When it comes to RPA implementation timeframes, there are multiple estimations given by several practitioners and vendors. The RPA tool vendor may state this time frame in weeks, an implementation specialist may say it is a few months (2–4), and you might also have heard of implementations that have taken a longer time. The right way to look at timeframes may depend on various factors. Here, we have discussed the key factors based on our RPA implementation experience at Imaginea.

First-time implementation

The reason for your RPA implementation could range from reducing costs and errors to improving process efficiency, or it could be a part of your digitization efforts. Whatever the reason, you need to do the following:

• Select the right automation tool(s)
• Choose the vendor(s)
• Make sure you have procured the right licenses (from RPA tool vendors or any other auxiliary tool (OCR) licenses)
• Do a Proof of Concept
• Conduct a pilot program
• Set up a Center of Excellence
• Set up a governance framework
• Ensure that infrastructure readiness assessment is done
• Set up 24/7 Operations Center for bot process maintenance

For first-timers, like any other initiative, chances are that the duration might be longer than the estimated schedule.

Using the existing environment setup and IT policies of the company

Often organizations tend to overlook existing infrastructure and policies, which leads to chaos at various stages. For example, using the same versions of the technology stack in all environments can ensure that the application runs properly after production. Likewise, a policy that does not allow code change in any environment, except Dev, can help prevent unwanted surprises. Here are the main points to keep in mind:

• Make sure all your environments (all or subsets of Dev, QA, Stage, Production) are ready. More the environments, more the deployment efforts and testing efforts (deployments/ testing can be automated) in each environment. But it is necessary since each and every environment has its own reason for existence.
• New policies might have to be defined for the bot accounts at the enterprise level along with modifications of all the application policies that the bot may interact with.
• In case of new bot account segment, come up with IT policies, which will determine how accounts are given permissions, levels of access, and expiration policies.

If environments are taken care of during the initial setup, the time taken for future deployment of new functionalities or new processes will be considerably lower.

Managing implementation partners 

There is a widespread notion that an RPA implementation does not require IT experts. This is not true. You will need IT experts at least to some extent. In fact, it becomes advantageous to have IT vendors who come with multiple enterprise implementation knowledge and experience. They also come with a host of reusable components that they already have built, which will speed up development. Let us see how RPA implementation timelines get affected according to the implementation teams.

• New — Inducting any new vendor into an enterprise takes time, be it for RPA or any other implementation. Starting from acquiring access to knowing the systems and processes, it takes time. This may not be true for seasoned IT vendors (like Imaginea), who have worked with multiple enterprises and are already aware of the protocols. In fact, such seasoned vendors may also add value to the existing system, as they might be aware of most of the enterprise needs — even though their processes may be different.
• Existing — With an existing partner, there may not be a learning curve to study the systems, processes, etc. Therefore, the implementation will be faster here as they may be well versed with the client’s environment.
• Internal — Some organizations will have an internal IT team that might have been ramped up for RPA implementation. This solves the problem of getting to know the infrastructure. The business process knowledge may or may not be a problem. However, caution must be exercised when dealing with an RPA implementation.

Procuring licenses

Acquiring the necessary licenses is a complex process since there are multiple parties at play. The governance team has to carefully consider these before making recommendations for procuring licenses:

• Based on business needs, identify the number of processes that are required to be automated.
• Estimate the bot requirement for each process.
• Now, forecast the volume of bot licenses that are required. If you get more and don’t use it, you lose. At the same time, if you get lesser, last minute buys would be a time killer and at the same time, you will lose on volume discounts.
• Consult with implementation partners on what type of license needs to be procured for which process type (e.g., assisted/unassisted bots). Identify if there is a necessity for additional licenses for OCR or other tools. Sometimes these auxiliary licenses are sold by the RPA tool vendors and sometimes they have to be procured separately.

The governance team plays a crucial role since RPA implementations cannot be technology driven nor can it be just business driven. It needs multiple teams — from the upper management to the people on the field — to be educated and aligned on the purpose of every RPA implementation.

Acquiring assets

Getting the infrastructure for an enterprise level implementation is a task on its own. The complexity increases with the compliance requirements of the organization (SOX, GDPR, etc…).

RPA vendors, Implementation vendors, IT infrastructure, and business teams have to work together, and they must all be bonded by a strong governance team. This team should be looking at these factors:

• Process complexity
• Number of steps in a process that needs to be automated
• Number of application integrations that the process has
• Data formats (structured/ unstructured)

RPA vs IT Automation vs Cognitive solutions

RPA is a surface technology that mimics human actions. In a typical enterprise process RPA implementation, the solutions quickly drift towards cognitive because most processes deal with data. We need to exercise caution since dealing with data is a different ball game.

Any solution that needs just an RPA implementation will be faster than any cognitive solutions. A process that deals with semi or unstructured data will need some degree of cognitive solutions, and this, in the strictest sense, goes beyond the traditional RPA. At least the expectation of RPA implementation in weeks does not take cognitive solutions into the picture.

Sometimes processes will need a level of IT Automation or a Business Process Automation along with RPA. The process selection for RPA needs to be carefully done for such processes. However, in the course of time, this scenario might become inevitable and take more time than a traditional RPA implementation.

Conclusion

So, how much time should an RPA implementation take? Any attempt to answer this question without considering the factors, mentioned in this discussion, becomes futile. After satisfying the needs and requirements of these key factors, achieving a closely realistic RPA implementation timeframe becomes easier.