BDA: Don’t be penny-wise, pound-foolish

  • Lite or free tools can smother a business with long delays, limited vision
  • Tools are useful, but humans are needed to figure out the unknowns
BDA: Don’t be penny-wise, pound-foolish

COMPANIES are stepping up their efforts to become more data-driven, hoping to glean insights and trends into their business. However, as costs mount up, some are tempted to use the free or ‘lite’ versions of big data analytics tools just to keep up.
But such tools can put limits on a company’s vision and what their data can do, according to Ng Fai-Keung, Asia Pacific group country manager at French web analytics company AT Internet.
“The free ones come with a lot of limitations, and I think people will be constrained by these limitations,” he says, speaking to Digital News Asia (DNA) in Singapore recently.
“For example, you get high-level reports, but you don’t get granular data. Sometimes … there is a long delay in getting the data, or some of the data is provided in real time while some is not.
“The implication is that if you don’t get granular data, you can’t do integration with other sources – so you have limited your view on what data can do,” he adds.
Man vs machine

BDA: Don’t be penny-wise, pound-foolish

Quoting analytics evangelist Avinash Kaushik, Ng (pic above) says there are three categories of analytics: Known knowns, known unknowns, and unknown unknowns.
“Each set requires different expertise and talent … data scientists are for the unknown unknowns,” he says.
Unknown unknowns is essentially: I don’t know what’s going on, but put the data in a black box and have algorithms to figure out a pattern and discover something.
The big opportunity lies in the known knowns and known unknowns, where tools can help.
The statement for known knowns is: I know that I need certain metrics and when I see it, there is something I need to do.
“Some tools could potentially serve this purpose by incorporating business [applications] and making the data easy to digest – with visualisation … graphs and dashboards being able to tell the story,” says Ng.
But for known unknowns, more exploration is needed.
“You need analysts, so if their hypothesis is ‘Maybe people who come from Facebook have a shorter engagement time,’ they need tools to verify the hypothesis,” says Ng.
“From there, they can say, ‘Okay, perhaps because of that, content served from Facebook needs to be shorter.’
“[For such uses], I think it is harder to let the tools figure out the cause and generate insights for you automatically – you do need analysts to synthesise the hypothesis, using the tools to crunch the numbers,” he adds.
In all three categories, tools can generate initial automated insights, but ultimately companies would still require a human behind the machine, Ng argues. “It requires talent on the vendor or client side to fill the gap.”
Asking the right question
All sorts of skills can be useful in data analytics, such as mathematical modelling or number crunching, but being able to ask the right question is the key. You need to understand the business to know what questions to ask, Ng advises.
When you know the questions you need to ask, you can go to the vendor ask, ‘How can you help solve these problems? How do I get the data to solve these problems?’
“If this vendor can’t solve it, it’s not the end of the world – there are so many other vendors which can help you,” says Ng.
When it comes to Asia, he feels that companies here need to go back to the basics, as without the proper foundation, nothing can be built.
“When we talk to clients or prospects here about how they collect data, most of the time they collect it as a standardised table,” says Ng.
“From the get-go, their data collection already limits their vision.
“It is in the simple metrics like the ‘known knowns’ part – getting the data collection correctly set up allows you to generate hypotheses. And it is through the generation of hypotheses that you know more about the data, and you then move to the next level.
“Right now, we are jumping to the most advanced level because of the hype … but we need to get the basics right,” he adds.
Related Stories:
What’s Next 2016: Data, models and asking the right questions
How to ensure your BDA is driven by business needs, not technology
Too much data, we’re only human after all
For more technology news and the latest updates, follow us on TwitterLinkedIn or Like us on Facebook.

Keyword(s) :
Author Name :
Download Digerati50 2020-2021 PDF

Digerati50 2020-2021

Get and download a digital copy of Digerati50 2020-2021