Companies are getting swamped with data. While businesses have more information about their customers, clients and broader environment than they could have possibly imagined a few years ago, they are struggling to make sense of it.
Although there has been a lot of talk about 'big data' giving magical insights, data analytics has yet to live up to the hype. This roundtable looks at how to best use it, what to use it for and how to get the C-suite to buy into it.
How big is the danger that we see patterns in data that are not really there? Is getting insight from 'noisy data' a problem?
Windsor Gumede, senior BI consultant, PBT Group: One of the fundamentals in working with data is making sure that it has been 'cleaned', so it does not give distorted results. There's truth in the saying, 'Garbage in. Garbage out.' So if you make a decision on poor-quality data, it's going to have an impact on your results.
You also have to make sure that the data is warehoused.
Zakes Socikwa, analytics, business intelligence and big data lead, Oracle South Africa: In my experience, you have to put a very good data collection infrastructure in place. This, for example, means keeping track of all the interactions with your customers. If you don't do this, you won't be able to analyse these relationships.
What's changed is that there is now a lot of information sitting outside of a company's control that can tell you a lot about the relationship with its customers. It's out there in social media, it's out there structured in other forms.
This is why big data is becoming so valuable. If this outside information is merged with that warehoused by the company, you get much better results.
Gary de Menezes, country GM, Micro Focus Software: I think we are looking at data analytics too much through a corporate perspective. If you take out your phone and search for images of trees, you will find every picture on your phone with a tree on it. This is a form of data analytics.
Same thing with Google. It advertises something you just searched for. It's part of our everyday lives without even realising it because it's about pushing the right thing, to the right person, from the right company.
The world of data analytics has changes remarkably over the past two years. The solutions have become more industry-wide than industry-specific. This means the same solutions can be used by multiple companies and in multiple use cases.
The world of data analytics has changes remarkably over the past two years. The solutions have become more industry-wide than industry-specific.
Gary de Menezes, Micro Focus Software
It's part of a broader evolution in artificial intelligence (AI), machine learning and robotics. It's actually about the full automation of business processes, regardless of what industry you're in.
Eshaanan Gounde, senior middleware specialist, SAP: I hear what you're saying, but how do you go about building a data set, which takes into account a machine learning model?
AI, for instance, works by following a set of instructions and knowing what patterns to look for.
Machine learning needs AI, but we have to give it free reign within a data set, so it gets an opportunity to learn.
But the data is not really set up for a machine learning model. People want the data collected, put into a data set and they want to be able to report on it.
Gustav Piater, marketing & sales director, AIGS Insights (Yellowfin BI South Africa): Just coming back to the question of noise. One of the reasons we don't get value out of data at this moment is because of noise. How do we address noise?
This is what the BI vendors are trying to address. They are using AI to clean the data as it's pulled from the source systems, instead of trying to clean it after the fact.
Are we making the best use of analytics?
Mike Munsie, head of analytics, Experian: I think the train of thought around this conversation is being driven around the collection of data, data warehousing, and the like. But in my interactions with corporates, although they talk a lot about collecting data, and throw around catch phrases like AI and robotics, I've yet to see this actually implemented.
There is a big difference between the collection of data and the practical use of it.
I see lot of dashboarding of data, but I don't see a lot of predictive analytics. You might see it in retail, but not really in financial services.
Sudhir Juggernath, head of Orange Applications for Business, at Orange Business Services: To me, it comes down to what you want to extrapolate from your data. What are your KPIs? What are you looking for? Once you know what you are looking for, that's when you get the value.
Chris Wigget, director of Insights and Analytics, Britehouse: One of the easiest ways to fail is to do analytics for the sake of analytics. I think there's a fair bit of that. 'There's a buzzword, so let's do it'. There are no clear objectives and no thought is given to what questions need to be answered.
Bill Hoggarth, enterprise information management (EIM) business unit manager for Gauteng, Datacentrix: I'm quite enjoying the debate, but I feel we are caught in a time warp. Virtually everything said today could have been said 15 years ago. We said back then that we have all this dirty data, we need to consolidate it, clean it, put it in a data warehouse, and then use it to find patterns.
That was 15 years ago and so much has changed. Yes, we have done a lot in that time period, but now we're coming to the core question of how we extract value out of data analytics.
For me, this is starting with the end in mind, and using the analytical processes to enhance the function points, the process and the strategy.
Zakes Socikwa, Oracle South Africa: We need to differentiate between different kinds of analytics out there. You have your basic analytics, which is basically your addition, subtraction, percentage kind of stuff. But when we talk about advanced analytics - in the past, we called it data mining - you don't use it with preconceived ideas. You go there with statistical techniques and have the data 'talk to you'. It will give you the feedback.
This is why we are bringing in forms of data not traditionally used. We are bringing in voice, text and images - unstructured data - to enhance the results.
Gary de Menezes, Micro Focus Software: I'm going to offer a different perspective. The reason why we have not gotten the best use out of data analytics is because of how our customers are internally structured. The CIO is mandated to build a data warehouse, but has no idea what to do with the data. No one inside the company tells the CIO who owns the data.
Take any of our customers; do any of them know how to extract value out of it?
The people who own it are traditionally business people, and they don't understand what to do with it. The technology to extract value out of the data is available, but there is gap in the understanding of how to use it at the C-level.
This means the people responsible for delivering the value are not able to see how to do so. They will continually do what they have been doing over the last 30 years, which is how they got into the C-suite in the first place.
This is a major problem in South Africa. While we are seeing several pilot projects, like the use of robots to automate mundane tasks, business in this country is falling behind when it comes to adopting new technologies.
Saying this, I see Bank Zero and Discovery Bank coming out with something that will rock the foundation on how business engages with customers.
This might spark a change in other businesses in South Africa.
Bill Hoggarth, Datacentrix: If this happens, it will have a profound impact on the business models of the vendors. If you look at data analytics through the lens of licensing fees and consulting hours, you are absolutely compromising the willingness, and the ability of the end-user, to adopt data analytics more broadly.
Look at Experian. It sells the end result of probably the most sophisticated analytics process out there today. If you buy a credit score from them, you don't care where they got their data or how they cleaned it. You are paying for a result.
Vendors have to follow their lead. They have to sell the results of data analytics solutions, rather than the solutions. They should, for instance, sell a prediction of the quality of sales lead.
If this happens, we will see the adoption of data analytics sky-rocket. At the moment, we are not doing this.
Why is data analytics not getting traction in South Africa?
Zakes Socikwa, Oracle South Africa: The way businesses are structured is an issue. In South Africa, we have the CIO reporting to the CFO. The CEO and CFO are on the board, but the CIO is not part of this decision-making structure.
This means, when vendors want to present a business case for their solutions to the key decision makers, it becomes a challenge because they only have a relationship with the CIO.
Mike Munsie, Experian: I just want to come back to my point that data as an asset truly hasn't been unlocked. People spend a large amount of money collecting it and storing it, but this is really driven by the need to be in keeping with the data regulations.
One of the easiest ways to fail is to do analytics for the sake of analytics.
Chris Wigget, Britehouse
This is particularly true in the banking space where they will use it to be compliant from a regulatory perspective. But what's really needed is to move to the next step, which is moving past regulations. It's moving to consumer interaction, where an enterprise is able to plot a journey for a consumer at an individual level.
I don't think enterprises have been able to sweat their data assets to get to this point. There are very few use cases, and you generally have to go abroad to find examples.
Windsor Gumede, PBT Group: I've been out of the country for the most part of the last decade, and on coming back, I expected us to be further ahead when it comes to using data analytics.
We are still using it at a descriptive level. We are not using it to make predictions. It's not where it should be. There are very few companies that are using it at the very edge.
Chris Wigget, Britehouse: There are pockets of intent to do proper analytics, but most people are caught up in a world where they have collected data from operations, but not with an eventual goal in mind. To get analytics to work for them, they have to rethink what they want from this data.
Windsor Gumede, PBT Group: I just want to agree with Bill, on defining the end objective. Yes, you can pull in all the data, but you have to define the end objective.
It needs to align with your business strategy.
Bill Hoggarth, Datacentrix: It worries me that we constrain the discussion around data analytics, within the box of decision making and action. I think it should also be used to find out things you don't know about your business. One of the untapped uses of analytics is to satisfy the curiosity of line and executive management, who want to know more about their business.
A machine is always better at spotting patterns in large streams of data, so it would be great at showing them something they don't know. We don't always have to make decisions.
Gary de Menezes, Micro Focus Software: I want to add to that. It can be used to not only track what's happening to your business, but also to monitor the sentiment of your customers.
I agree, it's not always about decision-making.
Mike Munsie, Experian: In the ideal world, it would work like this. But in terms of how corporates operate, they are driven by billable hours, or very measurable outputs, and as a result, they are not incentivised to be more curious about their business.
Bill Hoggarth, Datacentrix: I hear what you are saying, but if you make knowing more about your business a measurable outcome, you will end up designing a process to get this outcome.
I for one think it's a very, very underrated outcome.
There is a big difference between the collection of data and the practical use of it.
Mike Munsie, Experian
And to Mike's point, our challenge as the vendor community is to drive the right culture, and the right way of thinking, in where businesses manage data as an asset, in the same way they manage people and money.
We know we will be successful when organisations are spending on data science for non-data science people, as they are spending on finance for non-finance people.
This article was first published in the November 2018 edition of ITWeb Brainstorm magazine. To read more, go to the Brainstorm website.
Share