Getting bots right for inbound work and customer engagement
Software bots have leapt from a standstill to a full run in recent years. Last year, the robotic process automation (RPA) market grew by over 60% (Gartner), and chatbots are growing 28% year-on-year (Technavio). Both qualify as software bots, colloquially just called 'bots'. Definitions can become quite broad, particularly when you factor in automation tasks. Then bots can cover the gamut from basic RPA to highly sophisticated forms of artificial intelligence.
This has led to some confusion over what bots are and caused delays in implementing the right bot choices. Yet, as the figures show, bots are incredibly useful and definitely something that businesses must consider.
RPA's meteoric rise underscores this: RPA is a relatively simple technology that mimics human behaviour, so it can take less effort to train. As a consequence, many companies started adopting RPA to get quick results and hopefully shortcut their digital journeys. Yet RPA is just the tip of the iceberg, and if you can grasp the broader promises of bots, you can find a lot more uses for them.
Bots start with processes
"Whether it's a voice bot, an image bot, a mail bot, a chatbot or something else, it literally is a small piece of code that can augment a human or take over human process," explained Richard Firth, CEO of MIP Holdings, one of the pioneering developers of bot solutions. Their experiences have taught Firth and his team a few crucial things about bots – specifically that knowing if you need bots, and whether they will be successful, starts with your processes.
"The fundamental ingredient is getting back to the industrial engineering trend of really designing your work processes. But you don't have to design 100% of the work processes to start getting value. You can take a look at your work processes and say: 'Which is my least inefficient? Or what is something that's impacting my business?'"
Bots essentially automate processes, and the steps and complexity of processes will dictate how effectively the bot can deliver on expectations. This is a crucial point, as many bot projects fail or become very expensive. According to research from Forrester, the service-to-licence ratio of bots (what you pay for extra services outside of the licence to make software productive) is more than three-to-one in favour of services. That is very expensive, and can happen for several reasons.
Keeping it simple
Bots must have a high success rate to be useful. Even if your bot is right 90% of the time, that 10% could mean a lot of human intervention, so you still lose. Getting a bot to 95% or 100% accuracy, though, can be difficult.
It doesn't have to be – success depends on several factors, such as the state of your processes and how complicated different process steps are. Many people get very excited, jump in with both feet, and end up with complex bots in confusing environments that produce very little. With bots, you must start at the ground and work your way up.
The journey to successful bots starts with processes. You need documented processes that run on computers, even if they are entirely manual. Instead of throwing all the work at a bot, it's more sound to have a bot take over limited steps and then improve upwards. Or build many simple bots to handle different steps. This keeps a lid on complexity and enables re-usability.
A bad process leads to a bad bot – there's no way around that.
"The nuance of what it takes to introduce automated processes sometimes gets lost in the big picture of the output, because people are thinking so big-picture they actually lose the simplicity of what they could do," Firth added.
This lack of nuance also reveals why bots are confusing the market. Though they are often seen as mostly the same thing, there are major differences between bots and how they can be used. For example, a chatbot would require more training than RPA.
In one example, Firth cited, the bot didn't even need to read specific data – it just looked at the layout of an e-mail to determine if it's an invoice or something else. So one type of bot can do its job at a glance, while another needs significant tweaking to work right.
There is more to explore under the banner of bots. They can expose business processes to customers in very safe and convenient ways for both parties – a concept Firth calls 'business process socialisation'. There are also considerations around training specific bots, or how complex they should be.
But don't be scared away from bots because of claims around cost, complexity or training demands. Those are true, yet only tell part of the story. Bots are popular because they genuinely are game-changers in terms of digitisation and automation. Start small, and start with your processes, Firth advised: "The point of departure is putting very small proof of concept together, because volume also complicates the matter. Base these on mapped processes and take it one step at a time. A very simple bot can add massive value to a customer, if you do it right."