With the current threats to business and processes from COVID-19 and other similar threats it is highlighting the weakness of having humans in the process. Outsourcing to different countries for cheaper pools of labour now seems more like a problem as each country varies its responses to the issues and these are all out of your control.
Your company may have a laptops for everyone and fast home broadband, but your software development vendor or Business Process Outsourcing (BPO) vendor, that provides customer experiences and back office operations for your clients, may have desktops and the staff only have a mobile phone at home.
So, what now. How can we get the same savings without a reliance on humans?
More and more CTO’s, COO’s and CIO’s are turning to a fast and emerging technology development called robotic process automation (RPA) to protect the business in time of volatile change with a strong focus on reducing reliance on humans, streamline enterprise operations and reducing costs.
With RPA, businesses can automate mundane rules-based business processes. This is certainly the initial focus – embedding straight through processing into human intensive processes. The PR would say this will enable business users to devote more time to serving customers or other higher-value work. Which is great but the focus here is reducing reliance on humans – so the real aim is doing the same or more with less cost, less people.
Once businesses have gone through this level of automation then it puts them on a fast track to adding intelligent automation (IA) which uses machine learning (ML) and artificial intelligence (AI) tools, which can be trained to process data much faster make judgments though computer vision or about future outputs. This adds resilience to a business and allows it to control its customer and back office processes.
What is robotic process automation?
RPA is an application of technology, governed by business logic and structured inputs, aimed at automating business processes. Fair enough. For the average company this means working though existing processes and ‘cleaning’ it up to allow higher levels of automation whether by simply developing internal automated processes or using RPA tools.
In this was company can configure the software in a process to, manipulating the data, triggering responses and communicating with other parts of the system given correct trigger responses or invoking exception management where they don’t.
RPA scenarios range from something as simple as generating an automatic response to an email or clearing a payment given that all triggers are passed to deploying thousands of bots, each programmed to automate jobs in a complete process.
Implementing automation with AI, such as computer vision can lead to great improvements in quality, speed and customer journeys. An example would be where a system ‘views’ documents and checks for completeness, errors and fraud at point of upload. This can take a person say 10 minutes, where a configured AI system can do this in seconds, 24 hours a day with no sick days or holidays.
What are the benefits of RPA?
Benefits vary from business to business, but the core benefits are reducing staff costs, reducing (human) errors, increasing speed, opening up capacity, hardening system uptime.
One of our clients had the following benefits:
- 18% of client total data were in error NOW 100% of data is now correct
- Duplicated images NOW No duplicated images
- Fraudulent cases discovered NOW All suspicious cases are flagged 83% of total data is incorrectly labelled NOW 100% of data is labelled correctly
- 24 different image sizes NOW 1 standard
- Over 28% of total data is unusable NOW 100% data is usable
- Process lacked consistency NOW New working process
- Less than 50% of total data was verified NOW 100% of data is verified
- Manually process which was time consuming and took 8 minutes to process NOW Ai automated and can do 100’s of cases in 8 minutes
Bots or process code updates are typically low-cost and easy to implement, requiring limited or no custom software or deep systems integration. This can give companies some breathing room so they can serve their business better through automating the low-value tasks.
Once a company is on a path though automation they can accelerate their automation efforts by implementing AI and ML with computer vision, speech recognition, and optical character recognition (OCR); fully automating higher-order tasks that in the past required the perceptual and judgment capabilities of humans.
When businesses are automating 10, 15, 25 step processes this verges on IA, Intelligent Automation, this uses AI and ML tools and solutions that are embedded into the business process.
What are the pitfalls of RPA?
RPA isn’t for every enterprise. As with any automation technology, RPA has the potential to eliminate jobs, which presents CIOs with challenges managing talent. Some businesses adapt, along with their workforce, and others don’t. Many businesses face challenges of lack of investment resources, long term outsourced contracts or TPA’s or poor historic data. These businesses need to evolve as they will be overtaken by those able to automate and implement RPA.
Issues when looking at RPA:-
- Installing bots can take a lot longer and is more complex and costly than most organizations plan for.
- The platforms on which bots interact often change, and the necessary flexibility isn’t always possible in the underlying platform or configured into the bot.
- New regulations that would just require a staff meeting and a bit of training can mean changes to an application and throw off months of work.
- Underlying data is not always available or even owned by the company. Outsourced contracts often overlook the value of data and process and focus on human costs.
10 tips for effective robotic process automation
1. Set and manage expectations
Quick wins are possible with RPA, but propelling RPA to run at scale is a different animal. Bold claims about RPA from vendors and implementation consultants need to be tested fully. COO’s and CFO’s should test capability first and bench mark claims with their own data.
2. Consider business impact
RPA is often propped up as a mechanism to bolster return on investment or reduce costs. But often over looked is to improve customer experience and to scale a business efficiently. A bot to help onboard a claim or a customer request can speed up processing time and make sure the ‘order’ or ‘claim’ is fully actionable before it reaches a human to process. Validation is a key area for computer vision and can improve a business considerably.
3. Involve IT early and often
There is a problem with low code systems and RPA software solutions as many initially buy hit a wall during implementation, because they have not considered the underlying issues in their existing platforms and data. A new phrase of "citizen developers", like “citizen project manager”, without technical expertise, business heads must involve IT from the outset to ensure they get the resources they require.
4. Poor design, change management can wreak havoc
Many implementations fail because design and change are poorly managed, this is part of 3. above, but can also result from the rush to get something deployed. RPA is a long term process and one that should be taken in steps, it’s a journey. The way to think about it is that you are remodelling your whole business around increased availability, reduced costs, better customer experience and reskilling/upskilling staff. There is no rush, get it right bit by byte.
5. Don't fall down the data rabbit hole
Deploying bots and updates to automate manual data entry or to monitor software operations generates a ton of data. This is a good step. However, this can lure CIO/COOs to start looking to leverage the data. Suddenly, the RPA project has become an ML or AI project that hasn't been properly scoped or understood as an ML/AI project. The scope keeps moving and team struggle to catch up to it. Consider RPA as a long-term arc, rather than as piecemeal projects that evolve into something unwieldy. Big goals should frame a constant set of changes. It’s a journey, with a map and a destination.
6. Project governance is paramount
Another problem that pops up in RPA is the failure to plan across the company if you take to bigger step in one place it can cause issues in another. Teams must constantly check for chokepoints where their RPA solution can bog down. Monitoring and alert systems to watch for hiccups impacting performance should be built in as metrics can suddenly explode – you don’t want to install your own version of a virus that impacts across the business. Command and control is essential.
7. Control maintains compliance
There are lot of governance challenges related to instantiating a single bot in environment let alone many. Let’s say you have a customer facing bot – should it be female or male? Suddenly this is an issue you were not expecting.
8. Get an expert and test
Make sure you use an expert. Scalability is critical and many RPA projects do not deliver results because the solution looked great in POC but when faced with live data they fail. An experienced vendor should focus on the company solution not try and bend the company into their solution. The RPA team must own and develop business cases, calculating potential cost optimization and ROI, and measures progress against those goals. Ability to scale is essential IT leaders across different industries must look for opportunities and understand whether RPA will be transformative for their businesses.
9. Don’t forget the impact on people
Wooed by shiny new solutions, some organizations are so focused on implementation that they forget the fact that once implemented and scaled up, the human part of the business is scaled down. This may look good but what happens if something goes wrong or a vendor sells their business and they now don’t want to support your contract? Continuity planning becomes huge when AI / ML / RPA projects are in place.
10. Put RPA into your whole development lifecycle
Team must automate the entire development lifecycle or they may kill their bots and Apps during a big new launch or update. This needs to be built in. Again step by step is the best process building up capability and testing each step before it goes live is essential. An intelligent automation ethos that must be part of the long-term journey for businesses.
11. IPR ownership and a warning
Who owns the IPR and process? This is a big question that needs to be built into the kick off of the project. Imagine this scenario……
….. you have an outsourced process like a call and process centre dealing with claims or complaints or bookings or delivery.
You decide that the people part of this can be reduced by 75% though RPA / AI project and you contract a software as a service vendor to implement this. The vendor has a proven capability and service other companies similar to yours.
You sign a 3 year deal and you start to feed your process and data through their system and let go your people and divert the process on their process and you get the savings and reports. All good.
Over the three years you have become dependent on this vendor but you notice that they are starting to service more and more customers and your competitive edge is under threat. You want to move to a new vendor or bring this in house…..but how can you start….they own the neural networks and complex coding to process your businesses. You are faced with staring from scratch without a human back up plan and or the ability to transfer any of the lessons or learning over.
Make sure you own the IPR and your code and solutions are licensed in such away that you own the process. Even if the provider(s) is sold, goes out of business, becomes a competitive threat. Its not the same as transferring (an employee or group of employees) to another employer under TUPE regulations. You can’t ask for the code and process and transfer that! Once you handover the data and process to a vendor you have lost a lot of flexibility!