Skip to main content
The ai Corporation

Key steps to consider before starting your journey to automation

If you're not aligned with the right outcomes, and if leadership is lacking, technology can only take you so far.

The automation of processes has advantages in many areas of business. Helping to create predictable success, like the autopilot technology on a plane has been perfect over many years and by decreasing the time that it takes to complete manual processes and removing the chance of human error. Automating processes also has the benefit of streamlining workflows.

The largest gains can be achieved by automating very large or time-consuming processes. It is not unheard of for processes, which usually take days to perform or require continual manual input, to become almost completely automated. If the automated process involves any machine learning, it is not unusual to see performance increase, due to the speed at which models can be trained and deployed.

Businesses across multiple sectors are automating time consuming processes to drive efficiencies and reallocate their resources. Giving time back to staff, who usually carry out a process, which can then be used for more creative work, such as being able to do more for their customers. Which, in turn, leads to happier, engaged customers, who are likely to buy your products or recommend you to their peers.

A key criterion for a successful automation project is explain-ability. Understandably, humans do not trust a ‘score’ that lacks reasoning. Indeed, within the European Union, any AI decisions that have an impact on customers have a legal requirement to be ‘explainable’ (GDPR Recital 71). In the US, the Equal Credit Opportunity Act, Title 12, requires sharing the reasons for an adverse action. Regardless of the regulatory landscape, AI outcomes must be explained.

Customers are also looking for solutions from vendors, which augment and improve the systems and processes that they currently have in place, rather than replace them. This is primarily down to cost, both monetary and time wise; but AI remains on trial in many scenarios. With many checking whether the solution or service they have been sold can live up to expectations and requirements.

What to automate and when

Before starting your journey to automation, the first step is to identify which processes would benefit from being automated, and to assess whether automation is possible. A good example of a high value process for automation is a process that involves report generation. Reports can be similar each time they are published, yet there is still manual effort involved in generating the content (such as metrics, charts, etc.), adding that content into the report, and performing any editing. An example of a process where automation may not be possible are processes which are dynamic, relying on many people or departments.

Our team have been focused on automating the fraud management process. This has been an enormous task, and it has taken us many years of trial and error by automating manual processes into one that only requires 10 minutes of management and user attention per day. Automating the fraud management process is a good use case for automation, as there are several manual processes, including models and rules performance monitoring, fraud pattern discovery and fraud alert management.

While these manual processes may be manageable at first, as the number of payment types and channels increase, it can quickly become unmanageable to add more and more staff to manage and monitor those processes. With fraud on the rise, more rules and models also generally mean more alerts, which usually lead to higher incident rates. Combine those two aspects and managing fraud can become very expensive, which is why efficient management processes are so important.

Many rules-based fraud detection solutions require near round the clock monitoring to ensure effective fraud detection. Even with machine learning doing some of the ‘heavy lifting’, there is still a large amount of monitoring that is required to keep on top of fraud. This, coupled with the rise in electronic payments, means it is becoming too time consuming to manually discover fraud patterns and write effective rules, or create machine learning models to combat fraud. However, AI/ML based fraud systems are quickly becoming the standard in the fight against fraud. Automating the process of spotting current fraud patterns and flagging certain behaviors that keep occurring to be checked by analysts.

Automating processes involving machine learning model generation can also lead to improved performance, as more models can be created and only the best selected for deployment to production. This automated process can run continually and is not restricted to office hours.

Effective planning – Impact analysis

The next stage is to plot each of the processes that you are considering automating into 4-boxes, which will reveal the order of the processes to automate:

  • Low impact, low effort - (Convert quickly or stop)
  • High impact, low effort - (Quick wins)
  • low impact, high effort - (Stop)
  • High impact, high effort - (Longer wins)

It is worth remembering that automation does not require automating an entire process in one go. You can see the benefits of automating parts of a process. Automating a process gradually can bring tremendous value to your journey, especially if you are choosing to automate many processes at the same time. A gradual approach will also give you the tools and evidence to convince relevant stakeholders of the value of any wider automation activity.

Before jumping in and developing an automation tool, you will need to decide where and how the process will run. Any automation tooling needs to be deployed somewhere that fits into the process, otherwise a new process will take its place and require additional automation. You should also consider whether the automation tool that you plan to develop could be used to automate additional similar processes, thereby increasing the power of that tool.

It is certainly worth the time to research and thoroughly plan how each process can be automated. Determining if any processes can be grouped together and assessing how that might work. This ‘due diligence’ will save an enormous amount of effort down the line and will enable your team to become more effective, more quickly.

Working with SMEs

If one of the processes you are looking to automate involves a subject matter expert (SME), it is crucial they are consulted, to ensure you fully understand how their team performs a particular process, and all of their usual actions can be captured by the automation tool.

For example, do they use a spreadsheet downloaded from an email attachment to perform a task in a database? Does their activity depend on their expertise? Will they want to be able to change certain behaviors later?

All this needs to be fully understood and accounted for before a single line of code is written. By the way, it is OK to go back and update the tools over time, but this type of back tracking should be reduced to ensure the most efficient systems are in place and the most benefit is gained.

Dedicated automation teams

While not essential, it is a good idea to set up a dedicated team for the automation of any processes. Assign proper resources and treat any automation project, as you would any important customer project, because it is very likely that by automating a few simple projects, you will see returns in line with some of the revenues generated from top line customers.

It might sound simple, but your automation team’s only task should be automation. Task them with automating anything, which is repetitive or time consuming and watch the wider team’s frustrations disappear over time.

Problems along the way

Our automation journey has not been easy, we have had to learn as we have progressed. Which sometimes meant we had to back track and redevelop things for a variety of reasons. We have had challenges with nearly all aspects of dealing with data, from acquisition, right through to gaining insights.

Developing automation tools as products brings its own set of problems. One of the biggest issues my team faced was engineering ways to reduce or entirely remove user interactions, whilst still making our product perform how the user wanted it to and giving them choices when they were required.

We planned for the short term, but neglected the long term, meaning that it has taken us much longer than we initially wanted to get to where we are today. Remember to map your entire automation journey before you start, as you may find that there are several stages which can be cut out completely. For example, if your goal is to fully automate a large process consisting of several steps, you may wish to automate each step separately. Then work out how you can avoid automating the smaller processes individually, as there will most likely be further work to be done to migrate those smaller processes into the larger automation picture.

Automating processes is a long, often arduous task. Requiring as much planning as a full-scale project, but the benefits are clear. By taking the time to automate properly, you will be surprised at how much more slick your business processes will become, how much happier and effective your staff will be, and most importantly, how happy your customers will be with the extra time your team has to help them with their requirements.

Search