The verdict in three sentences
For a mid-market company processing 40,000 bank transactions a month across several entities, manual matching costs more than automating it within the first year. An off-the-shelf tool at GBP 250 to 1,300 a month is enough if your flows are simple, but a custom rules engine with AI (GBP 10,000 to 26,000) pays off when references are messy and accounts are numerous. The realistic target is 85 to 95% of entries matched automatically and a close brought forward by 3 days, in a 2 to 4 month project.
Where your finance team's time goes
In a typical London mid-market group (5 entities, 11 accounts spread across 4 banks), reconciliation always stalls on the same issues: bulk customer payments settling several invoices, references truncated by the bank, fees charged with no reference, intercompany flows to eliminate, failed Direct Debits. Each case requires a manual search in the ERP. At 30 to 45 seconds per difficult entry, two accountants spend most of their week on it.
Automation relies on three building blocks: statement retrieval (Open Banking APIs or bank host-to-host feeds, ideally camt.053 rather than legacy MT940), a rules engine (amount, date, reference, payer account, tolerances) and an AI layer that suggests likely matches for ambiguous references, with human approval of uncertain cases.
| Monthly volume | Entries to process | Estimated manual time | Full-time equivalent |
|---|---|---|---|
| Single-entity SME | 3,000 | 25 h | 0.2 FTE |
| Multi-account SME | 10,000 | 85 h | 0.6 FTE |
| Mid-market, 3 entities | 25,000 | 200 h | 1.3 FTE |
| Mid-market, 5 entities, 11 accounts | 40,000 | 310 h | 2 FTE |
| Group, 10 entities | 90,000 | 680 h | 4.5 FTE |
| After automation (5 entities) | 2,000 to 6,000 exceptions | 45 to 70 h | 0.3 to 0.45 FTE |
Off-the-shelf tool or custom engine: 2026 prices
SaaS reconciliation tools and native ERP modules (Sage, Xero, Microsoft Dynamics, NetSuite, Odoo) handle standard cases well. Their limits show on specific rules: allocating bulk payments, cross-entity matching, reconciling card settlements or payment terminal takings.
| Option | Setup cost | Recurring cost | Target auto-match rate | Timeline |
|---|---|---|---|---|
| Native ERP module | GBP 1,700 to 5,000 | included or GBP 90 to 260/month | 60 to 75% | 2 to 4 weeks |
| Entry-level reconciliation SaaS | GBP 1,300 to 3,500 | GBP 250 to 520/month | 70 to 85% | 3 to 6 weeks |
| Multi-entity SaaS | GBP 4,300 to 10,000 | GBP 700 to 1,300/month | 80 to 90% | 6 to 10 weeks |
| Custom engine, rules only | GBP 10,000 to 15,500 | GBP 130 to 260/month hosting | 80 to 88% | 2 to 3 months |
| Custom engine, rules and AI | GBP 17,000 to 26,000 | GBP 220 to 430/month | 85 to 95% | 3 to 4 months |
| Open Banking aggregation per account | none | GBP 1 to 3/month per account | n/a | 1 to 3 weeks |
Custom work is justified above 20,000 entries a month, with at least 3 entities, or with business rules vendors cannot configure. Below that, a well-configured SaaS remains the best value for cost and speed.
A 4-month project, step by step
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
A serious project starts with an audit of 3 months of history: entries already matched by hand are replayed to measure what rate the rules achieve. Then comes bank-by-bank connection, rules for each flow family, training the AI layer on historical references, then a month of parallel running before cut-over. The audit trail is kept: each automatic match stores the rule applied and its confidence score, which your auditors will ask for.
| Step | Duration | Deliverable |
|---|---|---|
| Audit and history replay | 2 to 3 weeks | achievable match rate per flow |
| Connecting the 4 banks | 3 to 6 weeks in parallel | daily camt.053 or API statements |
| Business rules per flow family | 3 to 4 weeks | 40 to 80 documented rules |
| AI layer on ambiguous references | 2 to 3 weeks | suggestions with confidence score |
| Parallel running | 4 weeks | measured gaps, tuned rules |
| Cut-over and training | 1 week | exception handling procedure |
Mini case study
Sarah, CFO of a technical distribution group in London, has 2 accountants working full time on matching, a loaded cost of GBP 95,000 a year. She chooses a custom AI engine at GBP 22,000 and GBP 350 a month to run. After cut-over, 91% of the 40,000 entries are matched automatically and exceptions take 0.4 FTE. The annual gain is 1.6 FTE, about GBP 76,000, minus GBP 4,200 running costs: GBP 71,800. The project pays back in under 4 months, and the close moves from working day 8 to working day 5.
FAQ
Do we need direct bank feeds or is Open Banking enough? Open Banking under the UK framework works well for most mid-market groups, at GBP 1 to 3 per account per month. Direct host-to-host feeds remain more robust for high volumes and long history, and some banks charge GBP 20 to 50 a month for them.
What auto-match rate is realistic? 85 to 95% for standard customer and supplier flows after 2 months of tuning. The remaining 5 to 15% are genuinely ambiguous cases that should stay under human review.
Can the AI match without supervision? We recommend a confidence threshold: above 95%, automatic match, between 70 and 95%, a one-click suggestion, below that, manual handling. This protects accounting quality.
What happens to the two accountants? In most projects they move to management accounting, credit control or variance analysis, higher-value tasks. Plan 2 to 3 days of training on exception handling.
How much does maintenance cost? Allow 10 to 15% of the project cost per year, GBP 2,200 to 3,900, to add rules, a new bank or a new entity.
Let's scope your project. Send us your number of entities, accounts and monthly entries: we will scope the work, give an indicative budget between GBP 10,000 and 26,000 and a 2 to 4 month timeline. Detailed quote within 48 h. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
