Intelligent Process Automation (IPA) is hot right now, there’s no doubt about that. McKinsey estimates that 85% of a typical firm’s 900+ processes can be automated. Suddenly every consulting firm on the planet has an RPA practice. And Gartner, Forrester, and many other analyst firms continue to pound RPA down the throats of the C-Suite at the Fortune 500.

But there are some serious issues that are giving RPA a black eye. As I wrote in a previous blog, nearly half of all RPA projects fail. Even more alarming is that on top of implementation failures, RPA simply has not scaled. Most organizations that have implemented some form of RPA have not made it beyond a handful of business processes after several years of effort.

RPA at Scale

According to a Deloitte study, organizations that have implemented RPA at a substantial scale (50 or more automated tasks) went from 3% to 4% between 2017 and 2018:

Deloitte Study - RPA Implementations

Maturity of RPA Implementations. Source: Deloitte

Top 3 Barriers to Successful RPA Implementations:

  1. Process Fragmentation – business processes can span many systems within an organization. Companies that have implemented RPA have typically taken a siloed approach by automating individual tasks inside individual systems without being able to logically bring these automations together into an end-to-end business process.
  2. Lack of IT Readiness – contrary to what you’re hearing on the street about the “citizen coder,” RPA is typically implemented by IT departments that are backlogged with many other competing initiatives.
  3. Lack of Clear Vision – executive teams are not engaged and goals are not being set, leading to many one-off initiatives that fail to move the productivity needle.

While #2 and #3 can be handled with better planning, prioritization, and engagement with senior leadership, Process Fragmentation is the silent killer in the world of process automation.

Connecting the Bots

The crux of the RPA scaling issue is that end-to-end business processes are almost never confined to a single system, while RPA bots typically are isolated to individual tasks in individual systems. In order to bring these bots together and expose them to business users in a meaningful way, organizations have a few options:

  1. Build their own user interface to connect the bots. We’ve seen custom .NET, Java, and HTML applications built to bring it all together.
  2. Use a Business Process Management (BPM) platform and integrate it with their RPA bots.
  3. Use an Attended RPA platform.

Attended RPA gets bots in the hands of users through the intersection of BPM and RPA, which Craig Le Clair at Forrester has called “the last mile” in process automation. In order to scale and get the most value out of an RPA implementation, Attended RPA needs to exist in some capacity.

Attended RPA provides a single pane of glass for business users to carry out their daily workflows with RPA bots at their fingertips. Automation Anywhere has called it the next frontier of process automation and has created a nice visual that breaks it down:

Attended vs Unattended RPA

Source: Automation Anywhere

Most case studies on Attended RPA point to call centers and help desks as the most common use cases, but we’ve found the benefits of Attended RPA extend far beyond those examples. Any task that an organization performs at scale – thousands of times a year to millions of times a year – can benefit from Attended RPA. The most common examples we’ve seen are:

  1. Sales Order Entry – Attended RPA allows sales reps and CSRs to quickly enter orders and ensures better data accuracy by defaulting data fields behind the scenes and automating downstream tasks.
  2. Customer Invoice Reconciliation – Attended RPA brings together all of the upstream and downstream data needed to allow finance teams to quickly resolve issues with customer invoices.
  3. Purchase Order Entry – Attended RPA can completely automate PO entry through e-commerce integration and also restrict the vendors that employees buy from, dramatically cutting down the amount of time it takes to buy while also ensuring complete data integrity.
  4. Vendor Invoice Reconciliation – Attended RPA can automate many of the rules and tribal knowledge used to match vendors invoices to purchase orders to cut processing time down on one of the most expensive business processes at any organization.

The sheer transactional volume of these uses cases makes the ROI obvious. Cutting the execution time down and reducing training can saves tens of millions of dollars a year, or even hundreds of millions of dollars, depending upon the size of the organization.


In order for RPA to scale, it needs to be paired with BPM to enable Attended RPA. As part of the roadmap to Attended RPA, organizations also need to consider that a fair amount of business process re-engineering will be required, which may require assistance from a consulting firm. These initiatives need to be carefully planned and need to have buy-in from executive leadership to be successful.