You prepare a quote in 15 minutes. The customer receives it two days later. Should automating its preparation really be your first step?

Imagine a company where the sales team prepares quotes quickly. Before a quote can be sent, the director has to approve it. Because the director has other commitments, the quote waits until the next day. Sometimes longer.

Management is considering AI that would cut quote preparation from 15 minutes to five. That would save some work. But the customer would still wait almost as long for a response.

Perhaps the company does not need a faster tool. Perhaps it needs a different approval rule.

This is an often-overlooked aspect of business process automation: a faster task does not necessarily mean a better process. Before choosing technology, it makes sense to establish what should be removed, what should be simplified and what is actually worth automating.

The quote example is an illustration, not a Yoptis client result.

1. What is business process automation?

Business process automation means having software, integrations, predefined rules or artificial intelligence systems carry out repetitive tasks partly or entirely. Examples include transferring data between applications, processing documents, routing requests and providing order status updates.

It is important to distinguish between three different things:

  • Digitalisation moves information or procedures into a digital environment. A paper form becomes an online form.
  • Automation enables a particular step to take place without manual intervention each time. Data are automatically transferred to another system.
  • Business process optimisation improves how work moves from start to finish. It reduces unnecessary waiting, handovers, errors or rework.

A process can be fully digital and still be slow. It can be automated and still require approvals that add no value.

Not every task that can be automated is a task worth keeping.

That is why, in business process optimisation, we first examine the path of the work and its constraints, rather than a list of available tools.

2. The most expensive part of a process may be the part where nobody is working

Let us return to the quote. Preparation takes 15 minutes, while waiting for approval takes two days. If we automate preparation and reduce it to five minutes, we have improved an individual task. But we have not necessarily improved what matters to the customer: when they receive the quote.

Three measures need to be distinguished:

Measure What does it tell you? Why does it matter?
Processing time How long someone actively works on a task. Reveals workload and opportunities for automation.
Waiting time How long a task sits before the next step. Shows bottlenecks that faster execution does not remove.
Throughput time How much time elapses from the start to the end of the process. Often best reflects the customer’s experience.

If a company prepares quotes faster but they all wait for the same approval, even more unfinished work may accumulate before that approval step.

There is also a less obvious possibility. If waiting cannot be shortened because a justified control is required, you can clearly tell the customer when to expect a response and what the status of their quote is. The actual time may stay the same, but uncertainty may decrease.

These are two different improvements: one changes the flow of work, the other the experience of waiting. Which matters more needs to be tested.

The customer does not judge how quickly you prepared a document. They notice how long they wait for the next step.

3. Check three options before automating

First, remove what is unnecessary

Duplicate data entry, reports nobody uses and approvals left over from an earlier organisational structure can become so familiar that nobody notices them anymore.

Ask yourself: What would happen if we did not carry out this step?

If removing it would not create an unacceptable risk, lose necessary information or breach a regulation, removal may make more sense than automation. Required controls and traceability must, of course, be assessed appropriately before any removal.

Then simplify what needs to remain

The director may need to approve quotes with non-standard terms. But does the director also need to approve every routine quote that already meets the agreed rules?

Clear limits of authority, fewer handovers and an established route for exceptions can reduce waiting without a new application.

Only then automate repetitive work

The most suitable tasks have a clear purpose, sufficiently stable rules, reliable inputs and predictable exception handling.

These may include transferring approved data between systems, preparing standard documents or sending order notifications. For sensitive decisions, the human role must also remain clear.

The order matters: remove, simplify, automate. Sometimes the right decision is also to keep a particular step manual.

4. Which business processes are most worth automating?

The best first candidate is not necessarily the task that takes employees the most minutes. What matters more is where the business is being harmed and which change you can meaningfully test.

What do you notice? A common first response What should you check before investing?
Quotes arrive too late. AI for preparing quotes. Are quotes waiting for preparation, information or approval?
Orders are piling up. Hire another employee. Which step limits order completion?
Employees are entering data manually. A new business application. Can the existing systems use the same approved data?
Complaints take too long to resolve. More support staff. Who owns the case, and who can make a decision?
Customers frequently ask for status updates. A chatbot. Is the problem a slow process or a lack of information?

None of the usual solutions is necessarily wrong. What is wrong is choosing one before checking the cause.

The most expensive improvement may be one that efficiently fixes the wrong part of the process.

5. Five questions before you buy a new tool

Take one recurring process, such as the path from receiving an enquiry to sending a quote. Then answer:

  1. Where does the process start and end? Measure the result experienced by the customer or the next department, rather than just an individual task.
  2. Where does work wait? Separate active execution from the time between steps, handovers and decisions.
  3. What is repeated? Look for duplicate entries, documents returned for correction and decisions made again.
  4. What could you safely remove or simplify? Check rules, responsibilities and exceptions before choosing new technology.
  5. How will you know the change is useful? Define the main success measure and at least one safeguard against unintended consequences.

First review one actual completed case, then check your findings against others. A single case can reveal a possible cause, but it cannot prove that this is the most important constraint in the entire process.

A small trial instead of a large project

Suppose you suspect that standard quotes wait too long for approval. Instead of immediately buying software, you could first test a change in approval authority for a clearly defined group of quotes.

The hypothesis could be:

“If routine quotes within the agreed terms no longer wait for additional approval, the time until a quote is sent will decrease without an increase in errors or unsuitable terms.”

Test the change on a limited scale, provided risks and internal rules are appropriately managed. Compare the time from enquiry to quote and monitor quality.

If the hypothesis does not hold, you have still gained an important answer before making a larger investment.

6. When should you use AI, and when should you use conventional automation?

Not every task needs artificial intelligence.

Conventional automation is often suitable when rules are clear and data are structured: for example, transferring an approved order to another system.

Artificial intelligence can help classify messages, identify information in documents and prepare drafts. It requires quality checks, data protection and a clear procedure for exceptions.

A process change without new technology makes sense when most time is lost through waiting, unclear responsibilities or unnecessary steps.

Do not start by asking “Where can we use AI?”. Start by asking “What do we want to improve, and why is it not happening today?”

7. How can you measure whether automation was worthwhile?

Saving ten minutes on a task does not necessarily mean the customer received the service sooner. Saving ten working hours does not automatically mean ten hours’ worth of lower costs either.

Before the change, choose three types of measures:

  • Business results: for example, time until a quote is sent, the share of orders completed on time or processing cost.
  • Workflow: for example, waiting time, number of handovers, amount of unfinished work or rework.
  • Safeguards: for example, errors, complaints, safety deviations or additional manual work caused by the new solution.

Measure before and after the intervention using cases that are as comparable as possible. Include implementation, maintenance and exception handling costs in the assessment.

If an individual task became faster but the overall process did not improve, this is not necessarily a failure of the technology. It may be evidence that the wrong constraint was chosen.

8. Frequently asked questions about business process automation

How much does business process automation cost?

The cost depends on existing applications, data quality, the number of exceptions, integrations and maintenance. Before estimating a price, you need to define the process, the expected benefit and how to measure it. Otherwise, you are comparing solutions to a problem that may not yet be properly defined.

Does automation also make sense for smaller companies?

Yes, when a task recurs frequently and has a significant business impact. In a smaller company, however, the best first improvement may be entirely non-technical: clearer authority, less duplication or a better handover.

What is the difference between business process optimisation and automation?

Optimisation establishes what makes sense to change and why. Automation is one way to implement that change. A good process is not necessarily the most automated one, but one that reliably achieves its purpose with the least unnecessary work.

Where should we start if we do not know which process is causing problems?

Start with recurring delays, rework, waiting for decisions or customer questions. Choose one process with a limited scope, review how it actually works and formulate a testable hypothesis about the constraint.

Before automating a problem, check whether you can remove it

You may indeed need a new tool. Perhaps another employee. But the biggest improvement may be one fewer unnecessary approval, clearer responsibility or timely information for the customer.

The right question is therefore not just:

“What can we automate?”

It is:

“What is stopping work from flowing better today, and what is the smallest change worth testing?”

At Yoptis, we first examine the actual flow of work, waiting, handovers and repetition. We then formulate a hypothesis about a change that can be tested.

If you have one recurring process where the team is busy but the result is delayed, Yoptis Flow Check may be a first step. Within a limited review of one process, involving up to 16 hours of specialist work, you receive a written Flow Report and one testable hypothesis for improvement. Implementing the change and measuring the effect of any pilot are separate steps.

Before investing in a new solution, check whether you are solving the right problem.

Check whether your process is suitable for Yoptis Flow Check.

An enquiry is not an order. We first check whether the process fits the scope of the service.