Automation for SMEs: a guide to start with criteria

When a request received by mail ends up copied into a spreadsheet, forwarded to another person and converted manually into a task, the problem is usually not just the time spent. Duplicate data, forgotten steps and doubts about what information is updated also appear.

The automation for SMBS consists of designing a flow so that an action or data activates repeatable tasks without having to execute them one by one. This guide answers a practical question: which processes it is best to automate first and how to decide it. When you finish, you can distinguish a useful opportunity from an automation that would only transfer the disorder to another tool.

If you still need to compare several tasks and decide which one deserves priority, use the diagnosis on what to automate. Here we start from an opportunity already identified to explain how to start the flow with scope and controls.

Direct response

It starts with a process that meets three conditions: it is repeated frequently, it follows fairly clear rules and generates a visible problem when done by hand. For example, register requests, assign a task, warn of a change of status or prepare information for review.

Don't automate a process just because it looks modern or because a tool allows it. Before, it clarifies what should happen, what data intervenes, who should act, and when a person has to review or decide. If the process is confusing, automating it can make the error repeat faster.

The aim is not to eliminate human intervention from any task. It is to reserve it for exceptions, trade decisions, validations and relationships that require criteria.

Context and scope: what can automate a SME

An automation connects actions and data. An event starts the flow; then defined steps are executed. For example, when you receive a full form, the system can create a record, notify the responsible team, and send a confirmation to the sender.

That flow can operate between existing tools or within an application. The right solution depends on how you work, the data you need to keep and the rules that the process must respect.

In a SME, automation is often useful in situations like these:

• A request always comes through a specific channel and must be registered with the same fields.
• A state changes and several people need to know.
• A document is prepared from already validated information.
• A task requires reminders according to a date, state or defined condition.
• Two systems contain related data and manual updating causes inconsistencies.

These are hypothetical examples. They do not imply that all businesses should automate those flows or that the same configuration serves everyone.

It is also important to differentiate between automation and digitization. Passing a paper task to a spreadsheet can sort the work, but not necessarily automate it. Automation appears when the system takes a subsequent step according to an agreed rule.

Practical criteria for choosing the first process

The best first automation is usually concrete. Instead of proposing "automating administration", it describes a complete sequence: from what happens at the beginning to what result should be ready.

Search for repetition with understandable rules

A process is a good candidate when you can explain it without vague terms. For example: "When an application is approved, a task is created for preparation and the assigned person is notified." If the explanation depends on "as the case may be" without the cases being specified, you first need to sort out the rules.

Everything doesn't have to be identical. There may be exceptions, provided they are identified and have an exit: stop the flow, ask for a review or send it to a responsible person.

Check the operating cost of manual work

You don't need to calculate an exact figure to prioritize. Note where waitings, data reintroductions, internal queries or corrections accumulate. A process that forces you to search for information in several sites may deserve attention even if it doesn't take many hours in a row.

Give priority to problems whose effect you can observe. For example, unresponsive requests, unallocated tasks, documents with different versions or changes that are not communicated.

Define the target before talking about tools

Separate goal, reach and solution avoid buying or building something before you know what you need.

If your goal is to better respond to requests, the scope may include capturing data, checking mandatory fields, assigning responsible ones and recording tracking. The specific tool is decided after understanding those needs.

Decide what should continue to validate a person

There are actions that require confirmation before continuing: approving an amount, modifying a commercial condition, accepting an exception or communicating a sensitive decision. An automation must make that point visible and not hide it behind an automatic rule.

It also defines what happens when data is missing, a connection fails, or a request arrives outside the expected assumptions. A useful flow not only describes the normal path; it contemplates how it stops and who receives the notice when something does not fit.

Application: from a repetitive task to a usable flow

Before setting up anything, it goes through a real or representative case from beginning to end. It notes the trigger, input data, actions, responsible ones and the result that must be recorded.

You can use this sequence to prepare the first range:

Describe the startup. Indicates what triggers the process: a form sent, a status change, a document received, or a date. 2.

Enumera the necessary data. Define what information should exist and what fields cannot be left empty. You prevent the system from generating tasks with insufficient information. 3.

Sort rules. Specifies what happens in the usual situation and what conditions change the route. 4.

Aresponsibles. Determines who receives each notice, who corrects incidents, and who makes decisions that the flow should not take. 5.

Define a verifiable result. It's not enough to say that the process will be more agile. Formulate a result that you can verify, such as that the request is registered, assigned and confirmed. 6.

Try normal cases and exceptions. Check what happens with incorrect data, duplicates, subsequent changes or bugs of a connected tool.

Imagine, hypothetically, a company that receives budget requests using a form. A first version does not have to generate budgets automatically. It can only validate the essential fields, create a request sheet, assign it according to the type of service and alert the team. The commercial criterion, the needs review and the proposal would remain human. That scope content allows checking if the information arrives complete and if the team can work with it.

The first version must include what is necessary to meet the defined goal, not all future possibilities. If you need new rules, integrations or panels later, you can evaluate them on a process that you know better.

Limits: when it is not appropriate to automate yet

Automation does not correct a misdefined process by itself. If different people perform the same task in incompatible ways and there is no shared rule, automating before agreeing to it can consolidate the conflict.

It is also not appropriate to delegate decisions that depend on interpretation, context or responsibility to an automatic flow. Technology can prepare information, classify according to agreed rules or warn of a situation. The final decision should be kept where a human assessment is necessary.

Pay attention to these limits:

-Untrustworthy data: If the source data is incomplete or contradictory, the result will be too. -Dependency of external tools: a connection can change, fail, or require maintenance. Define who monitors the flow. -Access and permissions: each person must have only the access necessary to perform their function. -Unresponsible exceptions: an unresponsible incidence ends up becoming an invisible task. -Excessive range: trying to connect all areas at once makes it difficult to understand what works and what needs to be corrected.

In addition, if the process processes personal data, contractual information, collections or other sensitive information, it is appropriate to review the applicable obligations and who can access each data. An automation does not replace the legal, fiscal or compliance validation that corresponds to your activity.

Next step: turns the idea into a defining project

Before looking for a solution, it brings together an example of the current process and answers four questions in writing: what activates it, what data it needs, what outcome you expect, and at what point a person should intervene. On that basis it will be easier to decide if you need to connect existing tools, adjust a flow, or develop a more specific solution.

If you already identify a repetitive task but are not clear about its scope, you can explain your automation project to AVSISTEC. The starting point will be to understand the process, its rules and its limits before proposing a solution.