3 key components for a successful big data strategy
An Interview with Mike Leverington, Head of Customer Data and Analytics at Guardian News & Media
Mike Leverington is Head of Customer Data and Analytics at Guardian News & Media. Prior to working at the Guardian Mike worked both client and agency side helping companies leverage value from their customer data. Ahead of the Chief Data Officer Summit, Leverington spoke to Andrea Charles from PEX Network about the key challenges transforming customer data into customer insight, the reality of customer experience and utilizing data analytics to understand customer need. Leverington also shares his 3 key components for a successful big data strategy.
PEX Network: What are some of the big current data trends that you're witnessing?
M Leverington: The main data trend that I'm witnessing at the moment is the move towards data federation within companies, primarily brought on by the growth of digital data and technology. Organizations are now collating far more detailed data and at greater volume than ever before.
There is a massive opportunity to leverage this data to improve customer experience and organizations are trying to take up this opportunity.
Technology has moved so fast, it's now within the capability of most companies to get access to their data and be able to analyse it at scale. It also doesn’t cost millions to get started; a lot of the technology is open source or pretty reasonably priced compared to ten years ago.
For example, you have tools, like R and Google Analytics that are both free to use. There are also some great dashboarding/data visualization tools, like Tableau or QlikView that are reasonably priced and fairly easy to implement.
This provides an opportunity for data and analytics to permeate throughout organizations more than ever before. Whether it is being used effectively is another matter though!
PEX Network: How are big data and customer experience coming together?
M Leverington: I've only seen a few digital examples of big data and customer experience coming together in a meaningful way. That's not to sound too negative about it, but I think the growth of data and technology has been so rapid, that execution against this opportunity hasn't really caught up yet. There are a few things out there that I think are interesting, but little has been done at scale yet.
What's really important in good customer experience is not only all about data either. For example, I recently purchased two new sofas from a department store and absolutely none of my data was used in this transaction. It was a very good customer experience but this was down to the salesperson, who had excellent product knowledge, met my requirements and talked to me through the process and what to look out for.
When this retailer did have my data they actually used it quite poorly. I received six months' worth of tea and cake vouchers about two weeks after I'd bought a sofa. Not too sure what the purpose of that was! When someone has purchased a sofa it is not too difficult to imagine what other products and services they are in the market for.
But overall, the customer experience of buying that sofa was brilliant despite the poor use of my data. I will still shop with this retailer despite the poor use of my data because the customer experience when making a purchase is extremely important.
PEX NETWORK: What are some of the key challenges in transforming customer data into customer insight?
M Leverington: I think the main challenge is trying to cut across data silos and create a single customer view. This allows you to measure and analyze individual customers behaviour and then execute against your findings. If you can't get all your data in one place and create a single view then you are running uphill from the start and everything becomes a lot harder.
I think the second thing to look out for is making insights actionable. Clever isn't always best for commercial needs and isn't always the most actionable at the same time. You can do lots of analytics and produce some great insights, but if the business can’t implement them or they don't benefit the customer, then how useful are they?
On an everyday level I think analysts need to be hard on themselves and keep on asking the ‘so what?’ questions. Why is this finding important? How can it be used to benefit the customer? You need to be ruthless with yourself regardless of how much work you put into analysing the data.
PEX Network: What would be your top tip for utilizing data analytics to understand customer need?
M Leverington: My top tip here would be to start with a helicopter view of your database. I think that too often, people dive straight into the detail and try to find interesting clusters of customers to analyze, or start building predictive models.
The most effective place to start is with top line customer behaviours. How are your customers or prospects interacting with you? How can you improve the customer experience? Start with the basics and then gradually work your way down to the detail.
Finally; make sure you incorporate other areas of insight as well and do not just rely on data analytics. Analytics can tell you ‘what’ your customers/prospects are doing and how they are doing it. It is a limited method though for finding out ‘why’ customers are behaving in certain ways.
PEX NETWORK: In your opinion, what are the key components for a successful big data strategy?
M Leverington: You have to start with the end in mind and then work backwards from there. Set your objectives and goals and then build the roadmap that allows you to get there. If you don’t have clearly defined aims you will invariably end up going off on a tangent.
Ensure you have shared KPI’s for your data strategy within the organization. I can’t stress this one enough you need a set of shared KPI’s that the organization agrees on from board level down. This allows teams across departments to collaborate better and work towards a shared goal. In short this helps the organization achieve its data strategy. If you don’t have shared KPI’s then you are likely to get conflicts between teams or department objectives.
I think another key tip is don't try and boil the ocean. It's quite easy to start doing that when you get into your data. You get a bunch of people who are interested in data and analytics around the table, and all of a sudden you're trying to do extremely complicated tasks with data.
It’s crucial to keep the 80:20 rule in mind. What are the simple things you can do that will deliver 80% of your value? Prioritize those tasks and keep them as simple as they need to be.
Finally, I would say it's people, people, people. You need to have good data orientated people to help you achieve your data strategy. If you don't, then you're not going to get anywhere fast regardless of how much fancy technology you have in the business. This is not just about having good analysts. It’s about the organization having a data driven culture and using data and analytics to help drive the decision making process.
PEX NETWORK: What's next for big data?
M Leverington: Definitely the rise of artificial intelligence being used in mainstream society; you just have to look at IBM Watson to see the potential in this area.
IBM Watson is a computer being used to help doctors diagnose patients. The doctor asks Watson questions or inputs symptoms and the computer analyzes the patient’s records and current medical knowledge to present several diagnosis scenarios back to the doctor.
The benefit of Watson is that it is able to evaluate a patient’s symptoms against an extensive database of past occurrences to come up with several possible scenarios with a measure of likelihood for each one occurring. This allows a doctor to draw on far more data when making a decision on how to treat a patient.
I think in the near future we are going to see computers like Watson used to perform a lot more tasks within society from customer service to helping catch tax cheats.
Another area where I expect to see growth in organizations is in the use of machine learning techniques to tackle business problems. The type of modelling that was once the realm of PhD level mathematicians and statisticians will be more readily available to the standard analyst in an organization. In much the same way that analysts use decision tree’s or regression modelling techniques at the moment I think more analysts will be start using machine learning techniques to help solve business problems.