, sister concerns, family UPI) distort both turnover analytics and, if this were a PL file, FOIR. For 44AB, identify them, disclose, and keep them out of operational cashflow analytics. Party-wise sheets from BSA give the first cut; judgement still names the related party.

Zoho Books: feeds lie by omission

Bank feeds in Zoho may be incomplete versus PDF, especially password statements and accounts the client connected late. Use the PDF pack as the twelve-month spine. Match Zoho bank transactions to the extract. Unexplained Zoho rows are usually duplicates, cash modules mis-posted to bank, or a feed that double-counted a settlement. Unexplained PDF rows are usually books that never saw the money — including bounced deposits booked as sales. Export CSV from Zoho if you must; do not type. The extract is the spine. Zoho is the assertion. Same as Tally.

Bounces, GST, cash, and window dressing

Outward returns: sales not realised. Inward returns: payables not honoured — going-concern and related-party clues. ECS/NACH fails: EMI and statutory standing instructions. Do not present a single bounce integer in annexure working papers. Unreversed bounce is inflated turnover. GST 2A/2B versus bank receipts will not match 1:1; your job is to explain cash sales, other accounts, timing, and non-GST receipts. Cash deposits that look like 269ST problems are a legal/tax issue the partner owns; the extract’s job is to list them without dropping wraps that contain the depositor identity. Year-end window dressing is a March credit spike reversed in April — continuity and circular flags help you see it. Do not “smooth” it in Excel to make ABB look like a lender file. This is an audit pack.

1. All operating accounts, 12 months, original CBS PDFs.
2. Engine extract; continuity OK or exceptions listed.
3. Bounce ledgers unmerged; reverse in books if income was booked.
4. Contras and related-party named; ops cashflow excludes own-account.
5. Tally/Zoho rec against extract, not against a retype.
6. Cover shows account type evidence (Law 5) and opening/closing numbers.
7. Partner signs only when the checklist is complete.

March crunch staffing is when firms hire typists who create the worst extracts. That is how you get openingBal=0, dropped Canara wraps, and a bounce that was sales. The sign-off checklist is: continuity OK or exceptions listed; bounces unmerged; all operating accounts present; Tally/Zoho rec performed against extract. If any box is empty, do not date the audit report against that bank book. Attach the PDF. Note exceptions. That is a 44AB bank file a reviewer can re-perform — not a junior’s dump.

Multi-account years, OD interest, and the cash module

Clients with five banks and one OD will hand you the OD and the “main” current account. The GST receipts may sit in a third HDFC the proprietor uses on UPI. If that third PDF is missing, your recon will invent a cash-sales story that is actually a missing ledger. Tick PDFs received against a bank list from last year’s audit file plus GST portal disclosures, not against the client’s memory in March. OD interest and bank charges must land in the books; they are not bounce events. Sweep transfers between OD and current are Law 6 contras for analytics and still exist as ledger lines for rec — do not delete them from All Txns to make cashflow look pretty. Pretty cashflow is a lender artefact. Audit wants the lines.

Cash module in Tally versus cash deposits in the PDF is where firms get creative. Large cash deposits need identity in narration wraps — Law 3 keeps the depositor text. If wraps were stripped by a converter, you cannot complete a 269ST conversation from the extract. Re-parse. Do not guess the depositor from the client’s WhatsApp. Window dressing that credits a related concern on 31 March and reverses on 2 April will survive a junior who recs monthly totals and dies on a daily continuity sheet. That is why Layer-4 and daily balance sheets belong in the audit file, not only a monthly summary the client already liked.

GST versus bank will never be a formula. It is a set of explanations: cash sales, other accounts, timing, non-GST receipts, advances, and bounce reversals. Your working paper should list unexplained bands, not a fake 100% match. A converter that drops UPI refs makes the unexplained band larger and the partner’s night longer. Institutional Excel with party-wise cuts is how the article assistant starts; the CA finishes with judgement. If the firm still types twelve months because “we have always typed,” you are paying for a process that hides the very differences 44AB is meant to find.

If the firm’s review checklist still says “bank rec done in Tally,” add a line: rec performed against CBS extract with continuity exceptions listed. That one line is the difference between a re-performable 44AB file and a circular workbook. Partners who skip it are signing the junior’s typing. Partners who keep it can answer a peer reviewer who opens page 19 of the Canara PDF and still sees the same rupees as the All Txns sheet.