I picked this little book up at a railway station for two reasons: as a trainer, I wanted to find out about the approach that business analytics people take to
introducing number-crunching, in comparison to someone like me, with a formal
statistical background, and also, I am always interested to pick up any ideas
about recruiting and resourcing an effective team, and I know those ideas are
often tucked away in unprepossesing crevices.
I wasn’t expecting much, to be honest, as I have a generally pretty low opinion
of articles on data analysis / data science / machine learning / artificial
intelligence published in magazines for business folk, but I was pleasantly
surprised in places.
Let’s say it’s a mixed bag of various articles in HBR that have been reproduced and sometimes re-edited here. Page counts and levels of technical detail vary
considerably. I’ll take each chapter in turn briefly, then talk about it in the
“Keep up with your quants”
This is insightful and well worth reading. Also the only place that talks seriously about recruiting.
“A simple exercise to help you think like a data scientist”
I agree 100% with the thrust of this chapter: engage in hands-on, playful data
collection, analysis and visualisation as a way of learning. It’s hard to do this
in the workplace, as they describe, without irking someone, or being told how to
do it, or being told to get back to your real work. I would suggest having a side
project in your spare time, but remember it is playful and a learning
opportunity. Don’t feel you must excel at it and share it.
“Do you need all that data?”
Well, probably not. This makes some reasonable points to dampen the enthusiasm of
data-lake-filling executives, though the writing is rather dull.
“How to ask your data scientists for data and analytics”
This is a really good idea. The boss needs to know how to interact with the
number-crunchers in order to get actionable insights. Unfortunately, this article
is not practically minded and sticks to theory.
“How to design a business experiment”
Vague, not practical. I hope you do a better job than this article did.
“Know the difference between your data and your metrics”
All pretty obvious stuff, though it has one good framework in a box called
“Picking Statistics,” by a different author, which importantly puts the summary
stat into a framework taking reliable data and leading to actions and continuous
“The fundamentals of A/B testing”
It’s OK. This would be helpful to managers thrown into running the data analysis
service in their organisation, and to introduce analysts to the terminology of
the business world.
“Can your data be trusted?”
All pretty obvious stuff. If you’re responsible for data analysis and learn
something from reading this, you are in way over your head, and need to hit the
ejector seat button before you do some serious damage.
“A predictive analytics primer”
Not bad, fairly obvious stuff that most readers will know. However, there’s a
couple of paragraphs on compatibility of data sources and warehousing that will
teach you more than the articles above that try to address those very issues.
“Understanding regression analysis”
Seriously, if you don’t know this, you need a proper course, not a HBR article of
8.5 pages plus a box on spurious correlations.
“When to act on a correlation, and when not to”
You can probably guess what this is going to say. Especially if you just read it in the previous chapter lolz.
“Can machine learning solve your business problem?”
This has one nice case study (reproduced from elsewhere), which I might use in
training, but otherwise it’s pretty flimsy. The question does not really get
“A refresher on statistical significance”
There are plenty of better places to learn this. There’s no mention of
controversies about replication, which are as important in marketing as they are
in psychology. Also, who does p-values in business? I thought that was very 20th
“Linear thinking in a non-linear world”
This has nothing to do with data analysis. Maybe the editors were taken in by the
“Pitfalls of data-driven decisions”
Confirmation bias, overfitting. There, you got it in three words. There’s plenty
more that could be talked about, and a gazillion insightful case studies. Their eyes are clear and bright, their voices are soft and cool, but they’re not here.
“Don’t let your analytics cheat the truth”
“Data is worthless if you don’t communicate it”
Well, ya-hah you guys.
“When data visualisation works — and when it doesn’t”
If you’re interested in business analytics, there will be nothing in here to tear
you away from your Tom Clancy.
“How to make charts that pop and persuade”
More obvious stuff of the most basic level imaginable.
“Why it’s so hard for us to communicate uncertainty”
This is an interview, not an article. It’s about a political poll, not business.
And it has several statistical misunderstandings. Avoid.
“Responding to someone who challenges your data”
This isn’t really about data. Apparently telling them to go screw themselves is
not the approach to take.
“Decisions don’t start with data”
It turns out that you should tell a motivating story and appeal to emotion if you
want to influence decision-makers. Now, there may well be readers who don’t know
that, but it’s best to cut their misery short by not telling them.
“Data scientist: the sexiest job of the 21st century”
An iconic and influential article. It’s worth reading it just to know what
everyone else has read. I’m not saying it’s right, or that it’s predictions
haven’t already unravelled a bit.
All in all, if you see this about, you might want to flick through it, but I wouldn’t cross the road to get a copy. If you’re in training, this sort of thing is important reading, if only because it encourages you to believe in your own potential. If you are considering a career in big business data analytics, skim read it (or don’t) and move on to The Signal And The Noise. Then hire people who know what they’re doing, keep your head down, and be nice to the boss. You might like to consider hiring me to come and do a training course for your team. You might like to hire me as a coach to help you define and progress in your career. I’m terribly discrete.
There’s a lot of content here which just quotes some consultant or trainer, and I
felt like I was missing out by not hearing from the ur-boss, but then the job of
a journalist is to boil this stuff down so we get to read it in reasonable time
scales. All in all, I didn’t mind this book too much, and it’s worth a few pounds to get some new case studies, frameworks vel sim that I might use. The articles are at least mercifully short.