What business processes should you automate?
Automate processes that run often, follow rules you could write down, start from a clear trigger and touch data that lives in a reachable system. In most businesses that means lead intake and follow-up, quote and invoice chasing, onboarding steps, data moving between tools, reporting and first-line replies. Leave processes that are still changing, run rarely or carry legal or financial risk without a human checkpoint until they are stable and you can put a person at the decision point.
How do you score a process?
Give each candidate a quick score from 1 to 3 on five questions:
High totals go first. A process that scores well on frequency and rules but poorly on data access usually needs a small data fix before it is worth automating. Keep the list and rescore it after each build, because every automation you add makes the next one easier to connect.
Which processes usually come first?
Lead intake and first reply. Trigger: new enquiry from any source → steps: deduplicate, log in CRM, classify, send acknowledgement, assign owner → output: every lead answered in minutes and owned by someone.
Quote and invoice chasing. Trigger: daily schedule → steps: find quotes with no response and invoices past due, send the next message in a polite sequence, log it, flag anything past the final reminder for a phone call → output: cash flow that does not depend on someone remembering.
Onboarding. Trigger: deal marked won → steps: send contract for signature, create the client folder, set up accounts, book the kickoff call, post a handover note to the delivery team → output: every new client starts the same way, even when the usual person is on leave.
Reporting. Trigger: Monday 7am → steps: pull numbers from the CRM, ad platforms and accounting tool, calculate the agreed figures, post a summary → output: the report is waiting before anyone asks for it.
Document handling. Trigger: attachment arrives in a shared inbox → steps: AI step extracts the fields from the invoice, order or form, checks them, files the document and writes the data to the right system → output: no re-typing, with anything unclear sent to a review queue.
What should you not automate yet?
- A process that changes every month. You will rebuild the workflow every time it changes. Run it manually, write it down, and automate once it has settled.
- A process that runs a few times a year. The build and maintenance cost more than the hours saved. A checklist is enough.
- Final decisions with legal, financial or reputational weight. Refunds above a limit, contract terms, anything a regulator would ask about. Automate the preparation (gathering documents, summarising, routing) and keep a person on the decision.
- A process nobody owns. If nobody can say how it should work, the automation will encode whatever happened last time.
What breaks when you automate the wrong thing first?
The most common mistake is starting with the most painful process instead of the most automatable one. The most painful process is often painful because it is full of exceptions, involves several departments and has inputs in ten formats. Automating it first means a long build, a shaky launch and a team that decides automation "doesn't work here".
The second mistake is automating a broken process. If quotes go out late because nobody agrees on pricing, an automation will send unclear quotes faster. Fix the rule, then automate it.
What does it take to keep these running?
Each workflow needs an owner, an alert when it fails and a short document saying what it does. Simple flows (a form to a CRM) need almost no attention once stable. Flows that touch several systems need someone to read the alerts and handle the occasional expired login or renamed field. Plan for that time from the start rather than discovering it later.
What would this look like for an accounting practice?
Say you run a 10-person accounting and VAT practice in Dubai with around 150 monthly clients. Scoring the processes might look like this:
- Chasing clients for monthly documents: runs every month for every client, clear rules, clear trigger (month end), high cost of delay. Automate first.
- Logging incoming documents: daily, rule-based once an AI step reads the files, data lands in a shared inbox. Automate second.
- VAT return preparation: frequent, but the judgement is the service. Automate the data gathering and leave preparation and sign-off with the accountants.
- Annual engagement letter updates: once a year, varies by client. Leave manual.
The first two builds remove the chasing and filing that fill the juniors' weeks, and give the partners a live view of which clients are behind.
How does WebBox approach this?
We sit with the people who do the work and list every repeated process, then put an hours figure against each one: how often it runs, how long it takes, how many people touch it. Each gets scored on the questions above, and we tell you plainly which ones to leave alone, including when the answer is "not yet".
The top one or two get built first, against your real data, with error handling and a named owner. Then we rescore. Most businesses find that the second and third builds go faster than the first, because the connections to the CRM, inbox and accounting tool already exist. For a sense of how far this can go, see the case studies, including an operations system that took over the incident work of a full-time role for a digital publishing business.
Related questions
Something else? Book a call and ask us directly.
Keep reading
How WebBox helps
Not sure where your hours are going?
We will list your repeated processes, score them and tell you which ones are worth automating and in what order.