AI 101: an introduction to automation and artificial intelligence in outsourcing
The traditional view of outsourcing has tended to see cost reduction as one of the primary drivers for any customer. The idea that the ‘total cost of ownership’ of a particular business function over the term of the outsourcing contract should be lower is very often part of the business case. Similarly, seeing outsourcing as a means of transforming a collection of assets on the balance sheet into a recurring service charge, and reducing (or at least apparently reducing) capital costs is another common refrain at the outset of deals. Whilst technology transformation / business change deals do happen where the aim is for a service upgrade at an increased overall cost, they are by no means as common as cost-led deals.
To date, a large portion of the cost savings delivered through outsourcing deals – especially those that involve any sort of offshoring, nearshoring or even (in the London-centric UK at least) ‘northshoring’ – come from labour arbitrage: the central idea being that the activity undertaken by the outsourcing customer’s relatively more expensive current local employees can be performed at a lower cost, but to the materially the same or even a better overall standard, by the outsourcing vendor’s resources. Ideally, the quality of the service is promoted via a tight service description, process maps / standard operating procedure manuals and a set of measured service levels backed by service credits and other contractual remedies for the customer if the service does not meet the required standard.
The continuing growth of outsourcing deal value, and its expansion from the domain of relatively commoditised non-core business functions into that of highly skilled business critical functions demonstrates the success of this formula. Anything that Harry can do from a desk in Basingstoke, Harun can do from a desk in Bangalore for a lower overall cost.
Look, no hands…
Whilst the labour cost may be lower, there are other significant costs associated with outsourcing deals. The most obvious new aspect is that the vendor needs to recover its overheads and make some margin, so the customer is always paying for that. Particularly where offshore resource is used, the cost of customer site visits to the vendor’s offshore facilities and the cost of compliance checks / processes (data protection, anti-money laundering, sector-specific regulatory regimes such as those in financial services or insurance etc.) add cost burden back into the equation. Add to that the cumulative impact over time of the much steeper inflation of salaries in traditional offshoring locations, and the day-one cost savings may have evaporated by the end of the expected term.
Conversely, technology costs everywhere continue to drop. Whilst Moore’s law may be slowing slightly as we approach the physical limits on how small a microprocessor feature size can be manufactured using current silicon photolithography techniques, the cost of computing continues to reduce dramatically year-on-year. And these costs are essentially equivalent everywhere – emerging economies enjoy no cost advantage when it comes to buying computing devices. Whereas once the cost of the computer on the desk may have been several multiples of the operator’s salary, that is no longer true even in lower-cost offshoring locations.
If you could get the computer on its own to do the job that you currently have a computer operated by a human doing, you can dramatically reduce the cost. What is more, you can then locate the computer anywhere in the world for an essentially equivalent cost. Moving services from the remote offshore locations back to being near to the customer’s other operations then reduces the travel, audit and compliance costs significantly.
The magic box
Speak to any large business process outsourcing vendor, and they will already be offering a number of automated product lines. For the most part these systems are not yet (in winter 2015/16) replacing human operators in significant numbers. And for vendors based in locations with vast and still relatively cheap labour pools, the incentives in this regard are not straightforward – this type of change in service delivery erodes the very advantages of labour cost that have allowed these vendors to enjoy such rapid and profitable growth in the last two decades.
But change is coming. Already genuinely smart systems are being demonstrated by a variety of different players. IBM famously won the US TV game show ‘Jeopardy’ with its Watson artificially intelligent system, and is deploying that same technology in the field of diagnostic medicine. The ramifications for the deployment of that technology in other ‘diagnostic’ fields, such as technology support and application maintenance are reasonably easy to see. Elsewhere, in fields as diverse as insurance claims handling and call centre management ‘Business Process Automation’ as opposed to outsourcing is a new term of art.
The key difference is the addition of some form of ‘artificial intelligence’, ‘machine learning’ or other ‘smart’ technology into the mix. The latest automated systems are self-improving and capable of increasingly sophisticated analysis and decision-making based on that analysis. This is not simple if-this-then-that automation; it is something capable of replacing a role previously filled by a person. And as the systems become more complex and capable of dealing with a more sophisticated set of inputs to make more sophisticated decisions, the roles replaced will go from the lowest-level jobs to the most senior roles.
The evolution from a human-led solution to a machine-led solution will not be an all-or-nothing change. We will see the service offerings of outsourcing vendors gradually include more and more automated elements over the next few years, and slowly approach a point where there are very few humans required to deliver a service that would be fairly labour intensive today.
Alongside that evolution, the relative importance of different aspects of the typical outsourcing contract will change. On the machine side, the importance of the scope of software licences, access to data, and access to systems and/or derived logic in an exit scenario take on a much deeper importance. On the human side, a smaller population of more critically skilled staff make retention and controls via key personnel and non-solicitation provisions more important. Similarly, the impact of (already very important) systems availability service levels are likely to increase, but will almost certainly be easier to mitigate as hardware cost reductions make the cost of running active-active systems to add resilience and redundancy earlier.
Other aspects of the contracts might move more toward being more akin to boilerplate. The attention currently paid to offshore compliance risks – especially around data protection type risks – diminish if systems can be located in the customer’s home jurisdiction.
Other risks need to be considered that are rarely part of the conversation during outsourcing negotiations at present. As systems become smarter, and more decision-making is automated based on large input datasets, the systems themselves might develop decision-making logic on grounds that are legally problematic. For example, anyone operating a system that treats people with particular surnames differently based on patterns it detects within the input dataset might be no less guilty of discrimination on racial grounds if the surnames happen to accord with a particular ethnic group than if that differential decision-making were based on simple prejudice. Responsibility for checking the decision-making logic and ensuring that it does not breach the law – in this example by discriminating on the basis of protected characteristics – needs to be addressed.
Last person standing
Technology change always seems to happen more quickly that we expect. The impact of smart automation on outsourcing deals will no doubt follow a similar path. The pace of change may mean that five-year deals written today will be replaced with essentially majority automated processes upon their renewal. Thinking about this now – especially with regard to employee transfer or redundancy issues – will no doubt stand both customers and suppliers in good stead.
About the Author
Gareth Stokes is a Partner at DLA Piper, focussing on information technology, outsourcing and intellectual property-driven contracts. Gareth’s experience spans numerous complex major procurement and sourcing projects within heavily regulated industries. Most such engagements involve elements of service delivery on-shore, with other elements delivered further afield, whether nearshore within the newer EU member states or offshore in India, China and emerging economies. These projects tend to involve innovative contractual structures, including joint venture arrangements, multi-level framework and call off structures, trust arrangements and other multi-contract arrangements.