# Analytics or Hair Splitting?

Advances in Big Data, Internet of Things and other technologies are contributing to an ever-increasing quantum of information in today's digital world. Not surprisingly, analytics has become a hot topic once again.

But I sometimes wonder if there's such a thing as "too much analytics".

I'm not thinking of "analysis paralysis" where the analysis leads to no action. I'm actually referring to analysis that does lead to action but of the type that resembles hair splitting. Not that there's anything wrong with hair splitting, but more on that later.

I recently went through this "analysis or hair splitting?" debate in my mind when I read the following average figures for a checking account in this Bank Director article:

• Balance: US\$ 5600
• Cost: US\$ 250
• Revenue: US\$ 413

It doesn't take a PhD in mathematics to calculate that the average profit is US\$ 163. Therefore, it should be obvious to everyone that an average checking account is profitable for banks.

But it's not, to the the authors of this article. They claimed that the average checking account "doesn't pay for itself". I pointed out their erroneous conclusion in my comment below the article. One of the authors emailed me to admit their faux paux, so I'll let that pass.

But what really got my goat was the article's assertion that averages don't tell the real story of checking accounts.

Well, why single out checking account? Isn't this true for everything in life?

Perhaps the best illustration of the danger of over-reliance on averages can be found in this episode where a mathematically-obsessed but swimming-challenged guy drowned in a lake thinking he could comfortably wade across it because he was six foot tall and the lake was only four feet deep - on an average.

However, if averages told the whole story, who'd read the rest of the story?

Anyway, proceeding with their platitude, the authors urge banks to take “action to cure the unprofitables and protect the profitables” by going beyond averages and drilling down at a more granular level.

There's nothing wrong about this advice since FIs are under constant pressure from Wall / Dalal Street to trim their unprofitable businesses while simultaneously bolstering their profitable franchises. But, it's while implementing this suggestion that analytics could turn into hair splitting, for the very next question would be, at which of the following levels should an FI carry out profitability analysis and recommend remedial action?

1. Product level? e.g. Be happy if the overall checking account business is profitable.
2. Geography level? e.g. Since checking accounts are unprofitable in rural areas, shut down all branches in villages.
3. Account level, as recommended in this article? e.g. Close all unprofitable checking accounts.

To use consultant-speak to answer the above question, each bank should decide the optimum level at which to measure profitability by comparing the cost of every incremental level of analysis with the benefit of the action coming out of it.

Unfortunately, this classical business case approach is challenged by several qualitative considerations in this context e.g. business ethics, reputation damage and regulatory rap on the knuckles.

For example, what'd happen if customers whose checking accounts are terminated unilaterally by banks - for whatever reason including low margin - take to social media to vent their fury against the bank, the way Brett King did when HSBC USA canceled his account? The resultant public backlash could snowball into a PR crisis that might thwart other people from becoming - potentially profitable - customers of the bank.

However, banks shouldn't be deterred by such challenges. I love analytics and, as I'd hinted at the beginning of this post, I'm quite ambivalent about hair-splitting. Therefore, I can't resist the temptation of urging banks to drill down even further and develop a strategy at the sub-account level. Let me illustrate this approach with the following example:

While John Doe's checking account is profitable on the whole, deeper analysis reveals that he uses ATMs a lot more than the average checking account customer, thereby increasing the bank's operating cost for servicing his account. Accordingly, the bank should take remedial action to boost profits by weaning John Doe away from ATMs (while "doing nothing" in the case of other customers).

In this pursuit, the easiest option for the bank would be to slap excess-ATM-usage fees on John Doe and other customers like him. However, that might invite the ire of the regulators. Therefore, let me suggest an alternative approach:

Float a ghost-like image on the screen as soon as John Doe inserts his debit card into the ATM slot. Undeterred, if John Doe proceeds to enter his PIN number, replace the customary ATM usage instructions by blood-curdling screams piped through the speakers. If John Doe still stands his ground, threaten to swallow his debit card on his next visit to an ATM.

Jokes apart, where do you draw the line between analytics and hair-splitting? Please share your thoughts in the comments below.

4890
###### Report abuse

Nick Collin - Collin Consulting Ltd - London 20 May, 2014, 18:25 0 likes

Nice blog Ketharaman

I once did a consulting job at a bank where, thanks to an unusually good IT system we were indeed able to measure profitability at account level.  We then simply sorted the accounts and looked at the most unprofitable.  The worst categories were bank staff, businesses wrongly categrised as consumers, and friends of the branch manager!

###### Report abuse
Ketharaman Swaminathan - GTM360 Marketing Solutions - Pune 21 May, 2014, 12:27 0 likes

@NickC: TY for your feedback. Just out of curiosity, did this bank shed this unprofitable segment?

###### Report abuse
Nick Collin - Collin Consulting Ltd - London 21 May, 2014, 12:55 0 likes

I don't know - I'm a consultant - I make my recommendations and move on swiftly, leaving others to do the dirty work ;-)

###### Report abuse
A Finextra member 21 May, 2014, 14:02 0 likes

As I recall in risk analysis it is, Identify Risks, Analyze Risks, Prioritize Risk.  Then plan mitigation, implement mitigation, and manage mitigation.  Then rempeat.  Priorizing what is important is the key.  Big Data is cool, but it's not meaningful without a plan.

###### Report abuse
Ketharaman Swaminathan - GTM360 Marketing Solutions - Pune 21 May, 2014, 18:01 0 likes

@NickC: LOL. Hope you didn't mean "move on SWIFTly"?:)

@TonyW: TY for your comment but I must respectfully disagree about the sequence of activities: The fun is in first drinking the Kool-Aid of Big Data because it's, well, cool. Then unraveling hitherto unknown trends. Then taking the recommended actions. Then deciding whether to have a plan or not!

### Ketharaman Swaminathan

Founder and CEO

GTM360 Marketing Solutions

Member since

17 Apr 2009

Location

Pune

Blog posts

137

4,246