A single bookkeeper managing bank reconciliation for 30 or 40 clients faces a throughput challenge that software alone does not solve. The bottleneck is not the individual transaction — it is the workflow overhead: switching context between clients, knowing which clients need attention, applying consistent coding decisions across multiple engagements, and catching exceptions before they become problems. This guide explains how to structure a multi-client reconciliation workflow that scales.
The Practice Dashboard: Your Entry Point Each Day
Reconlink's practice dashboard shows all your clients in a single view, with indicators for outstanding unreconciled transactions, statement import status, and clients whose last reconciliation is overdue. The correct workflow is to start here — not by jumping directly into a client's account.
A client with a red indicator means either the latest statement has not been imported yet (check whether the client has forwarded it to their inbox) or there are unreconciled transactions awaiting your review. Green means the client is current. The dashboard lets you prioritise the day's work in under 60 seconds.
Set a daily opening discipline: open the practice dashboard at the start of the day, sort by "oldest unreconciled transaction" to see which clients are falling behind, and triage before opening any individual client account.
Tiered Coding: Letting the Rules Do the Work
The coding rules engine in Reconlink operates in three layers:
Layer 1 — Deterministic rules: rules you have set, triggered by vendor key match, description keyword, or amount range. These auto-code transactions silently before you see them. A mature client engagement should have 70–80% of transactions auto-coded by Layer 1 rules.
Layer 2 — ML pattern matching: trained on the client's historical coding decisions. Transactions that don't match a Layer 1 rule are scored by the pattern model; high-confidence suggestions appear pre-filled in the coding UI.
Layer 3 — AI coding: for transactions where the pattern model confidence is below threshold, an LLM analyses the vendor description and amount to suggest a code. This applies to genuinely novel transactions.
Your job as the bookkeeper is primarily to review and confirm Layer 3 suggestions, investigate any transactions that produced no suggestion, and use those decisions to generate new Layer 1 rules (one click in the UI). Over time, the auto-code rate climbs and your manual review load shrinks.
Building Cross-Client Coding Rules
For clients in the same industry — say, five medical practices — many vendors are shared. Pathology payments, medical software subscriptions, professional indemnity insurance, and BMA memberships appear across all five. Reconlink's coding rule system operates per-client, but you can replicate rules across clients using the bulk rule copy feature.
When you create a rule for a vendor in one client's account and it resolves a pending exception for that vendor, check whether the same vendor appears as an exception in your other clients in the same industry. Applying the rule across those clients at once eliminates a batch of pending transactions without opening each client individually.
Batch Statement Import via Email Inbox
For clients who receive bank statements by email, Reconlink's per-client email inbox eliminates the manual import step. Each client has a unique email address — forward the bank statement from your registered address to that inbox, and the statement is automatically parsed and queued for reconciliation.
At scale, the workflow is: at the end of each month, send a quick email to each client reminding them to forward their statement to their inbox address. This outsources the data entry step entirely.
Exception Triage: The Inbox Zero Approach
Apply the "inbox zero" principle to coding exceptions. Every transaction that arrives in the review queue should be resolved the same day — either coded, matched to a rule, flagged for the client to explain, or parked with a note. Transactions that are not resolved accumulate and become harder to diagnose (the context is lost).
For transactions you cannot code without client input, use Reconlink's client portal to send a query directly from the transaction line. The client sees the transaction and the question; their response automatically associates with the transaction record. This keeps the communication in context rather than in an email thread where it will be lost.
Monthly Reconciliation Sign-Off
End each month with a formal reconciliation sign-off checklist:
- All transactions for the month are coded (zero unreconciled items)
- Closing bank balance in Reconlink matches the bank statement balance
- Any transactions in suspense (pending client response) are noted and the client has been notified
- The BAS worksheet (if applicable) is reviewed for anomalies — unusual spikes in GST collected or ITCs claimed
- Any new vendors identified this month have Layer 1 rules created
The BAS worksheet review is often skipped. It takes two minutes and catches two kinds of errors: miscoded transactions that should have a different GST code, and transactions coded with the wrong GST rate entirely. A GST error caught before the BAS is lodged is free; one caught after lodgement costs an amendment and potentially a penalty.
Client-Level Performance Metrics
Reconlink's practice dashboard surfaces per-client metrics that benchmark coding efficiency: auto-code rate (what proportion of transactions were coded without your review), average time-to-resolve (how quickly exceptions are cleared), and the number of active coding rules. These metrics tell you which client engagements are optimised and which need attention.
A client with a low auto-code rate and a high volume of manual reviews is either a new engagement (rules are still being built) or a client with unusually diverse transactions. The correct response is to invest time in rule-building for that client until the auto-code rate normalises — typically six to eight weeks of consistent rule creation.
Tracking these metrics monthly gives you the data to have a conversation with clients about fee adjustments: an engagement where your manual time has halved due to automation is one where the value narrative is very clear.