Big data analytics is becoming big business with waves of new agencies being launched. David Reed looks into what it takes to move from start-up status to profitable star by ensuring the ideal collision of data science, technology and consultancy.
If you live in Mountain View, California, then you might notice an almost imperceptible dimming of the lights at around 6pm. The reason? Workers from Google’s headquarters returning home and plugging in their electric cars to recharge. If you are the electricity supplier to the area, understanding that behaviour is vital for two reasons - firstly, to ensure you manage generation and distribution resources to meet this demand and, secondly, to encourage some changes in this activity which might smooth out the load.
This is an example of the way technology is generating data streams which provide entirely new insights into customer and business issues. Reading that data and making sense of it is no simple matter, however. It requires a combination of genuine data science, innovative software and commercial understanding.
Offering those skills is a new generation of data analytics agencies which have burst into life over the last five years or more. Start-ups abound in this space seeking to tap into the surging demand and investors are lining up to place their money behind such service providers. So what does it take to create, launch and sustain an agency and just how easy is it to move from start-up status to significant player in this sector?
Onzo is a good example of the shifting nature of analytics services. Set up in 2007 as a provider of in-home displays and sensors, it divested itself of the hardware side of the business within a few years. “We see the future of data analytics in utilities and energy as cloud-based, software-as-a-service using the patented algorithms we have created,” says chief executive Spencer Rigler.
As a technology-agnostic agency, Onzo specialises in extracting insights from smart meter reads down to one-second frequency, processing trillions of readings to extract value-driving business intelligence. “In Texas, 60 to 70 per cent of energy consumption is for air conditioning,” notes Rigler as one example of the behaviours and issues faced by his clients. Through applied data science, the agency is able to drill down into the data and identify what individuals are actually doing in terms of their energy usage and activity profile.
For Dr Katie Russell, who joined as head of data and analytics in 2013, this was a major part of the appeal. A graduate in physics from Edinburgh, she had been working in a water company, “then I joined Onzo because energy data felt like a major step up in big data’s evolution. It is required for accurate billing, but because consumers use energy for so many things, its real value is in what you can learn about their lifestyle.”
With utilities operating in a highly-regulated market and facing decreasing margins, tapping into data insights is being viewed as essential for creating sustainable new business models. As a GTM Research survey among 400 energy executives in the US recently discovered, behind distributed energy resources, customer engagement is ranked as the second biggest business opportunity.
Onzo is addressing this with its Personalised Consumer Insight and Personalised Consumer Engagement services, which even extend to a mobile phone app which one Dutch energy company is about to offer to its customers. “There is a lot of discussion about the Internet of Things and its impact on business. I personally believe it is the Internet of People that will matter,” says Russell.
The agency is in the process of investing in its human resources to scale up from its current level of just over 30 staff, with a nine-strong sales organisation and five-member consultancy team wrapped around its core data science and engineering capabilities. “We are looking to double in size in the next six months because we are in a very hot space. Everybody is interested at the moment,” says Rigler.
Blue Yonder is a prime example of just how hot the analytics sectors is from an investment point of view. Created in 2008 as a joint venture between German retailing giant Otto and the former CERN scientist, Prof. Dr. Michael Feindt, the big data analytics provider gained $75 million in backing from Warburg Pincus in December 2014, the largest private equity spend yet in this space.
The backbone of the company is its algorithms using neural network-type approaches to tackle problems in the retailing sector. Forward Pricing is a new solution it launched in July to help e-commerce retailers optimise their prices and model what effect price changes have on sales. “It has a tremendous impact on business to get the accuracy level of its forecasts high and make better decisions on how to optimise the supply chain and demand programme,” says UK managing director Rakesh Harji.
Data scientists who used to work on predicting where particular collisions might happen are now focused on what might happen with every product in every store every day across a retailer’s estate. That team has grown from five at start-up to 150 now, with 85 per cent holding PhDs. Says Harji: “Retailers have got data coming in so fast, but they need to service their customer better. Those data sets are extremely complex, which is why you need people who are used to that level of complexity.”
The company has a campus-style culture to help its scientists feel at home and has also created the Data Sciences Academy to help business executives understand this new area of activity. EAT’s chief financial officer Strahan Wilson is one graduate (see “Serving up better margins,” DataIQ Issue 18). The business has even generated its first spin-off in the form of sensalytics, a retail sensor technology vendor.
Harji says the new investment is important to help develop the company into a global player. “The partner we are working with is a very established investor in technology companies and is not in it for short-term gain - it is for the long-term, helping to take us to the next stage beyond Europe. We are very strong in Europe for predictive analytics which we are confident we can grow in other interesting markets, like the United States,” he says.
What these examples demonstrate is that the major players in data analytics in the future need to offer a combination of data science, technology and consultancy skills. Even those started by a like-minded team of analysts have to be capable of putting other core competencies in place, although it is the nature of the practitioners which makes the difference.
“Our greatest asset is our people,” points out George Frangou, founder and executive chairman of Massive Analytic. It is a start-up specialising in precognitive enterprise analytics, although it can trace its origins back to between 2010 and 2012 when Frangou began to explore whether data analytics could have predicted the financial crash. “I looked at the whole area of precognition of ‘Black Swan’ events using artificial intelligence,” he recalls.
Discovering that existing services would not have spotted the 2008 crisis, he still identified the power of probability when applied to specific problems with business forecasting. He already had a track record in the sector, having built and sold Pronus, an online analytics service provider.
After raising seed capital, he spent two years building the analytical engine, now called Oscar AP, which has already attracted Amazon and Lockheed Martin among others as clients. From the outset, however, Frangou has structured Massive Analytic for growth, drawing on his career experience at KPMG during the dot.com boom.
“We are not set up like a start-up - we have a board of directors with three executives and two non-executives. That line-up has already changed several times, but we are also developing people. One of our staff started as an intern and is now a vice-president. We also pay above market rates to recruit people who can accelerate things,” he says.
Although still relatively small at around 15 staff, the company has raised £1.5 million first round funding and a trade sale is on the agenda, according to Frangou. Massive Analytic is now migrating its technology out of AWS to allow it to work on rivals to Amazon. “Our idea is to be disruptive,” he says.
Aquila Insight has achieved the transition from start-up to established analytics agency in a relatively short period of time. “We wanted to get to 30 staff, but we are already at 43 - 18 months ahead of schedule,” says co-founder John Brodie. Already involved in client pitches against heavyweight rivals ranging from management consultancies to data services groups, he puts this ascent down to a clear-minded strategy adopted at the outset.
“Co-founder Warwick Beresford-Jones and I both felt there was a better way for things to be done, so we came up with three core pillars - technical skills, management consultancy polish and client governance,” says Brodie. While all start-ups have the first of these at their core, either in the form of PhDs and data scientists or practitioners with deep understanding of the problems, the latter two qualities are what he identifies as frequently missing.
“The big four management consultancies are very good at developing key messages for the C-suite, but when you scratch the surface, they are not necessarily technically excellent. Our view was of being able to pitch to stakeholders on the board, so we put a lot of effort into that polish,” says Brodie. Similarly, agencies are generally better at winning than managing their clients, something both partners were keen to improve on.
One direct consequence is that Aquila Insight has not lost a client, regularly winning project work which then converts into long-term engagements. Brodie also notes that, “we probably went after far bigger clients than we should have for a company our size.” Both have both worked at Bank of Scotland and other client and agency roles where they saw both the best and worst of analytic services providers.
“We are delivering something the bigger agencies don’t. Clients are facing ever higher volumes of data, buying more software to get value out of it and it is not happening. So offering a different way to do things is appealling,” he says.
Moving into this new data analytics space is not solely an opportunity for start-ups. Market research and insight agency Simpson Carpenter brought in Frank Hedler as its director of advanced analytics in September 2013, bringing a background of maths and physics as well as experience of working on marketing sciences at GfK in Germany.
“I have worked on traditional segmentations, conjoint modelling, risk simulations and the like. A few years ago, those hit a barrier because marketing had gone digital and it became about behavioural data. Clients started to question the value of market research. That was the first time I got into very large data sets,” he says.
Text mining proved to be a route into a new data sciences practice with the need to understand sentiment in ambiguous statements and third-party sources. Combining existing marketing sciences techniques with open source tools, like Python, and the addition of data science thinking created the new advanced analytics proposition.
Klondike uses Natural Language Processing (NLP) algorithms which are the result of extensive research at the Stanford NLP Group and the University of Massachusetts, and represent the state-of-the-art in text analysis and machine learning. The results, fine-tuned to a client’s specific data and requirements, are accessible via intuitive online platforms, with interactive, easy-to-interpret graphics and visuals.
“It takes us into a new market so now we are competing against operational CRM systems like Clarabridge that scan incoming emails and segment them according to their content. Our offering is in a similar space by operationalising data insights from any source,” says Hedler.
When a market sector starts to gather momentum, it begins to draw in other players from outside the space. This gravitational pull not only attracts investment, it maintains the excitement and interest among clients which is essential for growth. Advanced analytics right now is transforming how companies understand their business, markets and consumers. You don’t need a PhD in nuclear physics to realise what this big data Big Bang will lead to.