Predicting RPA by Identifying the Challenges to be Overcome
- May 25, 2020
- 0 Comments
Predictions are tough. Even when they are predictions about what appears to be the near-term or the next year. Because you not only have to get the substance of the prediction right, you have to get the timing at least somewhat correct. There’s a great saying (Allworth?) that being too early is the same as being wrong.
Back in 2017, Neil Cybart predicted “Apple Glasses are Inevitable” – it’s a great piece on why they are likely inevitable, assuming trends in batteries, displays, and miniaturization hold up. Also, he gave a long runway for this prediction, which was smart – and lots of leeway even on that. If his timing was right on the money, Apple Glasses would come out sometime between 2020 and 2022.
Historically, Apple has launched a major new product category every few years. While consensus thinks the gap between when Apple enters new product categories is something like three years, it is more likely five to seven years. However, there isn’t a large sample size to allow us to place much confidence in any particular pattern. Since the Apple Watch was unveiled in 2015, Apple still has some time before the inevitable pressure arises for the company to launch a new product category.
Not to be outdone, here’s another interesting take from 2017: “Bicycles are eating the lunch of China’s dominant car-sharing app“:
Didi Chuxing last year beat the car-sharing business model’s pioneer Uber Technologies at its own game, and bought out the New York company’s China business to become the dominant app for hailing taxis or sharing a ride in the world’s most populous country.
Less than a year on, Didi’s dominance is being challenged by an unlikely source — about 40 smartphone apps that have sprouted in major Chinese cities since late 2016 for commuters to share bicycles.
The article cites bicycle sharing users topping 50 million by the end of 2017. But since then, while bicycles did make a huge dent in the ride-sharing market, scooters came along and undermined the bicycle economics in many of the dense urban areas where bicycles made sense. There have been many stories about thousands upon thousands of bikes gathered in piles for disposal as those bike sharing startups in China have gone under and their inventory collected as waste. And then COVID-19 has come along and put a hole in the canoe of the scooter-sharing business model as well. May you live in interesting times, indeed!
In what seems like ancient times now for RPA, back in 2017, McKinsey Digital published a post titled “Burned by the bots. Why robotic automation is stumbling” – but it turns out the title is a bit misleading – it sounds like a prediction of bad things to come, but really it is a listing of the challenges and the approaches CIOs were taking to address those challenges:
A year or so ago, a lot of people around the world got very excited about this.[…]
Many companies, therefore, rushed to install bot armies, spinning up pilots to configure all sorts of processes and projecting large financial outcomes. [..]
However, in conversations with dozens of executives, it is clear that the first act in the ‘robotics evolution’ has not been a slam dunk for many, especially when companies try to scale the localized proofs of concept.
So not a slam dunk. But clearly, from 2017 to 2020, the world of RPA has exploded rather than stumbled. Perhaps RPA will stumble in 2020 or 2021, but that certainly wasn’t contemplated in the article in 2017, it was a more immediate issue.
Sometimes, the best way to predict what might come next is to just list out the challenges, and assume that the companies selling into that space will work hard to solve those… So what were the perceived stumbles in 2017 that prevented scaling RPA programs beyond the initial pilots?
- Structured data may be easy to retrieve, but not all the data we need is structured. This is a challenge that RPA vendors and image recognition partners have been particularly aggressive about taking on with not only OCR but also entity extraction and applying Machine Learning to the whole process. At BP3 we’ve built some really slick solutions for our clients over the last 3 years that leverage entity extraction technologies.
- What if the “standard process” has lots of variations? This is hard to handle natively inside RPA, but this is why BP3 takes a hybrid solution approach: we leverage decisioning technologies to help decide what bot should execute or what the next action is, and we leverage process orchestration technologies to coordinate exception processes and to keep the mainline process clear.
- Bots need maintenance and updating (security patches, upgrades, etc). This is still true, and one of the reasons we always make an enablement package available to our clients to help them manage this overhead. Mind you, this overhead is still less than the cost of managing a team of people through the same upgrades and patches, but it is a cost that needs to be managed.
- Changes upstream or downstream in the business can significantly delay bots in production or break them in production. This is definitely a challenge – but it is also a challenge with APIs in many organizations, because they treat APIs as interfaces of convenience, rather than contracts that can’t be broken without mutual agreement. I can’t tell you how many times we’ve written a solution against an API and then had a change to that API break the code at some point in the future. (We can have a discussion in the comments of the definition of an API! )
- The expectations of hard savings can be overstated. I think this is an issue with any rapidly growing sector of enterprise software.
The remedies to these challenges that are prescribed do stand the test of time pretty well:
- Treat the bot as a tool in the tool belt, rather than the only golden hammer you have. Leverage process, workflow, six-sigma, etc.
- Take an end-to-end view of the process and the outcome – you can start with the specific pain point, but then map from there to the end-to-end to make sure you understand the context.
- Treat employees as problem solvers, which can be culturally transforming – this is great advice but probably the hardest to follow is to enable your team to do what they can.
From observations in our own customer base, it feels like the RPA juggernaut continues to roll forward in various ways shapes and forms, and we’re still finding great opportunities to work with you on how to apply it to key parts of your business. We’re seeing, together, how much value RPA, in combination with other methods and technologies, can create. I’m more excited than ever about the future for what we can and will do with automation.