How Vestas Wind Systems used outsourced machine learning to transform contract management – Diginomica

Vestas wind turbines in Australia

Our diginomica inboxes are awash with machine learning PR pitches. But when I got the chance to talk to Vestas Wind Systems A/S about their lessons with machine learning in action via an outsourcing partner that got flagged, in a good way.

Henrik Stefansen, Senior Director, Global IT Sourcing at Vestas Wind Systems A/S, gave me the inside view. Founded in 1945, this Danish manufacturer and servicer of wind turbines has become a global player in wind energy. Now with turbines in more than 70 countries, Vestas bills itself as the only global energy company dedicated exclusively to wind energy.

Five years ago, Vestas Wind Systems was dealing with the complications of declining government subsidies. The global economy was working its way out of a recession. Higher operational costs combined with sluggish energy demand compelled Vestas to push hard for new efficiencies. Stefansen has been an IT leader at Vestas for sixteen years. In the last four years, hes led a drastic change:

Weve gone from being a fully insourced company on the IT side about four years ago, to today being more or less fully outsourced. So thats been quite a journey.

Managing outsourced processes has brought a learning curve:

Ive come to realize that a lot of the other stuff that we need to be able to handle and the processes you need to have in place to manage an outsourced setup is quite different from when you run everything yourself.

That opens up a chance to improve processes:

Thats where really we got into looking at, How can we optimize and automate some of these processes instead of doing everything manually?

Stefansen handles these operations with an internal team of twenty, and about a dozen externals. 27,000 employees count on his teams IT services. If you cant handle the breeze, dont be in the renewable energy business:

We went through a bit of a dip through the financial crisis around 2011, where we cut the company in half. We had to reduce that much. But we recovered from that, and had a record year last year.

How has the wind energy business from Stefansens early days at Vestas?

When I joined the company, we were still sort of an entrepreneurial startup. Over the last five, seven years its been much more industrialized. Now wind is a competitor, and its a subsidy to all of the known coal and gas sources as well.

Today, I would say, wind is more or less on par with coal and gas, also from a cost perspective. And thats of course what weve been working towards the last many years If you want to sustain a business like this, it has to be comparable on a cost level to the other energy sources out there thats roughly where we are now.

Success brings its complications:

Looking at it from a country or global perspective, theres no doubt that renewable energy is high on the agenda in most countries these days. That makes it a nice place to be in a company like this. But its also a highly regulated environment Theres a lot of restrictions from local governments that we need to also work with to promote this kind of energy.

Stefansens approach to outsourcing has changed also. At first, outsourcing was a tactical decision in response to the economic downturn: We had to reduce head count, we had to reduce cost, and we had to do it fast. As Vestas bounced back, Stefansen decided that outsourcing was their future course but now they approach it more strategically.

Outsourcing makes sense for Vestas on several fronts. It solves the challenge of needing to staff up internal IT in Denmark. Stefansen also likes the flexibility on cost and exposure to new technologies:

We also saw the possibilities of joining forces with some of the big outsourcing vendors out there that have thousands of people. They can bring us those new technologies much faster and better than we could develop it ourselves.

And thats where SirionLabs comes in. Stefansen found the downside of outsourcing was managing the services. Ideally, he could automate a big chunk of contract management, and have it delivered as a service. During his research, Stefansen found SirionLabs. He evaluated a range of providers:

I looked a few of the big ones, including IBM and SAP. They had good capabilities in some of the areas that I needed. but none of them really had the view and connectivity between the different parts of the process that I saw with Sirion.

Stefansen also liked SirionLabs cloud emphasis:

Their software as a service comes pre-configured out of the box, so you dont have to do the on-site installations and set up. Basically, I just ship my contracts to Sirion, They upload them in India, and we are live.

Vestas started working with SironLabs in 2015. They spent the first few months uploading contracts, but that wasnt the biggest change:

Once you start working with a tool like this, there is a set of processes that enables you to get the benefit out of the tool. That was the main part of the implementation to get those implementations within our own organization.

The big surprise wasnt process change; it was the people side.

Thats probably one element that surprised me a little bit how much energy I had to put into my own organization to get my own colleagues to work in these new processes.

What changes did Stefansen see after going live with SirionLabs? One big change: tracking of deliveries and obligations. Sirion pulls all of the outsourcing vendors obligations from their contracts, and puts it into a calendar view for tracking:

All of that is alert-based. Alerts tell us that, This is supposed to be delivered now. Did you receive it, or is it still pending? In the past, we would have missed that, because it would have taken a lot of manual effort to track all of this.

On the IT side, SirionLabs is now handling Vestas four main outsourcing partners, comprising 70-80 percent of all outsourced services. Its really a shift to pro-active way to manage outsourcers, Stefansen has already seen cost reductions:

[Another part] of our cost savings is the invoice reconciliation. Basically, matching invoices to what weve agreed in the contract, and making sure that we are paying them correctly. Thats where we see a lot of the direct cost savings.

The savings arent small: Stefansens first year calculations on the SirionLabs investment: a 300 percent ROI.We talked about the machine learning aspect. Stefansen doesnt need to know the inner workings of Sirions machine learning capabilities to see the value on his side.

SirionLabs applies machine learning to areas Vestas would have struggled to monitor on their own, from incorrect invoicing to avoiding SLA penalties that are invoked when a usage threshold is reached. As the SirionLabs PR team put it to me, SirionLabs uses machine learning to cull through the mind-numbing tedium of contracts to ensure everyone is doing their job.

Looking ahead, Stefansen wants to get his outsourcing partners to use SirionLabs to collaborate and address contractual issues. So far, weve seen good benefits from that, where weve managed to convince our outsourcing partners that this is a good idea.

Today, SirionLabs manages contracts valuing $160 million for Vestas. For Stefansen, better control over back office IT means his team can be more strategic, and less caught up in administrivia:

If I hadnt implemented this I would probably of had to hire say four people to manage these things manually. So it gives me a lot of flexibility from an organizational point of view.

Image credit - Image of Vestas wind turbine in Macarthur, Australia from the Vestas.com web site, model number V112-3.0 MW.

See the original post:

How Vestas Wind Systems used outsourced machine learning to transform contract management - Diginomica