Offshore or home alone?

If conditioning data does not add any value to your analytics, then why not get it done offshore? David Reed finds out

Great ideas are often said to be the result of 1 per cent inspiration and 99 per cent perspiration. That is what Thomas Edison thought, and he knew a thing or two about genius.

If that is the case, can you then split those two elements up and get the hard work done separately, stepping it at the end to turn the functional into the breakthrough? That question is currently being hotly debated in the realm of customer insight and data analytics.

The high-level trend at work is for knowledge process outsourcing – moving activities that create intellectual property and added value (such as analytics) into external providers, especially those operating in low-cost environments offshore. Market analysis firm Datamonitor has identified this as a growth opportunity, provided companies are willing to relinquish direct control over core business functions.

“KPO represents the next stage in the evolution of the outsourcing market,” says Ed Thomas, analyst for business process outsourcing at Datamonitor and author of its recent report. “Unlike BPO, which refers to the transfer of mainly transactional, non-core processes to specialist providers, KPO involves the outsourcing of core business processes, for example planning and auditing, which require a high level of domain expertise.”

The main driver of KPO is, of course, cost. In the realm of customer insight, salaries for experienced analysts have been rising rapidly northwards of £50,000 per year. Add in the cost of software licences and ancillary support and a single-person analytics team could set you back six figures.

Alternatives are emerging which appear to offer all the convenience of a UK-based marketing services provider with the low cost-base of an offshored business. Cogent Analytics is one example of this new hybrid KPO solution.

“There is a two-fold driver for us,” says general manager Matt Hutchison. “We can be very competitive in terms of the pricing levels we can set by having a low-cost offshore base. We also offer the same skills and resources of a local operation.”

His company came into existence when its parent group acquired an operation in India, part of which was already running statistical models. The skills base to be found among Mumbai-based graduates is particularly strong, yet salary expectations are a long way behind those in the UK. “It is a particular feature of analytics that it is often difficult to find those skills here,” adds Hutchison.

For the Indian staff, the takeover has changed the type of work they do. “We have brought a whole new set of work to them. From an employee development point of view, that is very positive,” he says.

Part of the approach being used by Cogent is to ensure its offshore teams get exposure to the UK market. During a project for a health club chain, for example, it brought an analyst over for one month in order to get to grips with the issues and market. Conversely, Hutchison will visit the Indian office two or three times per year.

Strong local management is now in place to keep the operation focused, while communications technology such as VoIP and Citrix are a routine part of the interaction between the two locations. Of the things that tend to concern clients about offshoring knowledge processes, the two main ones are resilience and context.

“Events can happen anywhere. Our office is in Mumbai – it went offline for about two hours when the attacks happened. But we had a plan which involved people working from home. And we were in the middle of a big deliverable at the time,” says Hutchison.

But can an analyst working offshore and experienced in an entirely different culture really pull out a major business insight from a data model? Few argue that the data mining and modelling techniques are not the same wherever you are. What matters is the sense you make out of those numbers at the end.

Huw Davis, founder of Huw Davis Partnership and one of the pioneers of KPO for analytics, says that this is precisely the reason for having a dual location set-up. His business has its data engineering operation in Singapore and the client-facing team in the UK.

“We still want one or two people managing projects this end. It is like creative work in an agency – it is the creative director who takes the meetings,” says Davis. “In my experience of running UK teams, only the top 10 per cent of people go out to see clients.”

In other words, the hard work that happens behind the scenes could already have been done elsewhere (and often was, generally by freelancers). What the client needs to focus on is getting the brief right and leave the service provider to ensure they match outcomes to requirements. “That’s where our knowledge and experience is important,” he says.

For Davis, a bigger gain from using KPO is transferring the burden of getting data into the right condition before any profiling, modelling or analytics can be carried out. He notes that in many client organisations, their in-house insight manager may be carrying out a lot of workarounds to correct underlying data errors and omissions, based on their personal knowledge of the data set.

Part of the process his company has built is called Knowledge Discovery and Data which is designed to uncover where these issues exist and will need to be dealt with. “Part of the problem is that a lot of clients and agencies have a fear of completely letting go,” says Davis.

To address this, HDP ensures that all personally indentifiable information, such as name and address, is stripped out of the data. A unique reference number is applied and a fully-encrypted set of anonymised records is sent to Singapore for processing and analysis. Only after this step is the identity of the subject added back when the data returns to the UK.

This division of work is critical to the emerging KPO model for insight and analytics. It brings to light one of the often overlooked issues with data – that at least a third of the time spent on any project will be just for conditioning data. This is not value-adding and can be done by anybody with a good understanding of what data should look like.

Given a trained mathematician or statistician, briefing them about the correct condition for that data is easy. So why not locate that individual offshore where their cost base will be much lower than in the UK?

“It has got to be an option for everybody providing the services we do,” agrees Gareth Mitchell-Jones, director of analytics at Experian Integrated Marketing, “particularly for bigger companies who may be getting pressure on margins and are looking for cost-savings.”

He notes that the initial set-up costs for an offshore operation may be considerable and will therefore deter all but the largest suppliers from entering the KPO market. That gives a global company like Experian a built-in advantage if clients prove willing to work in this way.

“We take advantage of ‘dark hours’ services to work around the clock for clients who want faster turnaround,” he says. Experian has a set of internal standards and processes that all of its offices around the world work to, making the transfer of projects across time zones easier.

“The brief is critical. We need to make sure it is detailed and there have to be clear arrangements for workflow to hand over at the end of each day,” says Mitchell-Jones. At the moment, however, he says that this is more of an opportunity than a reality, with only a small number of clients wanting to take advantage of this global resource to speed up their insights.

But not everybody is convinced that the market will move in this direction. Richard Higginbotham, marketing manager at CDMS, says that, “it is something that is quite worrying. Analytics is not just number crunching – it is a combination of understanding the context and applying commercial awareness.”

Programming an application to produce a model can be done anywhere, but will the resulting model make sense and be commercial appllcable, he wonders. Equally, the underlying data set may have definitions and variables that are quite specific to the market in which they were generated and are hard to understand outside of it.

“When you start from scratch, you have got an idea that might lead to a useful insight. That initial train of thought has to stay very close to the business,” says Higginbotham. Those ground-breaking insights often emerge from explorations of the data which do not start with a clear objective, but which suddenly spot a trend or pattern that can be acted on.

In other words, the inspiration often comes as a result of the perspiration. There is an historical argument in favour of that view. Yet the current move towards KPO and the suppliers who are driving it seems to contradict it. It is too early to say whether you should or should not sweat the small stuff.


    Leave a comment