Initial RPA engagements are like dating.  Everyone is on their best behavior, and very eager for a successful outcome.  Expectations are high, experience is low.  Are there topics of conversation that are particularly sensitive?  Although we both speak the same language, we may not share a common vocabulary about important matters. Worse, we may use the same words to describe different concepts.

But from what I remember from my own dating life, we first would meet for coffee, followed by a few get-togethers involving dinner or a movie. Only after assurance of mutual interest would we foray into more lengthy excursions.

What I’m seeing in many initial RPA pilots these days, however, is akin to planning a month-long trip to Paris for a first date. There are way too many moving parts, expectations are incredibly high, and it’s impossible to live up to any reasonable definition of success.

There are many reasons for this. Service providers are eager for big engagements, for example, and clients are excited to get moving, particularly to meet leadership expectations for efficiency, quality or savings targets.  Tools vendors aren’t helping either, with their compulsion to sign customers to big engagements in order to convince leadership about the speed and ease of implementation that RPA can deliver.  I still hear salespeople making claims such as, “a good Excel person can build a working process in 4 weeks!” *Shudder*

Two Principles for Success

Whether a company is just starting out in RPA, or simply rolling out to a new line of business, here are two principles to ensure success by keeping expectations and ambitions in scope while also ensuring fast results:

  1. “Hello World”

The first robot to be deployed in any new line of business should be a “Hello World” bot.  By this, I mean a trivial robot that nonetheless goes through the entire development lifecycle.  A great example is a bot that downloads the company’s Web page and verifies that the company name is found somewhere on that page (and throws a business exception if not).  This bot takes five minutes to build and exhibits the same infrastructure requirements that often derail a larger scale bot, such as virtual machines, bot credentials, monitoring and so on.   By removing any cost-savings expectations for the initial projects, however, businesses can concentrate on solving these problems in a repeatable way, without the added pressure of attaining performance results, nor the attendant defensiveness and finger pointing when things do not go exactly as planned.

  1.  Six Weeks to Glory: Minimum Viable Automation

Just as in Agile development, where we place a limit on the size (in points) of any particular user story, businesses should not pursue any RPA project that cannot be at least partially brought to production within six weeks.  This doesn’t mean the full value must be realized, or that ROI is achieved in that timeframe.  In fact, the business might continue to add to the RPA implementation for years to increase its functionality.  However, the concept should be simple enough to produce something in a short time. 

Typically, this is accomplished by going live with exception handling and a small but meaningful subset of the transaction types that will eventually be supported, and adding transaction types with subsequent releases.   For example, imagine you are building an indexing bot that processes an e-mail inbox, and handles the requests that come in.  The first release of this bot might include the logic dealing with error conditions, processing the e-mail box, forwarding unrecognized messages to the human processors, and responding to some very simple requests, like password resets.  Subsequent releases might handle more and more complicated requests.  Only after a significant time in production will the most complicated messages be handled (perhaps aided by machine learning algorithms).

Starting Slow to Finish Fast

As with dating, a slow start equals fast benefits with initial RPA implementations.  The most successful RPA program I’m aware of started off with a trickle of tiny projects (reducing FTE by 0.25 and then 0.3), and grew to tens of millions of dollars of annualized savings in a matter of four months.

Are you hitting those kinds of numbers?  If not, it might be time to change your “dating approach” by invoking a little “Hello World” and “Six Weeks to Glory.”

Shaun Dawson

Shaun Dawson

Shaun Dawson runs Cognizant’s Global Center of Excellence for Intelligent Process Automation, which includes RPA. He has over 20 years of experience... Read more