There are at least 25 ways to screw up analytics and AI

analytics AI
Image credit: Artur Szczybylo / Shutterstock.com

ITEM: Delegates attending the first AI/analytics track at Digital Transformation Asia may now have a better idea of what not to do with AI and analytics technologies than what they can do with them.

The track session – “Unleashing the Potential of AI and Analytics”, which took place on Tuesday – aimed to advise CSPs on how they should be formulating their AI adoption strategies for maximum advantage, where they should consider starting, what sorts of AI technologies or solutions might be appropriate for this, and so on.

And there was some of that (albeit occasionally in the form of barely disguised adverts).

However, perhaps the most useful presentation came from Pedro Uria-Recio, head of the Axiata Analytics Center at Axiata Group, who was scheduled to talk about the AI technologies CSPs should place their bets on, but ultimately decided that was the wrong question to ask.

“You’re not betting on technology, you’re betting on your ability to identify and solve problems,” he explained.

So rather than list technologies telcos should invest in, he changed his topic to “The Top 25 Mistakes You’re Making In Your Advanced Analytics Program”.

In terms of strategy, for example, telcos are assuming analytics is a plug and play solution that will provide short-term ROI (it’s not, and it won’t), or using it to solve problems that don’t need solving (the classic ‘solution in search of a problem’ conundrum).

Other strategy mistakes include relying solely on vendors or consultants, failing to prioritize, and failing to incorporate external data in their data sets.

Companies also tend to make lots of personnel/organizational mistakes like organizing analytics under non-operational functions, having too many analytics teams that create their own silos (which defeats the entire purpose), hiring only PhDs or CDOs who can’t explain analytics to anyone in the company regardless of technical know-how, and not creating the data-driven culture necessary to embed analytics into the company’s processes, which in turn is necessary for the company to turn results into action.

Failing to scale up pilot, neglecting automation, not using Hadoop, not allocating a big enough budget (or allocating too much) – the list goes on.

The fact that Uria-Recio had enough material to fill out a Top 25 list (when he could have settled for a more traditional and economical Top 10) is itself a useful takeaway. Granted, some of these sound so basic – especially in these days where data is under a harsh spotlight in terms of theft, misuse, governance, etc – that it’s hard to believe any halfway competent CxO would commit any of them. (Neglecting data governance and security? In 2018? Seriously?)

Then again, if everyone understood the basics of big data analytics and AI, we wouldn’t need conferences like this to explain it to everyone. In which case, it’s arguably helpful to explain what organizations are getting wrong before you start telling delegates how to do it right.

Be the first to comment

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.