When Dr Yasmeen Ahmad took over as head of advanced analytics and data science, UK and Ireland, for Teradata in Jaunary 2016 it was a time of significant change for the company, as well as for the analytics marketplace more widely. Having sold its marketing automation arm, Teradata is now focused on analytical services, including its recently-launched Insight-as-a-service (IaaS).
“My role is vey much bringing that to life for clients and delivering their projects. It has been a great first half of the year and I am looking forward to the second half,” says Ahmad. “Teradata has been a technology company building fast, massively parallel processing platforms. What we have seen in the market is clients changing the way they manage and store data with the arrival of big data, Hadoop environments and other technologies. Now they are struggling with how to make sense of all that data coming at them.”
Deriving value from big data is not as straightforward as in the established realm of transactional data. Digital interactions can yield significant benefits if they are leveraged properly, especially if they are combined with those interactions to give insight into customer behaviour as well as value. For many organisations, investing in the resources necessary to gain that insight is not feasible, whether it is installing an analytical platform on-premise or hiring in a data scientist.
This is where Teradata’s IaaS is gaining traction. “Our clients have been running big data projects using Hadoop and other big data platforms for the last two or three years as proofs of concept. Now they are asking where the value is. The only reason to store data is to analyse it and solve business problems, otherwise throw the data away,” says Ahmad.
She acknowledges that achieving this goal is not just about introducing new skills sets, such as data scientists and advanced analytics, but requires a blend of overlapping skills. “You do need data scientists because they understand the data and anaytical techniques, but you also need business consultants to translate that into a business case and show the return on investment. That is vital to get buy-in from stakeholders,” she says.
Data engineers are also a core component to wrangle all of the data, hook up to APIs and third-party sources, as well as bring in internal sets which may be stored in email, PDFs or spreadsheets. The final part is to apply data visualisation expertise so the outputs from this process can be understood by the business and non-technical users.
Ahmad notes: “That is one of the main reasons we exist because there is a massive skills shortage in all of those areas. It is difficult to find data scientists who can do everything - they are expensive. So are data engineers and business consultants.”
This is why Teradata already has consultants working alongside clients on-premise. “If a client has got the platform and data in-house, but not the skills, we can help by going on site and working within that existing framework. Companies that do not have the technology or ability to do that can use the IaaS package and get access to our scalable technology without the capital expenditure,” she says.
A key dimension of this service is that data is not just absorbed into the IaaS set-up, analysed and then used to create a black box model. Ahmad is very clear that outputs have to be capable of being operationalised, running on existing systems where they have an impact. Uplift can also result from the innovative thinking which Teradata brings to bear. “Clients may have hundreds of analysts in-house, but we have explored lots of different business problems across our client base which allows us to bring in ideas that are new to them,” she says.
Using IaaS can reduce the time-to-value, whether it is adopted purely on an outsourced basis or as part of insourcing while resources are built up. Business problems will not wait until a fully-operational data and analytics function is in place.
Teradata applies its RACE methodology to keep the focus on delivering solutions. A roadmap is created which captures key performance indicators and business challenges, mapping them against IT systems and data sources. The alignment phase identifies the resources required, where gaps exist and how they will be filled.
Then the create phase identifies what the impact will be on the business, something Ahmad says, “makes us very different from other service providers,” and adopts a “fail fast” agile process. Evaluation then tests whether all of the preceding steps are working to deliver against the business case. Ahmad adds that, “I feel too many analysts have forgotten this step. It shouldn’t be an academic, data science exercise. You may get the best algorithm ever, but if you can’t operationalise it, it is a failed project.”
A distinction between data science and conventional analytics is often that something will go live at 90%, rather than waiting for it to be 100% perfect. This helps to avoid the pitfalls of data science which Netflix experienced, for example, when it paid out $1 million in competition prize money for a data science-driven recommendation engine solution that was never put into production.
As Ahmad discussed Teradata’s IaaS proposition with DataIQ, she was already busy forecasting demand for 2017 and the hiring requirement it would create. Unlike many of its clients who have to rely on their own internal resources, this is where she believes the company’s service has a real advantage. “As a global organisation, I am able to draw on expertise in other countries, not just UK and Ireland. That is a great advantage of being a start-up within a large organisation.”