Automation in Outsourcing: Are the Robots about to Eat our Lunch?

Kit Burden looks at the potential impact of robotics process automation and related technology on the outsourcing industry and highlights some aspects that should be getting the attention of tech lawyers operating in the outsourcing sphere.

Continuous improvement and innovation have long been buzz words in the context of outsourcing arrangements. At the time that the relevant deal is first done, the customer will usually be expecting improvements on the status quo (including but not always limited to cost) and the contract will need to reflect that in terms of commitments in the charging regime, service level agreement and so on. Equally, as such contracts will usually cover multi-year terms, the customer will often push for at least a soft commitment that further improvements will be forthcoming during the term of the contract.

However, in more recent deals, such commitments have often been concentrated upon a particular theme, namely robotics, or automation. This goes beyond the 'physical' robots which have already transformed the factory floors of manufacturing industry all around the world, but concentrates instead on the new 'proto-A.I' of application software which is able to learn on the job by analysing patterns, data and behaviours, and auto-adjust to operate faster and more efficiently in future. To quote Chetan Dube, the President and CEO of IPSoft (one of the leading producers of automation software):

'We are approaching a seminal moment….the basic platform for how services are delivered is shifting. Not replacing labour with cheaper labour, but replacing labour with cognitive systems'

So what does this mean for the outsourcing industry, bearing in mind the fact that labour arbitrage has been such a significant driver for the offshored element of outsourced services, in particular?

Suitability for Automation

Not every outsourcing project has the capacity to be impacted to any significant degree by the use of automation; for automation to be material, a number of characteristics would need to be present.

Digital Feeds

Automation ideally needs data in electronic format which can be fed into the processing applications on a constant or near constant basis. It follows that tasks which do not depend upon such digital feeds will be less susceptible to automation (including for example break-fix services for physical IT infrastructure). In the BPO world, therefore, there may be various types of transactions which depend upon data feeds coming in from other parts of the business in a standardised format, which can therefore be readily configured to feed into the automation software.

Rule Based Processing

We are not in the age of fully fledged artificial intelligence, and so the processing software will not be able to think for itself just yet. Efficiencies will therefore be best achieved where the underlying transactions follow a consistent set of rules which the software can reflect and apply (eg if X and Y are present, then do Z). The software will be more sophisticated in terms of then being able to combine information from multiple sources in order to determine whether X and Y are in fact present, and can potentially eliminate human involvement in so doing, but the underlying rule set still needs to be there.

Poorly Integrated Systems

It may be obvious, but if the customer's systems are already well engineered for straight through processing, then the potential for a further automation layer to be introduced as part of an outsourcing solution will be much diminished. Automation as a function of outsourcing therefore works best where there is some form of break in the chain in the customer's current process, and/or elements of human involvement where both the time and cost of such a link is then susceptible to being removed.

Processes Prone to Human Error

Even if there are well understood rules and requirements, humans sometimes make mistakes; figures can be transcribed incorrectly, decimal places missed, words misspelt, or the 'fat finger' syndrome can strike such that the wrong character on a keyboard is inadvertently pressed. The automated assessment of digital feeds removes the possibility of such errors, so removing the need for both rework and potential compensation for mistakes.

High Volumes

Human beings have a finite capacity for work; people will usually work a maximum number of hours in a given day (eg in shift patterns, as part of an outsourced operation), and can only process so much work in the course of a given hour, even if they may be able to improve this through repetition over a period of time. By contrast, software never sleeps, does not get tired, and does not need to go home at any particular time. It works at high speeds, 24/7.

In the BPO world in particular, therefore, there will be a wide range of services which will fit into the 'sweet spot' highlighted by the characteristics set out above; for example, invoice receivable and payable functions, benefits processing re HRO, supply chain management functions and SLA monitoring and reporting, to name but a few. Equally in the ITO world, automation software is playing an increasing role in terms of helpdesk functions and incident reporting and diagnosis.

Benefits of Automation

Some of the benefits of automation are obvious, others perhaps less so.

Cost is the immediate payback. If transactions can be processed with a lower involvement of human beings and with a lower incidence of need for rework or repetition, it follows that the outsource service provider will need fewer people to deliver the outsourced service, and that will be reflected in its underlying cost (from which the customer will then expect to benefit).

An automated solution should also be able to facilitate significant improvements in process times, as the (slower) human links in the overall process chain are removed; a KPMG study put the potential improvement in some cases as being as much as 80%.

Not only will the relevant transactions be completed faster, but overall, they should also be completed more accurately. As noted above, human beings are prone to making mistakes, especially in high-volume set-ups where there are significant quantities of data to be assessed and multiple choices to be made on the back of the information being received. Software does not make mistakes born out of lack of attention to detail or tiredness; the KPMG study referred to above claims that in certain processes the level of errors could be reduced by as much as 90%.

On top of these compelling benefits, an automated function will be better placed to capture data regarding the relevant service, even in real time, and present this data back to both the outsource service provider and ultimately the customer in a consolidated and user friendly manner, so providing improved analytics and M.I.

The Flipsides?

So far so good; it would appear from the above analysis that automation is a 'slam dunk' and will necessarily be a part of any outsourcing engagement where the criteria applicable for the implementation of automation are met. However, automation is not without its legal and contractual challenges.

Firstly, we need to identify how the implementation of the automated solution fits within the scope of the wider outsourced service. Will, for example, the outsource service provider simply incorporate it as part of the outsource service itself (ie as part and parcel of its service commitment, and with the resulting financial benefits already reflected in its contractual charges), or will this be something which it will simply help to integrate/implement, but on the basis that the customer retains the licence relationship with the licensor of the automation software? This is an important consideration, as the contract will need to reflect what continued use the customer may have of the automation software beyond the end of the initial term of the outsourced contract, any restrictions on the scope of its use (eg with other group companies or in other territories, etc), and who will have responsibility for the provision of support and maintenance services for the software.

There is then the question of who has responsibility for the operation of the software if it goes wrong. As noted above, the proponents of automated solutions will make much of the projected reductions in error rates when one compares automation with a comparable high volume processing service undertaken by human beings. However, whilst it is true that human beings can get tired and can make honest mistakes, they also have the ability to spot an error that they have made and to go back to correct it, or to stop themselves making it again. Whilst automated solutions may also claim to have this capacity to a degree, a mis-configuration or programming error in the solution in the first place can result in the software making the same mistake over and over, at high speeds and on a 24/7 basis. So for example if the solution was operating in relation to a payments function and triggered incorrect payments being made, instead of just one incorrect payment being made (as may have been the case with a manual solution), there could literally be hundreds if not thousands made before anyone realised what was going on.

As the licensor of the automated solution will not be enjoying the size of revenue which one might expect the outsource service provider to be receiving, it is highly unlikely that their licence agreement will include a liability clause which will allow much - if any - of this kind of loss to be passed back to it. If the customer has a direct licence agreement with the software licensor, therefore, it will likely face a sub-optimal level of recovery. If, on the other hand, the automation software has been licensed by the outsource service provider and simply forms part of its delivery model, it may find itself 'squeezed' by way of having a significant liability cap with its customer (set by reference to the value of the outsourcing deal as a whole), but with only a very limited ability to flow much if any of its exposure down to the licensor of the automation software.

The use of an automated solution may also have significant people impacts. Clearly one of the aims of implementing an automated solution is to do away with - or at the very least streamline - processes which are dependent on human labour. It follows that the implementation of such a solution would be more likely than not to result in potential redundancies, and the outsourcing agreement will accordingly need to cater for the assignment of financial liabilities associated with this. Moreover, the move may have unforeseen consequences in this regard as the parties may have worked on an assumption that TUPE would apply and that staff within the scope of the affected function would transfer across to the new service provider. But the reality may be that the nature of the service has changed so much by virtue of its transformation from a manual process to one which is fundamentally driven by software that there is in fact no 'undertaking' of the type that the customer was originally conducting being carried on any more at all. If so, there may then be no TUPE transfer, and the customer may instead be left with an unexpected challenge of having to find new roles for its affected personnel, or going through a potentially costly redundancy process.

Strategies for Automation

The approach to be taken to automation in outsourcing - at least from a legal and contractual perspective - will differ depending on whether one is looking at new projects, or existing ones.

If you are looking to incorporate automation into a new project, then the following questions will need to be addressed:

·         Has the service provider already 'baked in' the cost savings and efficiencies which are anticipated as flowing from the use of automation, ie so that the customer is seeing the benefit of lower pricing, improved service levels etc which are anticipated as flowing from the use of the automated software? It may be more beneficial for the customer to have the contract drafted in this manner so that the risk associated with implementation passes across to the service provider but, in such circumstances, the customer will want to check:

o   whether the improved pricing etc kicks in only when a certain milestone or implementation action has been completed (which will beg the question of what will happen if this is delayed)

o   whether it is happy that it is getting a fair 'share' of the financial benefit of the use of automation, or whether an undue proportion of it is being retained by the service provider.

·         Will the automated solution be licensed via the service provider as part of the overall solution, or by way of a separate licence engagement direct with the end licensor?

·         What are the terms for support and maintenance of the solution, and what guarantees are there that such support will continue to be available for the envisaged term of use of the software?

·         What escrow arrangements are in place, should there be a need to access the underlying source code 'in extremis' in the future?

If there is no new project under consideration, that is not the end of the story. Automation could - and perhaps should - have a significant impact across a range of currently outsourced services, provided that they possess some or all of the criteria to make automation potentially relevant, as summarised previously.

Many outsourcing agreements will contain a range of provisions which require the service provider to 'keep pace' with new technological and process developments. Some may be in the form of commitments to 'continuously improve' the services (irrespective of changes to the scope of the services as may be agreed via any form of change control process). Others may focus on the pricing aspects, for example by reserving for the client the right to appoint an independent third party to benchmark the service provider's charges against those of 'equivalent' contracts and services, which may then of course take into account the fact that those equivalent projects have deployed, in whole or in part, automation-enabled solutions. The contract then may (or may not!) require that the charges of the outsource service provider be automatically adjusted down to whatever benchmark comparison level the contract may have set.

Even in the absence of express contractual provisions of this type, the customer may still have some leverage. A contract extension is always attractive and of potential value to a service provider, for example, as may be the opportunity to add additional scope to an existing deal.

The options available to a customer who is suspicious as to whether their service provider is passing on to them the full benefit of potential automation might accordingly be summarised in the diagram below:

robotics effects on outsourcing

Where the customer ends up will depend upon the nature of the outsourced services, its bargaining leverage with the relevant service provider, and of course the contents of the contract terms it is able to unearth.


Robotics Process Automation (RPA) is likely to be a buzz phrase and an ever more prevalent part of the outsourcing lexicon in the months to come. Whilst not relevant or appropriate for every flavour of outsourcing project, it will plainly have a role to play for many types of ITO and BPO deal. The key for the lawyers engaged with such projects (either when first being set up, or upon subsequent review and renegotiation) will be to be aware of some of the risks and issues associated with the use of such software, and to ensure that the contract is drafted in such a way as to address them in future. 

Kit Burden is Head of Technology and Sourcing Group at DLA Piper and an SCL Fellow.

Published: 2016-02-11T13:29:23


    • Kit Great article. Builds on the outsourcing session you and I participated in at the ITechLaw London Conference last November when Ole Horsfeldt spoke on this subject. I suspect that we are only at the beginning of robotics automation and that it will have an increasingly profound impact on outsourcing.
      Clive Davies, 14:24:13 17/02/2016
    Please wait...