Stop chasing money.
Automate the chase.

Invoices were created in Excel, saved as PDF, and emailed to client contacts one by one. Follow-up calls were made after 7 days, and again after 14 days. There was no consistent process — some clients received two follow-ups, some received none. The debtor book sat at an average of R180,000 per month, with R62,000 more than 30 days overdue. The owner had no real-time visibility into who had paid and who hadn't.

The agent pulls client details and invoice amounts from Google Sheets on the 1st of each month, generates a branded PDF invoice per client, and sends it via WhatsApp with a payment reference. If payment isn't confirmed within 3 days, the T+3 reminder fires automatically. Escalation at T+14 copies the owner. Payment confirmation is handled by the client replying "PAID" with their proof of payment — which the agent logs automatically.

The three-stage approval chain existed only on email. If one approver was underground or on leave, the invoice stalled. Contractors were routinely paid beyond 30-day terms, damaging supplier relationships. There was no central visibility into which invoices were pending, approved or paid. Finance had to compile a manual status report every Friday.

Disputed invoices (wrong quantities, missing delivery notes) caused further delays — the dispute process was entirely informal and handled by phone.

Attorneys logged their time in a shared spreadsheet — but they forgot, delayed, or described matters inconsistently, making it hard to generate accurate fee notes. The bookkeeper compiled fee notes manually at month-end from the spreadsheet, had them approved by the responsible attorney, converted them to PDF, emailed them to clients, and then began the phone-based follow-up process. The entire billing cycle took 3–4 weeks from work completion to invoice out the door.

Trust account reconciliation was done monthly by the bookkeeper manually comparing bank statements to the trust register — a process that took a full day and frequently found discrepancies that required attorney attention.

We built a matter-closing trigger: when an attorney marks a matter as closed in the shared Google Sheet, the system automatically pulls the time entries, generates a fee note using the firm's template, routes it to the responsible attorney for WhatsApp approval (one tap — Approve or Query), and sends the approved fee note to the client via WhatsApp and email simultaneously. The entire process runs without the bookkeeper.

The debtor collection sequence then runs automatically: T+3 friendly reminder, T+7 reminder with PDF re-attached, T+14 formal notice with the firm's letter of demand template, T+21 escalation to the responsible attorney. Payment confirmation via WhatsApp stops the sequence and updates the debtor register. For trust account matters, trust receipts and refunds are automatically logged to a reconciliation sheet that flags any discrepancy above R100 immediately.

The billing process was entirely manual and dependent on site managers remembering to submit progress quantities to the QS on time. When they forgot, the claim went out late or the principal used the delay to reduce the certified amount. Retention releases required the site manager to notify the QS, who had to verify practical completion, issue a retention release certificate, and submit it to the principal's finance department — a process that happened months late if it happened at all.

When medical aid paid less than the practice's tariff, a shortfall statement was generated and sent to the patient by email. Follow-up was manual — a phone call after 14 days, another after 30 days. With 200+ consultations per month across three locations, only 65% of shortfalls were collected — the rest were written off or forgotten. Medical aid rejections (claims returned for correction) sat in the billing clerk's inbox for days before being addressed, delaying payment by weeks.

WhatsApp is how their trade clients communicate — but statements and reminders went by email, which many smaller trade clients checked sporadically. Phone calls reached some clients immediately and others never. There was no consistent escalation process — the same large clients were called repeatedly while smaller overdue accounts were ignored because clerks focused on the biggest balances. DSO of 52 days meant R1.2 million was sitting in debtors at any time.

Debit order failures generate a notification from the insurer — but these arrived by email and were actioned manually, sometimes 5–7 days after the failure. By then, a second debit attempt had often also failed and the policy was in lapse. The practice was absorbing commission clawbacks of R15,000–R20,000 per month on policies that could have been saved with earlier intervention. Advisory fees were invoiced by the bookkeeper monthly and followed up by phone — a process that was inconsistent and time-consuming.

The director invoiced herself on an ad-hoc basis using a Word template, emailing invoices from her personal inbox. There was no systematic follow-up. Clients who were slow payers remained slow payers because no one chased them consistently. Debtors over 90 days — money fully earned, services fully delivered — stood at R220,000. This was cash the business had already spent on salaries and costs but hadn't collected.