RPA adoption is rapidly increasing in businesses from different sectors. From banking to automotive, RPA’s potential to automate error-prone, repetitive, and time-consuming tasks can save companies time, money, and human resources that can be spent better elsewhere.
However, given all the benefits of RPA, RPA is not suitable for processes that:
- Need constant human oversight
- Are too complex to be automated
- Deal with unstructured data
- Will not yield a profitable ROI
- Immature processes
See the 5 unsuitable processes for RPA and a guide for transforming those processes into ones that can be automated.
1. Processes that need constant human involvement
Some processes, such as quality control (QC), customer care, sales, or customer feedback analysis, are examples of tasks that cannot, by their very nature, be fully automated. The important keyword here is “fully.”
For a more efficient customer care process, RPA can assist the rep in quickly retrieving all the customer’s previous interactions with the brand, their personal information, and other relevant data, such as their customer journey data, for more personalized recommendations.
However, the next step, which is actually addressing the specific concern, requires human intelligence, social skills, empathy, and finesse.
If a company uses an RPA-based chatbot, customers can quickly describe their issues, like “I’m having a shipping delay.” The command catalog includes trigger words such as “shipping delay” and offers limited responses to address customer concerns and suggest solutions.
So, processes that always need human involvement cannot be completely automated with RPA and then forgotten about. Instead, the RPA solution can assist the human staff in doing their tasks more efficiently and with fewer errors.
Human-in-the-loop-automation is a feature that can help companies partially automate their customer-facing processes, while simultaneously having the capability to delegate important tasks to humans when the robots can not go any further.
2. Processes that deal with unstructured data
RPA is not suitable for processes that deal with unstructured data. Unfortunately or not, it is estimated that 80-90% of data at the disposal of businesses is in an unstructured format. And, according to a Deloitte survey, only 18% of businesses take advantage of unstructured data.
So if a company wishes to automate a process that’s built upon unstructured data – such as invoice automation, credit scoring, and resume screening – they should first start sorting and categorizing them. And that is not something that can be done by RPA.
The solution lies in intelligent automation: businesses should augment their RPA bots with other AI capabilities, such as OCR and NLP, that will allow the software robot to pinpoint, read, extract, and convert user data into a machine-readable format.
So if businesses embark upon automating a process with RPA without data type transformation, there will likely not be much success.
You can also read agentic process automation to learn how to handle unstructured data.
3. Processes that are too complex
RPA is not suitable for processes that are too complex to begin with. So automation would only increase the complexity by adding more layers to it.
For instance, a bank in Southeast Asia had around a total of 2,000 RPA bots installed on employees’ computers. And most of what the bots did was copy-pasting data from one field onto the other, on predetermined schedules.
The issue is that there is always a chance of operational discontinuity if the PCs’ user interface (UI) changed as a result of an update. RPA bots function on pre-written, mostly simple, scripts and lack the ML capability to react to change.
Furthermore, the bank was unable to track which bot performed what because multiple departments and their bots were intertwined with one another.
Prior to RPA implementation, the IT team can leverage process mining to understand the as-is processes of the company to realize their inefficiency, and the actual level of their complexity, and get a realistic outlook on automation success possibility.
Failure to do such due diligence prior to embarking on RPA implementation can be costly and result in automation-disillusionment, as was the case with the South Asian bank.
Learn more about how you can start prioritizing and choosing the suitable processes for RPA automation.
4. Processes that will return a low ROI after automation
The cost of an RPA implementation process does not start and end with the solution itself. The pricing models of different vendors are only the first factor to consider in your cost model. Companies should also take into account:
- Maintenance costs,
- infrastructure costs,
- Any future programming costs are to be paid to an RPA developer to curate a more customized solution if the company feels like the (off-the-rack) no-code RPA solution purchased from the vendor is not helping the company reach its specific objective.
All these costs should be accounted for with respect to the economic and monetary output that automation would bring to the specific department or to those that are downstream. Overestimating the RPA’s ROI for a process could backfire on the company, cost them resources, and derail their future automation projects.
Since RPA bot prices are declining, this group of processes needs to be evaluated in line with prices.
5. Immature processes
Not all business processes stay constant. The way certain processes are done – for instance, the way that intercompany accounting is carried out – could keep changing in terms of the technical infrastructure of the software applications, employee and department head turnover, governmental regulations, financial close season timing, and more.
Processes that suffer from these constant changes are said to be immature for automation. And companies are not advised to spend resources automating such processes that are highly prone to overhauls in the short term.
We suggest companies take advantage of RPA vendors’ free automation-maturity-assessment-tests to get a realistic and unbiased overview of whether their to-be-automated process is actually ready to be automated or not.
The questions that are often on these tests include, but are not limited to, the following:
- Do you have your automation KPIs established for assessment following automation?
- Do you have, or plan to have, a department to oversee the automation process, such as setting up an RPA Center of Excellence (CoE)?
- Have you set up a change management department to assist the CoE in the transitory period?
- Do you have any automated workflows currently in place? Do you have past experiences with automation?
- Do you know the differences between attended, unattended, and hybrid RPA?
- Have you drafted an operating model for automation, whereby each department’s or employee’s task in the automation loop is clearly defined (and for which they can specifically be trained)?
- How familiar are your employees with automation?
- Have you clearly explained to your employees the consequences of automation on the operational outcome of the company and their future employment prospects?
For more on RPA
To learn more about RPA implementation, read:
FAQ
Cem's work has been cited by leading global publications including Business Insider, Forbes, Washington Post, global firms like Deloitte, HPE and NGOs like World Economic Forum and supranational organizations like European Commission. You can see more reputable companies and resources that referenced AIMultiple.
Throughout his career, Cem served as a tech consultant, tech buyer and tech entrepreneur. He advised enterprises on their technology decisions at McKinsey & Company and Altman Solon for more than a decade. He also published a McKinsey report on digitalization.
He led technology strategy and procurement of a telco while reporting to the CEO. He has also led commercial growth of deep tech company Hypatos that reached a 7 digit annual recurring revenue and a 9 digit valuation from 0 within 2 years. Cem's work in Hypatos was covered by leading technology publications like TechCrunch and Business Insider.
Cem regularly speaks at international technology conferences. He graduated from Bogazici University as a computer engineer and holds an MBA from Columbia Business School.
Comments 0
Share Your Thoughts
Your email address will not be published. All fields are required.