What is our primary use case?
We use unattended automation for the bulk of our contact center processes and our financial shared-service center. We don't have attended robots, although they're on our roadmap for the next three months.
We use it on-premise, currently, but we are moving, over the next two or three months, to Amazon AWS.
How has it helped my organization?
A good example of an unattended bot is for what we call the "three two two two error," It has saved us £450,000 almost overnight. We allow customers to book seats on our flights so that a family of four can all sit together. We charge our customers for that privilege and it should work fine. However, if the original aircraft assigned to that flight is replaced, as far as the customer's concerned the plane is there and everything seems fine, until they get on board. They find that their seat booking hasn't transferred across to the new aircraft. They end up sitting in various parts of the plane, which causes complaints. We end up issuing refunds and compensation to customers. By having an unattended robot move the seat booking from one flight to another, we saved, overnight, £450,000. It's a great story to be able to tell.
Another process we've automated in the contact center is called our "Disney Calendar." Disney manages all of its hotel rooms on a big spreadsheet, effectively. They send that out to every tour operator which then manually updates its systems to show which rooms are available and which ones are booked. We previously outsourced that to a company called WNS, where they had a team of 18 people processing the spreadsheet. They came in three times a week. But because of the volume of rooms and the amount of manual work, we were constantly behind. From our customers' perspective, we were selling them rooms for the holiday of a lifetime — so it's a big, expensive thing — that had already been booked elsewhere. Then we were having to manually call them back up and tell them that the room had been double-booked and that we could put them in this hotel or that room. We were starting off on the wrong foot and providing an awful customer experience. We automated that process.
Now, we have saved the cost of those 18 people that we were dependent on from an outsourcing partner. And processing the calendar, instead of taking three or four days to do one iteration of it, takes three hours and runs overnight. The customer experience is fantastic because everything is quick to market. Everything is absolutely correct because the robot is 100 percent accurate every time. We've seen an increase in our sales to Disney and a drop in complaints about Disney holidays, as a result of the automation going in.
The solution has improved employee productivity, but not directly. We've taken a process that a contact center agent, or somebody in the financial shared center, was doing, which was taking some 50 percent of their time. We've now freed them up to do other more interesting work, to be more productive and more innovative. We don't have any metrics that we can share to show what they've done with that time yet. We're not quite that mature yet. But if I were to approximate the increase in productivity in the contact center, I would say it's about 30 percent.
In terms of the solution improving customer experience by helping employees stay focused on the customer rather than on desktop complexities, that's not happening at the moment. We don't have attended automation. But by removing things like the Disney Calendar issue, where our contact center agents would be dealing with complaints, and the finance guys would be processing refunds and compensation, the impact of that automation has meant that those calls aren't coming in, so our agents are able to focus better on the customer experience than they would have previously been able to do.
What is most valuable?
NICE is one of the only vendors that does attended and unattended out-of-the-box. Using the unattended processes we've been able to build a "feature library." We break each process down into workable chunks that we can save into a big library. The next time we come to automate a task, we already have chunks of that automation built.
For example, logging into an application is what we call a feature. We build it and we put it into the library. The next time somebody comes along to automate a process that uses that application, the login feature is already there. That's one piece of work that they don't have to do. That feature-based build is something that's unique to NICE and we find it most valuable.
The API connectivity is strong, although we don't use it a lot. We tend to go with the screen connectivity. However, we use APIs to connect into SharePoint, and they work seamlessly.
What needs improvement?
The one thing I'd like to see, and NICE is already heavily investing in it, is improvement in the user interface itself. They call it the Designer and it's what the developers use. It is a bit clunky; that is the polite way to put it. I'd like to see it be a bit more user-friendly, a bit more intuitive, and to move to something a bit more web-based, so they could develop though the browser for Excel. I believe NICE is already heavily investing in that, and I'm looking forward to getting my hands on that to have a quick play.
For how long have I used the solution?
I've been using it since the beginning of 2017, so just under two years now.
What do I think about the stability of the solution?
It's robust. I couldn't fault it.
What do I think about the scalability of the solution?
The scalability is also very good, very robust. We, with our on-premise solution, are constrained by the size of our servers, which is why we're moving to Amazon WorkSpaces. That way we can scale as quickly as we possibly can, using the AWS toolset and NICE together.
With hindsight, if we were to go back two years, I would have built it in Amazon to start with. But we needed to get a proof of concept over the line and keep costs down. We learned through our mistakes over the past couple of years that we have reached a point where we have outgrown the infrastructure we're on. Because NICE is so scalable, we need to get onto infrastructure to support that capability.
How are customer service and technical support?
NICE technical support is fantastic. They're very responsive. They're very quick. They're really keen to learn if there's a problem that we've identified in the tool itself. They then put their research and development department on it to work with us, almost instantly, to address the problem so they can fix it. They're brilliant. I can't fault them.
Which solution did I use previously and why did I switch?
This is our first dabble into automation.
How was the initial setup?
The technical setup of the solution was actually really straightforward. The part that's more complex, is the way that we at Thomas Cook set up our projects around it. We have multiple competing projects all wanting to use a shared infrastructure in slightly different ways. So the NICE tool setup was probably the easiest part. It's more our management and governance at the start, where we were quite immature and didn't have much experience for automation and that's how the CoE was born in Thomas Cook, to overcome those hurdles and provide a better service.
There was a learning curve about how to manage business automation. Take IT, for example. The change management processes, which we have in place to protect the business from itself, don't fit with automation that can be turned around in a couple of weeks. Traditionally the processes have been more archaic. A change request form would be submitted and then, three weeks later, you'd sit and present it to the board, and then your change would be approved or disapproved. If it was approved, you would go ahead the following week. So that timeline for implementation, traditionally, was four or five weeks. It would take less than that to build an automation and make it live. We had challenges around those ingrained processes from day one. So the CoE was brought in to make sure that the people within IT, and out in the business, who were not confident or were worried that we'd be impacting business applications or processes adversely, were brought along on that journey and that helped them overcome those issues quite well.
Those challenges are still there. There are still a few. It's more a case of reassurance. We've hit our target and it's starting to prove to the business that automation it works, that it does what it says on the tin. It makes everyone's life easier. We're just creating automations that replicate what a human is doing rather than building something new.
For example, and here I come back to change management, they were really worried that if we put a robot onto an application that the application would suffer because the robot was going too fast. Bringing them on that journey and showing them that we can actually speed the robot up or slow it down, that we can put more robots on or fewer, and that helps to make sure that we're not breaking anything. It wasn't a way of working that they'd previously seen or agreed to. But by showing them it is that easy, that we could just turn the thing off if we started to see something break or started to feel there's a problem. By showing them and bringing them along with us, we managed to turn it. They have all bought into what we're doing.
To get the first robot live took a lot longer than it should have. Again, not because of the NICE toolset, but because of the way we structured our projects. It took us the best part of six months to get the first robot live.
At the moment, we have a pipeline of work that's coming in from all over the group, around the world. That includes a contact center, a finance shared service center, human resources, IT, e-commerce, and our yield department. Pretty much everyone's seen the success that we have delivered over the past year and are keen to come on board. They have each assigned someone to be a project manager for automation and to manage their own pipeline of work within their own transformation programs. Each of the automation managers send through their roadmap to me. We then assign a business analyst to work with them to track some metrics, to find out what the benefits will be from automating.
Once we've got all that information together, the business analyst goes through in fine detail to understand what that process actually is. The process is then translated it into a technical document that we hand over to our developers who then build the solution within the NICE toolset. We go through a very small change management process within the center of excellence to make sure the documentation is up to date, that it can be supported, and to make sure there's capacity available on the servers. Then we put the automation live and monitor it for two or three weeks, depending on what the business wants. Once they've signed off on it — after two or three weeks of it being successful — it moves into our run team which then looks after the automation and any ongoing improvements going forward.
It took us a while to get there, but it's quite robust and we've got a lot of checkpoints in there to make sure that all our stakeholders are fully up to date and that it's all signed off and everyone has approved and is happy.
What about the implementation team?
We used a company called PAteam. I think they are ex-NICE employees and they essentially showed us how to build automations. I have a team, internally now, of developers and BAs that the PAteam, and NICE, trained to show them how to do this stuff. They helped us turn it into factory so that we can repeat it. I thoroughly recommend PAteam to anybody. We wouldn't be where we are without them.
What was our ROI?
In terms of scaling and increasing the impact of automation, our center of excellence has been around for 11 months now. Our target for the year was a saving of £2 million and we've just hit £2.4 million. That's given us just over a two-to-one return on investment. By using that feature-based build that I mentioned earlier, where we take those component parts for login and logout, or copy and paste from this spreadsheet to that one, we are forecasting a five-to-one return on investment within the next 24 months.
We use a range of metrics for ROI. The team is a fixed cost, the centers of excellence are fixed costs. But we claim benefits in a variety of ways, such as the more traditional things like FTE savings, margin increases, cost avoidance, avoidance of brand damage, and risk-avoidance. Those are among the metrics that we use to build up that entire ROI. We see ROI on all those metrics.
What's my experience with pricing, setup cost, and licensing?
We just moved to an enterprise license agreement. It's an all-you-can-eat license, so we don't have to count the licenses anymore. We have perpetual licenses with NICE and we pay an annual maintenance fee on them.
We've made a significant financial commitment to NICE. It shows that we are fully committed to NICE and to continuing the growth of automation within the group.
Which other solutions did I evaluate?
We looked at some of the other vendors, including UiPath, Blue Prism and Automation Anywhere. There were a few reasons that we took on the NICE product.
First, they're the only vendor that supplies attended and unattended automation out-of-the-box, which was a big tick in the box for us. Their levels of support were fantastic, and the level of training that they provided was what we needed. They were aware that we were completely new to this two years ago and they really heavily invested in us, providing onsite training, remote courses, and remote support for the developers. They couldn't have been better as far as the support package goes. Those were the key reasons we chose them.
What other advice do I have?
Listen to people who have the experience, guys that have gone through it; guys like us at Thomas Cook who didn't know on day one how to do this, at all. Also, do research online, read all the horror stories of things that have gone wrong and believe them. Take your time at the beginning and make sure you do a proof of concept on something small or a series of small initiatives, rather than going in for the big fish on day one. Find your footings, get the foundations there, and then build on that. Think big but start small and scale fast.
In terms of lessons learned, it's been such a big journey, at such a fast pace of learning, that almost everything is a big lesson. The biggest thing we've learned from the mistakes we made back on day one was that we picked the wrong processes. If I were to go back and speak to myself in 2017, when this all started in Thomas Cook, I wouldn't go for the biggest, most complex process. I would take smaller, more intricate ones as a proof of concept to show that it works, and then scale up quickly. We did the reverse. We went for the benefits of automating huge processes and then went straight into trying to tackle them. That's why it took six months to get the first one over the line.
We haven't used the NICE Employee Virtual Assistant to help our employees with onscreen guidance or in-context recommendations. It's on our roadmap for the first quarter of the next financial year. Similarly, regarding automating employees' routine processes, it is on the roadmap for the first quarter as we move into attended solutions. It will include things like password resets and the like. We're using the Automation Finder to identify processes best suited for automation, but only as a proof of concept. It will be running for the next couple of weeks. We haven't had the results yet, but we're excited find out what we don't know.
Since we don't have any attended users, there's nobody within the business itself using the toolset. But we have a team of 11 that includes me. I have a gentleman called Jit Patel who heads up our RPA development group, and Kay Ellis who looks after a team of business analysts. Beneath them, we have four developers reporting directly to Jit, and four BAs reporting directly to Kay. I manage the center of excellence.
Deployment requires a BA and a developer working with the business unit itself. But as far as the NICE toolset and the automation side of things go, it's actually very quick for the business analyst. That person makes sure we've got the process nailed down and then translated into a technical document so that a process of, say, 50 business steps is actually 12 features. They work on building those features in the document that the developer then brings to life.
As for increased usage of NICE, we started off with just two business areas in mind, the UK contact center and our financial shared service center which is in Palma, Mallorca. At the point that we got the first two robots live, we realized that we needed to build an internal function to manage these things. That's when the center of excellence was formed, and that's allowed us to then move out to lots of other areas. We have automations live for human resources, IT, costing and yield areas, finance contact center, and we have plans over the next six months to move across into our other source markets. That includes Germany, the Nordics, and everywhere else that we essentially have a Thomas Cook footprint. We're starting to expand out into a group for attended solutions. When we move to the Amazon infrastructure in AWS, we'll look to deploy attended automation from that central infrastructure as well. We have big plans to expand.
Overall, I would give NICE a solid ten out of ten. I can't fault them. It has been fantastic.