Experienced accountant interviews skip definitions like accruals and ask which judgement you made, why you made it, what it cost, and what happened when someone senior pushed back. Expect questions on revenue splits, suspense accounts, intercompany gaps, currency revaluation, provisions, prior-year errors, audit adjustments, control failures and coaching juniors. This page is written for accountants with roughly three to ten years behind them, who own a ledger, a close or an entity and review other people's work. Each answer below is a first-person story with a situation, what you did, the result and the trade-off. Swap in your own figures and systems before you say it out loud.
Search all questions by round, difficulty and level, or save the ones you want to practice.
Scope: what you owned, how big it was and who relied on it.
Decisions: two or three calls you made yourself, not ones handed to you.
Hindsight: one concrete thing you'd change and why.
“For the last three years I've owned the full close for our largest trading entity, from the payables and accruals through to the balance sheet reconciliations and the pack that goes to the finance director. Two people prepare work for me, and I sign off everything before it leaves the team. The biggest decision I made was moving the balance sheet reviews from quarter end to every month, which found a mis-posted customer refund balance two months earlier than we'd have caught it before. If I did it again, I'd document the judgement areas much earlier. When I took two weeks off last year, my cover had the checklists but not the reasoning behind the provisions, so a couple of calls waited for me. Now each judgement has a short note explaining the method and the evidence.”
Describing a team achievement with no clear sense of what you personally owned or decided.
Spot it: the gain or loss that didn't fit the rate movement.
Root cause: which accounts were flagged for revaluation and whether they should have been.
Fix and guard: correct the setup and add a check to the close.
“One month the system showed a large currency loss even though the rate had barely moved. I pulled the revaluation report by account and found it was revaluing supplier prepayments and a foreign currency inventory account. Those are non-monetary balances, so they should stay at the rate on the day of the transaction. Only monetary items like cash, receivables, payables and loans get restated at the closing rate. Someone had ticked every foreign currency account for revaluation when the system was set up. I reversed the wrong entries, fixed the account settings with the systems team, and checked the previous months, where the effect was small enough to correct in the current period. Now the close checklist includes a quick reasonableness check: expected gain or loss from the rate change against what the system posts.”
Saying every foreign currency balance gets revalued at the closing rate.
The difference: the tax rules write the asset off faster than the accounts do.
The effect: less tax now, more tax later, so the accounts carry the later tax as a liability.
Keep it concrete: one asset, simple numbers, no jargon.
“I'd use one machine as the example. In our accounts we spread its cost over ten years, because that's how long we'll use it. The tax rules let us claim it much faster, say most of it in the first few years. So in those early years our taxable profit is lower than our accounting profit, and we pay less tax now. But that isn't a saving forever. In the later years we'll have nothing left to claim for tax while the accounts are still charging depreciation, so we'll pay more tax then. The deferred tax liability is that future tax, recognised now so our profit isn't flattered by the early benefit. I've given roughly that explanation to a plant manager who asked why tax was on his balance sheet, and it landed well.”
Calling it tax the company owes today or tax it has avoided for good.
What looked wrong: the line, the size and why it didn't fit.
How you traced it: from the summary down to the entries.
The fix: the correction and the control that stops it recurring.
“Gross margin at one branch jumped in a single month, and the branch manager was happy to take the credit. Sales hadn't changed much, so I looked at cost of sales. It had dropped sharply. I drilled into the entries and found that a large batch of supplier invoices for stock had been coded to a prepayments account instead of purchases, because a new clerk had picked the wrong code from a similar-looking list. So the margin gain wasn't real. I moved the invoices to the right account before the pack went out, and the margin went back to normal. Then I asked for the two confusing codes to be renamed and added a margin check by branch to my review. It was a slightly awkward call with the branch manager, but much better than reporting it upward.”
Accepting a big favourable variance without checking it, because good news rarely gets questioned.
Why: time spent chasing small items that don't change any decision.
How: a level based on the size of the results, agreed with the controller and auditors.
Guardrails: year-end stays fuller, and many small items of one kind still get booked together.
“Our team was spending most of close day chasing tiny accruals like taxi receipts and small subscriptions. I proposed a threshold: at month end we'd only accrue items above a set level, based on how small it was compared with monthly costs. I agreed it with the financial controller and mentioned it to the auditors so nobody was surprised. Two guardrails mattered. At year-end we went back to a much lower threshold, and if many small items came from one source, like a recurring service billed late, we'd accrue them together. It saved most of a day each month. The cost was a little noise between months in a few cost centres, and one budget holder complained, so I showed him the effect was tiny and it settled down.”
Skipping accruals to save time with no agreed level, no approval and no year-end safeguard.
Separate the promises: which items were distinct and which were one bundle.
Allocate the price: on relative standalone selling prices, not the invoice lines.
Time each part: point in time versus over time, and the evidence for each.
“At my last company we sold equipment with installation and two years of servicing, all on one invoice. Sales wanted to book the whole amount on delivery. I read the contract and treated the equipment and the servicing as separate promises, because customers could buy servicing from others. Installation was small and done on the delivery day, so keeping it with the equipment made no difference to the timing. Then I allocated the total price using what we charged for each part when sold alone, not the discounted line on the invoice, so the discount was spread across both. The equipment revenue went in when the customer took control on delivery, and the servicing portion went to deferred revenue and was released evenly over the two years. It moved revenue out of the quarter, which sales didn't love, but the auditors accepted the method with no changes.”
Allocating revenue by whatever the invoice lines say, or booking everything on delivery because that's when the invoice went out.
Base method: loss rates by ageing bucket from your own history.
Adjust: for current conditions and specific customers you know are at risk.
Document: the data, the calculation and why it's reasonable.
“When I took over receivables, the provision was a flat round number carried forward every year. I replaced it with a matrix. I took three years of history, worked out how much of each ageing bucket was eventually written off, and applied those loss rates to the current ageing. The framework we used expects the provision to reflect expected losses, including current conditions, so I adjusted the rates upward for one sector where customers were clearly struggling. On top of that, I set specific provisions for two customers in formal insolvency, and took them out of the matrix so they weren't counted twice. The new figure was higher, which the finance director questioned, but the history and the working made it easy to defend, and the auditors accepted it. It now updates each quarter.”
Picking a round number or a flat rate with no data behind it.
Break it down: list every open item by age, source and size.
Clear by size: trace the big items properly, handle the small tail with an agreed threshold.
Stop the inflow: find why items land there and fix the cause.
“I had exactly this at my last job. First I exported every open item and grouped them by source: most came from unmatched bank receipts and a few from a payroll interface. I sorted by size and traced the largest items first, because a handful made up most of the balance. Those turned out to be customer receipts nobody had matched, so I allocated them to the right accounts and told credit control. For the long tail of tiny old items, I agreed a threshold with the financial controller, got approval, and wrote them off with a documented list. Then I fixed the cause: bank receipts without a reference now go to a named person the same week. The trade-off was accepting a small write-off instead of spending weeks on pennies, and I made sure the approval was on file.”
Writing the whole balance off in one journal without tracing the large items or getting approval.
Find the pattern: timing, currency, miscoding or charges one side never booked.
Agree a rule: cut-off dates, one rate, who invoices and who accepts.
Make it routine: a matching step before close, not after.
“When I joined my last company, our entity and a sister company were out by a different amount every month, and the group team just posted a balancing entry on consolidation. I matched the two ledgers line by line for three months. Three causes came up again and again: goods in transit booked by the seller but not yet by the buyer, each side using a different exchange rate for the same invoice, and management charges we booked that they never received. I agreed with their accountant that invoices would be exchanged three days before month end, both sides would use the group month-end rate, and any charge had to be accepted before it was booked. Within two months the difference was down to a few small timing items we could explain. It meant a slightly earlier cut-off for both teams, which they accepted.”
Treating a plug on consolidation as the fix, or blaming the other entity without matching the detail.
Understand their view: what they think is misstated and why.
Build the evidence: documents, the policy and the standard, in writing.
Resolve: agree, compromise on disclosure, or escalate to the right level.
“At year-end the audit senior proposed releasing a warranty provision, saying claims had been low for two years. I asked for their reasoning in writing first, so we were arguing about the same thing. Then I pulled the claims history by product and showed that a new product line had launched that year, and its early claims were already running higher than the old line. I also showed the provision method we'd used consistently and the sales volumes still under warranty. The manager reviewed it and agreed to keep the provision, but asked us to add a note on how the estimate was made, which was fair. The lesson I took was that our file hadn't explained the estimate well enough. Now every provision has a one-page note with the method and the data before the auditors ask.”
Treating the auditor as the enemy, or giving in on every point to finish the audit faster.
Size it: the effect on profit, net assets and key lines, including tax.
Apply the rule: material errors are usually corrected by restating comparatives; immaterial ones go through the current year.
Tell people early: controller, auditors and anyone who relied on the numbers.
“At my last company I found that a customer rebate agreement had been missed the year before, so revenue was overstated. First I worked out the full effect on revenue, profit and tax, and checked whether it touched any loan covenants. Under the framework we reported in, a material prior period error is corrected by restating the comparative figures and opening balances, not by pushing it through this year's profit. An immaterial one can be corrected in the current year with a note in the file. Ours was small against profit, so after discussing it with the financial controller and the auditors, we corrected it in the current year and disclosed nothing extra. I also added rebate agreements to the year-end accruals checklist, because the real failure was that sales signed it without telling finance.”
Quietly fixing it this year without telling anyone or checking whether it was material.
The rule: does the spend create an asset you control that brings future benefit, and has the project moved past the exploring stage?
The evidence: project approvals, plans and timesheets that show when the real build started.
The outcome: what you booked, when amortisation starts, and how you explained it.
“Our product team built a customer portal with our own developers, and the head of technology wanted every hour of the year capitalised, because it would lift profit. I asked for the project plan and the timesheets. The early months went on comparing options and throwaway prototypes, and under our framework that exploring stage is expensed. Once the design was approved, funded and clearly workable, the developers' time building it met our policy, so I capitalised that, with the timesheets as evidence. Training, data clean-up and the bug fixes after launch went to the income statement. So roughly half the cost went on the balance sheet, amortised from the day the portal went live. He wasn't thrilled, but I showed him the policy and the split, and the auditors tested it later and agreed. The trade-off was stricter timesheets for the developers, which they grumbled about for a month.”
Capitalising the whole project because a senior manager asked, or expensing it all without looking at what each stage of the work did.
Initial entry: lease liability at the present value of payments, and a matching right-of-use asset.
Judgements: lease term with extension options, and the discount rate.
After that: interest on the liability, depreciation on the asset, and the effect on profit timing.
“We signed a ten-year office lease with an option to extend for five more. Under the standard we reported in, I recognised a lease liability at the present value of the future payments and a right-of-use asset for the same amount, adjusted for any incentives received and our direct costs of signing. The two judgement calls were the term and the rate. On the term, we'd spent heavily on fit-out, so I argued extension was reasonably certain and used fifteen years, and documented why. For the rate, there was no rate in the lease, so I used our incremental borrowing rate, backed by a quote from our bank. Each month the liability carries interest and the asset is depreciated, so the total cost is higher in the early years than the rent paid. I explained that to the budget holders before it surprised them. Rules differ by framework, so I'd always check which one applies.”
Treating the lease as a monthly rent expense without knowing whether the framework requires it on the balance sheet.
Know the rule: revenue follows transfer of control, not the invoice date.
Offer real options: can the goods ship today, or does the order meet the strict bill-and-hold tests?
Escalate cleanly: in writing, to the controller, if the pressure continues.
“I'd say no to booking it, but I'd try to help within the rules. Revenue goes in when the customer gets control of the goods, and an invoice dated this quarter doesn't change when that happens. So I'd ask two questions. Can it actually ship today, with the customer accepting delivery? If so, it's genuine revenue this quarter. Or has the customer asked us to hold goods that are finished, set aside for them and can't go to anyone else? That's bill-and-hold, and the tests are strict, so I'd want the customer's written request. If neither fits, I'd explain that it lands next week and it's still a sale. If he pushed, I'd take it to the financial controller in writing. I had a version of this once, and the order shipped two days later with no harm done.”
Booking it and planning to reverse it next month, or refusing without explaining what would make it legitimate.
Recover: contact the supplier and get the money back or credited.
Trace: exactly which step let it through.
Fix the control: a change that would have stopped it, and a look back for others.
“A supplier was paid twice for the same delivery. First I called them and agreed a credit against the next invoice, which came through within the week. Then I traced it. The supplier had emailed a copy invoice after chasing payment, and it was keyed with a slightly different invoice number, an extra letter at the end, so the system's duplicate check didn't flag it. I changed two things. The duplicate check now also flags the same supplier, amount and date even when the invoice number differs, and copy invoices go to one person who checks the payment history first. I also ran that same test over the past year of payments and found two smaller duplicates, both recovered. The trade-off is a few false alarms each week, which the team clears in minutes.”
Getting the money back and stopping there, with no look at why the control failed or whether it happened before.
The change and why: the problem it solved for finance and for them.
The resistance: what they objected to, taken seriously.
Getting adoption: changes you made to the process and how you tracked it.
“We needed purchase orders raised before spending, because invoices were arriving for costs nobody had approved and our accruals were guesswork. Operations pushed back hard, saying it slowed down urgent repairs. Instead of pushing the policy, I spent a morning with their supervisors. The real problem was that raising an order took several screens and approval sat with one person who was often away. So I got a short form approved for repairs under a set amount, and added a second approver. I also showed their manager a report of their spend without orders each week, which made it visible without me chasing. Within a few months, most invoices had an order, and our month-end accruals became far more accurate. The trade-off was a looser rule for small repairs, which I thought was worth it.”
Announcing the rule and rejecting invoices without understanding why people weren't following it.
Build: receipts and payments by week from the ledgers, not from profit.
Test: compare each week's forecast with the actual and learn why they differ.
Use it: the decisions it drove, like payment timing or chasing debt.
“When a big customer paid late for three months in a row, our finance director asked me for a weekly cash view. I built a thirteen-week forecast from the receivables and payables ledgers, with payroll, tax payments, rent and loan repayments added on their real dates. Receipts were the hard part, so I used each large customer's actual payment behaviour instead of their terms. Every Monday I compared last week's forecast with the bank and noted why they differed. The first few weeks were off quite a bit because I'd been too hopeful about two customers, so I adjusted. After that it was usually close. It showed a tight week coming a month ahead, so we agreed new terms with two suppliers and chased a few overdue customers early, and we never had to draw on the overdraft.”
Building the forecast from budgeted profit, or never checking it against what actually happened.
Why change: reporting needs the old structure couldn't meet.
Design rules: accounts for what something is, cost centres or dimensions for where.
Mapping: old to new, tested, so comparisons still work.
“At my last company the chart had grown to well over a thousand accounts, with separate accounts for the same expense in each department. Reporting took ages and coding errors were constant. I led the redesign with the controller. The main rule was that the account says what the cost is and the cost centre says where it happened, so dozens of department-specific accounts collapsed into one each. We split a few accounts that were too broad, like professional fees into legal, audit and consulting, because management kept asking. I built a mapping table from every old account to a new one, ran the last two years of trial balances through it, and checked the totals matched to the penny. It took three months, and we lost a little detail in a couple of areas, but coding errors dropped and the monthly pack became much faster.”
Changing the structure without a tested mapping, so last year's figures can't be compared with this year's.
The task: what it was and how long it took by hand.
The build: the tool and the logic in plain words.
The controls: parallel run, built-in checks and documentation.
“Every month we built the sales commission accrual by copying reports from three systems into one workbook, which took nearly two days. I rebuilt it with Power Query so the three exports load and join on their own, and the calculation runs off a rules table instead of formulas buried in cells. Before trusting it, I ran the old and new versions side by side for two months and explained every difference, and one of those turned out to be an error in the old manual version. I added check totals that must agree with the source reports, and a clear warning if a salesperson appears with no rule. Then I wrote a one-page guide and had a colleague run it alone while I watched. It now takes about an hour. The trade-off is that someone needs to know Power Query when the rules change.”
Switching straight to the automated version with no parallel run and no checks against the source.
Tie-out: the balance agrees to the ledger and the support is real, not just a printout of the same number.
Reconciling items: age, size and whether each will actually clear.
Judgement: does the balance make sense for the business?
“First I check that the balance agrees to the ledger at the right date and that the support comes from outside the ledger, like a bank statement or a supplier statement. Then I look hard at the reconciling items: anything old, anything round, and anything that appears every month with the same amount. Last, I ask whether the balance makes sense. Once a junior's accruals reconciliation was perfectly tidy, but one accrual had been sitting there for eight months with a note saying invoice expected. I asked her to call the supplier, who said the work had been cancelled. So we released it. I didn't just fix it myself; I sat with her and showed her why the age of an item matters more than whether the sheet balances. Her reconciliations got much sharper after that.”
Checking only that the reconciliation totals to the ledger and signing it off.
The pattern: what kept going wrong and its effect.
The cause: knowledge, process, workload or attention.
The support: what you changed and how you checked progress.
“A payables clerk on my team kept posting invoices to the wrong period near month end. My first instinct was to correct them in review, but that just hid the problem. So I sat with him for an hour during close and watched how he worked. The issue wasn't carelessness: he didn't really understand that the service date, not the invoice date, decides the period, and the system defaulted to the invoice date. I explained the principle with his own examples, gave him a simple rule to check on each invoice, and asked the systems team to show the service period on the entry screen. For the next two closes I checked a sample of his work and gave feedback the same day. The errors mostly stopped by the second month, and later he was the one explaining cut-off to new starters.”
Quietly fixing their work every month, or going straight to the manager without trying to understand why it happens.
Triage: what must be right for the audit, what can wait, what can be simplified.
Redistribute: who covers what, with support, and what you take yourself.
Communicate: tell the controller and the auditors early, with a revised timetable.
“First I'd list everything that person owned and sort it by what the auditors need first and where the risk sits. The big balance sheet reconciliations, revenue cut-off and provisions come first. Management reporting extras and nice-to-have analysis can slip. Then I'd redistribute: I'd take the judgement-heavy items myself, like the provisions, and give the routine reconciliations to a colleague with clear notes, checking in twice a day. I'd call the audit manager the same day and agree which areas they'd test first, so we deliver those on time and push others back a few days. And I'd tell the financial controller what's at risk, instead of hoping to catch up quietly. I did something like this two years ago, and asking the auditors to reorder their plan saved us, because they were happy to start with sales testing.”
Trying to do it all yourself without telling anyone, then missing the audit deadline.
ClapAssist is an AI interview assistant for Mac and Windows. It listens to the interview on your computer and shows you what to say, in short lines you can read while you talk. Your live interview audio and screen are never stored. Your resume and notes are saved to your account so the app fills them in on any computer. It stays out of screen share on every plan, including Free; only you can see it.