BUSINESS PROCESS

RPA makes waves Contracts make a difference Xchanging makes it work Morocco makes its presence felt

REPORT

01 cover supp.indd 7 26/11/2015 17:26:06 RPA Poppy appeal: RPA at BPO service providers

Just how beneficial can RPA be in a BPO environment? Leslie Willcocks and Mary Lacity investigate Xchanging’s introduction of robots

changing is a provider of technol- BPO service provider Capita was seeking Donaldson was a Six Sigma black belt ogy-enabled business processing, to acquire Xchanging with a recommend- himself, so a suitable champion for the X technology and procurement ed cash offer of £412m. project. RPA also seemed to fit well services internationally to customers with Xchanging’s core values, including across many industry sectors. Listed on Xchanging’s RPA journey customer focus, innovation, speed and ef- the London Stock Exchange, at the end of With Xchanging’s stress on “technology ficiency, and people empowered to make 2014 it had more than 7,400 employees at our core”1 and with his own deep a difference through teamwork. Further- (4,600 business process services, 2,000 experience in technology services, for more, RPA matched with Xchanging’s technology, 800 procurement) in 15 Adrian Guttridge, executive director at offerings of innovation and technology, but countries, providing services to customers Xchanging Insurance, the step into RPA also with the promise of new, valuable ex- globally. Net revenue for 2014 was seemed obvious, but prototyping was pertise working for the customer, together £406.8m, of which £282.4m was from necessary. In early 2014 he placed his data with added service delivery flexibility. business process services. As of mid- and information manager, Paul Donaldson, There was another prize. If effective, RPA 2015, Xchanging had automated 14 core in charge of an RPA project to identify and could also be exploited beyond insurance, processes, deployed 27 robots that pro- automate 10 processes in the insurance thus tapping into Xchanging’s relatively cessed 120,000 transactions per month, business while establishing a long-term new Group-based focus: “Our deliverable for an average total cost savings of 30%. governance and support competency wasn’t only towards processes, but to put Besides cost savings, Xchanging reported for the Group: “We did not choose an a framework in place that could be lever- many other business benefits including IT person, and it had to be someone aged for the Group – to institutionalise it.”3 improved service quality, faster service de- who understood process reengineering. livery and scalability. By investing early and Though I have an IT group of more than Start out and launch, 2013-14 heavily in RPA, Xchanging became well 100 people onshore and offshore many Xchanging has a huge amount of back positioned as a service provider to use hundreds more, I put it into the business office, high-volume, repetitive data collec- its RPA capabilities strategically to attract processing part under the operations tion and processing tasks, many of them more clients. By November 2015, major director.”2 still manual and many still taking data from

12 Professional Outsourcing Special Report

12-19 Willcocks and Lacity RPO.indd 12 26/11/2015 21:30:23 RPA matched with Xchanging’s offerings of innovation and technology, but also “ with the promise of new, valuable expertise working for the customer, together with added service delivery flexibility.”

12-19 Willcocks and Lacity RPO.indd 13 26/11/2015 21:30:27 RPA

t non-integrated legacy mainframe sys- process, the RPA business case will natu- “run” function of two people took changes tems. Moreover, information is extracted rally follow; the economics will usually add into business operations. From Group from various sources e.g. Excel, Access, up. It’s not going to cost you long-term to technology there was a dedicated systems PDF, and input into another system or deploy.” manager and two support staff, responsi- used to generate reports. There is a lot ble for servers, architecture and technolo- of manual comparison of information Bringing IT on board gy policy. Project management staff were across different screens in a system before The business case was approved in early also involved, along with Paul Donaldson netting and closing a transaction can occur. 2014 and securing the resources and as project lead. Up to August 2014, when Entry into a target system has to be based mobilising the project became the main the project went live, Xchanging also on certain business rules. tasks to March 2014. At this stage one of utilised the RPA provider to educate and Blue Prism RPA products seemed emi- the problems was the relationship with IT; support, the plan being to build the in- nently suitable for addressing these issues this was resolved by organising RPA as a house capability and become self-sufficient and achieving efficiencies from moving technology project with a business driver, as quickly as possible. data from any source system to one or being done for, and sitting in, the business. more destination systems. In particular, a Technology was responsible for delivering Building the robotic operating number of claims presented to Xchanging the underpinning infrastructure and archi- model seemed attractive. Robotic FTEs could be tecture. This was translated into how the Xchanging gained implementation speed one-third the price of offshored FTEs and project members were assembled and from selectively applying the provider’s could work 24/7 without errors. It took organised. robotic operating model and enterprise only several weeks to automate, with no The RPA initiative had 20 people maturity model, which represented struc- need for IT specialists. Superusers in oper- involved at various times, working under tured accumulated learning from previous ations could train the robots. The robots the head of operational change – 10 corporate deployments at, for example, do repetitive clerical tasks and fit into ex- from the insurance business and 10 from Barclays Bank, Shop Direct, the NHS and isting operations. Working in a virtualised Group technology. Initially, four were Telefonica/O2. The model is particularly environment off a secure, audited and developed by Blue Prism to perform the strong on building long-term, resilient RPA managed platform, the robots would run key role of process modeller – basically capability, according to Blue Prism’s Patrick in a virtualised environment and so could trainers of the software and system, and Geary: “You have to plan for where this is be scaled up and down rapidly while owners of the change activity. A separate going to be, not where it is now. You have working in any jurisdiction. to build a foundation for a tower block, not a bungalow.” Identifying processes for automa- Xchanging drew selectively on this op- tion erating model and Blue Prism’s advisory, The RPA software seemed most suitable operating and training resources, together where degree of process standardisation, One of the problems with Xchanging’s own extant process/ transaction volumes, rules-based process was the relationship technology knowledge and resources to and process maturity were all high. with IT; this was create a strong development and imple- Ten candidate processes were chosen. “ mentation roadmap and team. Xchanging found it challenging to identify resolved by organising what “high” meant and made some initial RPA as a technology Training the staff and the robots miscalculations. A learning point was that project with a During May-July 2014, the RPA group you had to identify suitable processes: RPA focused on technology design and build, fitted more with high-volume, low-com- business including the architecture, and server and plexity work. Paul Donaldson explained driver. software support, as well as training the

his learning point: “If you define the right process modellers and all staff. A key part t 14 Professional Outsourcing Special Report ”

12-19 Willcocks and Lacity RPO.indd 14 26/11/2015 21:30:27 Original Aviso USS Professional Outsourcing Magazine.pdf 1 23/11/15 11:20 RPA

t was designing and testing processes to Poppy decides whether to complete or 3. Operations staff did not fear robotisa- get the most out of the robots and making make an exception of the request. While tion, but named and welcomed Poppy components efficient, easily maintain- Poppy creates the validated LPANs, as a team member and, indeed, asked if able and reusable. By August 2014, humans check the exceptions. In this Poppy could be trained to do more work preparations were sufficiently advanced scenario, note that Poppy has to be for them. to launch four automated processes using trained correctly to carry out its tasks. 4. The robot can outperform a human 10 robots. A notable feature, unusual in Continuous improvement is also part on quality, speed and error-rate metrics, other implementations we have seen, was of the work, with first-time completion but can only work at the pace the overall Xchanging’s own introduction of manage- reaching 93% by May 2015. Where a process allows it to work at. ment information reports underpinning 500 LPAN process previously took days, The launch also saw an intensification the operation of the four processes. a properly trained robot can now do this of the internal messaging process, with in around 30 minutes, without error. The many roadshows in the UK and India. Automating example: London robot can easily scale up and down to Donaldson recognised that people would premium advice notes meet changing workloads, without human see RPA as a threat, but Xchanging was Figure 1’s example of London premium resource issues, e.g. staff availability, train- never expecting to lose anyone from the advice notes (LPANs) aids understanding ing, overtime cost. business through redundancies, having here. One new robotised process was seen automation coming and planned the validation and creation of LPANs, Initial learnings around it. The messaging was that RPA which insurance brokers use to submit Xchanging learned four things from auto- gave people the opportunity to move premiums to Xchanging for processing. mating this process: on to other, more interesting work. The Once an LPAN is created it needs to be 1. Continuous improvement beyond roadshows gave evidence of people taking uploaded to the central Insurers’ Market deployment maximises benefits. The ratio on new, expanded roles. Repository. The original process involved between robot and human process times While there is a big debate around the customer sending Xchanging an increases significantly when no web-based whether automation would see the unstructured data file. The file has to applications are involved. repatriation of work from offshore sites, be opened and validated. The operator 2. High-volume repetitive tasks are better Xchanging argued in its internal messaging then has to collect additional data from a performed by robots, not least due to that in its case there was no strong system called “Account Enquiry”. Next, removal of human error. rationale for this. Xchanging had no quality the LPAN is manually created and, with problem with its many offshore sites, supporting documentation, uploaded to which spanned work in business process- the Insurers’ Market Repository. This is a ing services, technology and procurement. high-volume process that the operators Nor would repatriation produce a signif- did not really like doing, but Xchanging is icant cost differential; offshore processing contracted to do it. Enter Poppy, a robot was already highly efficient. Rather, in Enter Poppy, a robot named by Xchang- named by Xchanging’s practice, automation could be applied in ing’s PAS technician Amanda Barnes after PAS technician those offshore sites to further improve Remembrance Day 2014: the day the “ performance where needed, for example Amanda Barnes after RPA process went live. In the automated in speed, and may well mean new job process, brokers still submit premiums Remembrance Day opportunities. and the human role is to structure this 2014, the day the data into a standardised template and process The journey 2014-15 – ramp up, hand it over to a pre-scheduled Poppy, went live. improve and beyond which reads the request. After various In the ramp-up period from September checks previously done by a human, to December 2014, the RPA team auto- 16 Professional Outsourcing Special Report ”

12-19 Willcocks and Lacity RPO.indd 16 26/11/2015 21:30:36 Figure 1 London premium advice notices (LPANs): before RPA (left) and after RPA (right)

INSURANCE BROKERS INSURANCE BROKERS

Submit LPAN Submit LPAN

Structure submissions Structure submissions

Humans find Deal with RPA finds Humans deal errors exceptions errors with exceptions

Humans add more RPA adds more data data from Systems from Systems of record of record

Humans create LPAN RPA creates LPAN

Humans add to RPA adds to Insurance Market Insurance Market Repository Repository

mated a further six processes. Advanced Paul Donalson: “Platform disaster recov- first quarter of the year. RPA had also be- training for all the operators took place, ery is in, with every robot having a robot come part of Group operations, reflected and volume ramp-up occurred across all friend sitting in another site somewhere; in Paul Donaldson’s appointment as group the ten processes. Says Paul Donaldson: it’s as simple as that. We’ve got an exact product manager for robotic automation. “Since really ramping up, we started to copy in a separate site. When we were a Meanwhile, the RPA initiative was being upskill our people even more and we’ve certain size it didn’t make sense to divide extended into the procurement division of really started to escalate volume. We’re the ‘run’ and ‘change’ functions, but we’re the Xchanging business. The F&A financial working about 70,000 cases per month now growing to a stage where it makes services area was also pushing hard to using our robotic workforce. We leave to sense. ‘Run’ is now an India-based team. implement RPA. human interaction about 7% of the pro- The continuous improvement thing has cesses we’ve automated, mainly business been quite major for us. We did a whole Lessons learned exceptions, or things process users don’t raft of changes at the start of the year and According to Everest Group, four key want us to do. System exceptions are we’ve just got more benefits out of the factors are driving the need for more incredibly low, usually down-time in the process that we’d never planned to at cost-effective operations in the insurance application or an unexpected behaviour.” the start by really tweaking in a controlled market.4 Macroeconomic pressures By May 2015 the automated processes manner.” include low interest rates, low GDP were achieving a success rate of 93% growth and high unemployment ratios. against the original 80% target. That came Automating example: e-policy Meanwhile, post-2008 regulatory changes from all the continuous improvement. By June 2015 the RPA team was doing globally have been causing massive During 2015 Xchanging started to insti- a lot more work in insurance, looking to upheaval in the insurance sector. Thirdly,

tutionalise its RPA capability. According to double what it had already achieved in the rising fraud incidents are increasing t

Professional Outsourcing Special Report 17

12-19 Willcocks and Lacity RPO.indd 17 26/11/2015 21:30:36 RPA

t the cost of operations for insurers. analysis, eight lessons for other companies Fourthly, as the modern consumer moves considering automation: towards digital experience, insurers need l RPA needs a sponsor, a project champi- to respond, this being translated into on and piloting; increased pressure for more efficient and Xchanging could l A culture of business innovation and cost-effective operations. see from previous technology accelerates adoption; l Business process service providers like implementations RPA should sit in, certainly feel owned Xchanging need to respond to these “ by, the business; combined pressures. At one level of of the Blue Prisms l Standardise and stabilise processes analysis, Xchanging had gone as far as it software that the before automation; could with existing methodologies and operational l RPA must comply with the technology technologies. The company’s continuous function’s governance and architecture improvement capability was still strong, payoffs were policies; but a lot of effort was needed for marginal considerable. l Build internal RPA capability to evolve, improvement, while offshoring further leverage scale and increase business value; work would weaken both the onshore l Multi-skill the robots; presence and also Xchanging’s image in In our analysis, Xchanging did not have l Pay careful attention to internal com- the London insurance market. Meanwhile, to face the major barriers to deployment munications. RPA fitted extremely well with Xchang- often encountered. Drawing on her This case study is a much shortened version ing’s three-year 2011-14 transformation Everest Group work, Sarah” Burnett has of a chapter in the authors’ forthcoming (offering a further improvement lever) and outlined these as legacy systems and book, Service Automation: Robots and The its strategic positioning against major shifts service delivery, lack of adequate process Future of Work, published by SB Publishing in the business process services market. documentation, lack of knowledge and/ in February 2016. Contact: l.p.willcocks@ It also fitted well with its innovation and or buy-in, employment sensitivities and lse.ac.uk or [email protected] technology capability and messaging to service provider hesitation (to protect demanding customers desiring those very their more traditional FTE-based pricing Endnotes; Technology At Our Core is the attributes. Additionally, Xchanging could models). title of Xchanging’s 2014 Annual Report. see, from previous implementations of the Xchanging has gained multiple benefits 1. Interview with Adrian Guttridge, execu- Blue Prism software, that the operational from its RPA deployment so far. Cost tive director, Xchanging Insurance, 18 May payoffs were considerable if it took the savings run from 11-30% depending on 2015. accumulated learnings, married them to the process being automated. Meanwhile, 2. Interview with Paul Donaldson, Xchang- their relevant in-house capabilities and there is significant and still improving ing, group product manager for robotic built an RPA capability for the organisation service delivery in terms of quality and automation, 22 April 2015. as a whole, for the long-term. Interesting- speed, and better ability to manage 3. See Everest Group (2015). Service ly, while RPA is sold most often on large in terms of governance, security and Delivery Automation: The Business Case For savings on FTE costs, this did not emerge business continuity. Processing activity has RPA in Insurance Services. Market report, in Xchanging’s case as a primary driver. become more flexible, scalable up, down March 2015. Xchanging seemed to have a mature and across activities, within wide ranges. 4. Sarah Burnett, presentation at the Ev- awareness of the multiple, even strategic Starting in the insurance business, Xchang- erest Group Webinar “Service Delivery Auto- payoffs that were possible and, in our ing is also extending RPA to Group-wide mation: The Next ‘Big Thing’”, 26 February view, this gave RPA adoption dimensions adoption and planning strategic, compet- 2015. The barriers are detailed in Everest of innovation, learning, and organisational itive edge payoffs. Xchanging provides Group (2015). Service Delivery Automation: acceptance – lacking in less successful rich learning; as a pioneer of RPA with The Business Case For RPA in Insurance Ser- cases. remarkable results, its case offers, on our vices. Market report, March 2015.

18 Professional Outsourcing Special Report

12-19 Willcocks and Lacity RPO.indd 18 26/11/2015 21:30:37 A lot of effort was needed for marginal improvement, while offshoring further “ work would weaken both the onshore presence and also Xchanging’s image. ”

12-19 Willcocks and Lacity RPO.indd 19 26/11/2015 21:30:40