The Hidden Cost of 30 Minutes Per Case

Finance Connect ran a piece last week on bank statement automation in SME lending. Buried in the operational section was a number worth pulling out: an analyst or underwriter spends 30 to 60 minutes reviewing statements on each SME case, multiplied across hundreds of monthly applications.

Most lenders nod at that range and move on. It feels small. Half an hour here, an hour there. Embedded in salaries that are already being paid. The hidden cost only becomes visible when you actually multiply it out — and when you do, it’s larger than almost anyone in the building realises.

This piece does the multiplication.

The Single-Case Maths

Start with the conservative end of the Finance Connect range: 30 minutes per case.

That figure covers the underwriter time only. It doesn’t include the case manager who chases the documents, the file-prep step before the underwriter opens the PDF, the second pair of eyes for sign-off, or the rework when the wrong statement was uploaded and the right one needs requesting.

A more honest fully-loaded estimate for an SME application — across all hands that touch the statements — sits closer to 45 minutes. We’ll use both numbers below.

A typical SME lender doing 200 applications a month is therefore spending:

• 100 underwriter-hours per month on statement review (200 × 30 min)
• 150 total person-hours per month on the full statement workflow (200 × 45 min)

At a fully-loaded underwriter cost of £55,000 — salary, NI, pension, software, and overhead — the hourly cost is roughly £30. So:

• £3,000 per month in underwriter time on statement review alone
• £4,500 per month in total workflow cost

Across a year, that’s £36,000 to £54,000 for a lender doing 200 cases a month. For a lender doing 500 a month, it’s £90,000 to £135,000.

These numbers are unremarkable in isolation. They become interesting when you compare them to what the work actually produces.

What Those 30 Minutes Actually Buy

Here’s the uncomfortable part. The 30 minutes per case isn’t 30 minutes of judgement. It’s 30 minutes of administration to get to a position where judgement can begin.

The Finance Connect piece names this directly: before any real credit judgement could be made, teams often needed to establish what the genuine monthly income is, whether there are existing finance repayments, whether cash flow is stable or deteriorating, whether there are uncommitted funds available, and whether liabilities are understated elsewhere in the application.

That’s data gathering, not underwriting. It’s a baseline an experienced credit professional has to assemble before the actual analysis starts.

The £36,000 to £54,000 a year you’re spending on statement review at 200 cases a month is overwhelmingly being spent on the baseline, not the decision. You’re paying senior credit salaries to do data entry.

The Second-Order Costs

The direct labour cost is the visible one. There are three hidden costs that don’t appear on any P&L line but matter more.

Decision latency. Every hour an application sits waiting for statement review is an hour the broker is calling a competitor. Faster turnaround wins deals from lenders with identical risk appetite and identical rate cards. The cost of losing a deal isn’t the lost interest — it’s the lost relationship with the broker who sent it, and the next twenty deals they don’t send.

Inconsistent thresholds. Manual review produces variable outputs. Two underwriters reading the same statement will summarise it differently — different categorisations, different “average monthly turnover” figures, different judgements about what counts as a returned item. That inconsistency leaks into credit decisions. Some files get approved that shouldn’t; some files get declined that should have cleared. Neither error is visible at the time. Both show up in the portfolio later.

Senior time on junior work. The most expensive people in the credit team are the ones who least want to spend their day on transcription. When experienced underwriters are pulled into statement review during peak weeks, the cost isn’t just their hourly rate — it’s the strategic and exception work that doesn’t get done because they’re heads-down in PDFs.

What Automation Actually Changes

The point of automating bank statement extraction isn’t to fire underwriters. It’s to move them up the value chain.

The Finance Connect piece arrives at the same conclusion: AI agents do not remove the need for people; they automate parts of the process so that people can focus on maximising where they add value. That’s correct. The economic argument for automation isn’t headcount reduction; it’s redirecting the same headcount to work that actually moves the business forward.

What you get back, concretely:

• The 30 minutes per case becomes 2 minutes. Not zero — the underwriter still reviews the engine’s output before acting on it. But the work shrinks to checking and deciding, rather than reading and transcribing.
• The 100 underwriter-hours a month becomes 7. At 200 cases × 2 minutes, that’s 6.7 hours. The remaining ~93 hours redirect to relationship management, exception handling, portfolio monitoring, and the strategic work that doesn’t get done today.
• The cost line drops from £3,000 a month to roughly £200. The £2,800 a month difference is roughly £34,000 a year recovered at 200 cases — and it scales linearly with volume.

At 500 cases a month, the recovered cost is closer to £85,000 a year. At 1,000 cases a month, it’s £170,000. These numbers do not include the second-order benefits — faster decisions, more consistent thresholds, senior time freed up — which are typically larger than the direct cost saving but harder to put on a line.

The Calculation You Should Actually Run

The maths above is generic. The maths that matters is yours.

Three numbers:

1. Your monthly SME application volume
2. Your honest estimate of the fully-loaded time per case across all hands that touch the statements
3. Your fully-loaded underwriter cost per hour

Multiply them. The answer is what you’re spending today on a workflow that is, in the Finance Connect piece’s own framing, not the best use of time, cost and skilled resource.

That number is your starting position. What it could be, with the data layer automated, is between 5% and 10% of it.