This lesson involves Canada, ice, and time…
Ever hear of Wayne Gretsky?
He holds the all-time record for points in professional hockey, scoring 49% MORE points than the person in 2nd place.
How did he do this?
He got really good at anticipating.
Gretsky was incredible because he would skate to where the hockey puck was going to be, not where it was. He had seen enough pucks bounce off boards, pucks shot on the net, and pucks passed around to know where he needed to be to make the biggest impact.
In the last email, I wrote about building a career moat in the data field. To summarize quickly, the most desirable stage of that moat is the trusted advisor, and to get there you need a combination of technical skills and business acumen..
One of the key skills on the business acumen side is the ability to anticipate hidden questions. Let me illustrate with an example.
Suppose a stakeholder (VP, PM, Director, etc) comes to you with a question about website engagement. They’ve even gone as far as to mock up the exact report they want in Excel. All you have to do is fill in the data.
Easy right?
Yes it’s easy, but it puts you on stage 1 (dashboard monkey.) You’re not build a career moat out of that.
The key to moving up is to figure out what they’re REALLY asking.
Here’s a hint: They probably don’t care about website engagement!
There’s a lot of hidden assumptions, questions, hypotheses behind their question. It will behoove your career tremendously to figure out what they are.
You will find out what’s most important to that stakeholder:
What metrics are they looking to drive?
What hypotheses do they have and how do they intend to validate them?
What additional data are available that they don’t know about?
What potential follow up questions might they ask after you deliver the report?
These are extremely valuable to a stakeholder but they’re likely not thinking about them. It’s up to you as a data analyst or scientist to unearth these questions and hypotheses.
It will not be possible to get all this on a first try, and often they won’t even tell you. Not because they’re sinister or anything like that, but they’re likely unaware of their assumptions or even embarrassed about them.
Sometimes you’ll have to become a sleuth and figure this stuff out indirectly.
Next time they ask for a report, you can deliver a richer analysis with more useful recommendations and follow up questions already answered.
As a side effect you also get to learn more about the business.
This is how you gain their trust. This is how you build a moat.
On the data engineering side, when someone is asking for a new data set, you should figure out how they intend to use it. That way you could suggest additional data they may not be aware exists or avoid duplicating existing data.
You’ll gain knowledge over time, but the faster you learn to anticipate these hidden questions the faster you can become a trusted advisor. And that my friend is a moat nobody can take away from you.
SQL Pattern of the Week (SQLPOW)
In this newsletter I’m trying something new.
I’m including a section on SQL Patterns I have noticed or written about. Let me know if you enjoy these and I’ll keep going otherwise I’ll keep the email clean.
This week we’ll talk about granularity and finding duplicates.
Granularity is very important when joining tables. If you don’t use the table’s primary key or you choose the wrong combination of columns and the table has duplicates, your analysis will be flawed and your numbers will be artificially inflated.
The SQL pattern to check the granularity is very simple and you should use it often when exploring new tables or when debugging incorrect numbers. Simply choose the combination of columns you expect to be unique, aggregate them and check the count.
SELECT col1, col2, col3 —> combination of columns expected to be unique
FROM table
GROUP BY 1,2,3
HAVING COUNT(*) > 1
Until next time.