You’ve completed your proof-of-concept. You’ve launched your first process automations into production, using RPA or other forms of automation. You’ve seen efficiency gains in teams supported by a virtual workforce. Interest in automation has grown as a transformative force and the reputation of your digital assistants to efficiently deal with repetitive work has spread. Now ideas are flowing in. So how do you sift through them and scale your intelligent process automation practice?
In this multi-part article series, I’ll share some tips and techniques for taking your automation intake process to an enterprise scale and optimizing your assessments.
Tip #1: Take a Holistic View.
The journey of intelligent process automation starts with technology-agnostic Process Discovery. After all, you can’t automate what you don’t know about. Often, IT departments are unaware of the full extent of the manual processes and workarounds that business users actually go through to accomplish their tasks due to limitations in system functionality.
Engage business leaders about the transformative power of automation, and perform a breadth-first cataloguing of processes. Some key data points at the initial stages to collect:
- “Elevator Description” – how would you describe the goal and summarize the current steps of the process to someone unfamiliar with the domain?
- Metrics – how many people perform the process, how many transactions are processed, and how long does each transaction typically take?
- Systems Involved – what systems and applications are currently used to complete the process?
- Processing Windows – what is the earliest the work can be started? When is the latest it can be completed? How fast does a transaction need to be processed?
- Benefits – what benefits would automating this process bring?
A common temptation at this point is to market a particular type of solution during these discussions. Coming to that conclusion is a deeper level solution assessment, and in order to streamline the discovery process and not commit to a significantly sub-optimal solution, I highly suggest not pitching particular solutions during process discovery and to remain open-minded about what the solution(s) will be.
Tip #2 – Ask Why. Ask Whether.
The Parable of the Pot Roast has circulated on the Internet since the mid-1990’s:
The lesson of this story applies to process assessment.
- Ask why users perform the process
- Ask whether the process needs to be done the same way, or at all
Automating a bad process is still a bad process, and automating an unnecessary process is still unnecessary work.
Tip #3 – Use Intelligent Economics.
With a top-down approach, you’ll have a more comprehensive holistic view of many different processes. A natural tendency is to individually select the one with the highest business potential for savings and exclusively focus on that for deeper-dive analysis.
Regardless of whether you use a basic value calculation or WSJF, overall value is a calculation that not only looks at the benefits, but also considers the cost/duration of the solution. Filtering processes without an indicator of cost can result in coming up short on value potential.
Both basic value equations (Benefits – Cost) and WSJF (Cost of Delay / Duration) have a weakness in that they do not factor in economies of scale sufficiently. This is fine for projects that don’t feature a high-degree of commonality and reuse, but with automation technologies like RPA, commonality and reuse across automations can be high.
Using the data collected by taking a holistic view (Tip #1),
- identify all unique systems for the process set, and
- group processes that use similar systems.
Instead of assessing processes individually, assess with consideration of the groups. The first automation solution in a group will have a normal estimate, while the subsequent automation deliveries will be accelerated. You may find a group of processes that provides significant more business value when considering the economy of scale.
Take these first steps in cataloging your processes, and in the next article I’ll expand more on how this model can result in higher velocity of automation delivery.