Musings on using data...
Musings on using data...
But fear not, I stuck some of the Depp/Heard shenanigans in here for those wondering what in the hell is #muffingate.
Curious Minds… When do we get to data driven decision making for the rest of us?
Data driven decision making… same tired story decades after inception.
Years ago, I worked with a client that was assessing their organization. They wanted to understand their sales performance. They wanted to understand their inventory. And they wanted to understand their customers.
The company had failed to invest in data analytics but had a solid database of information. Customer information, orders, sales, and servicing, were all available.
I remember sitting down with the sales manager to discuss their customer metrics. The story the data told was all to familiar. Sales agents focused on the wrong customer groups, spending inordinate amount of time on customers whose buying patterns were receding.
At the same time the customers who were posed to grow or who could be nurtured into deeper relationships were being ignored.
No cultivation.
Worse yet, “preferred” customers received the lion’s share (and then some) of time and energy from the business. Some of these “preferred” customers required so much service and hand holding, the business was essentially paying them to be a customer!
”No… that can’t be right” the sales manager told me.
“I’m not sure I get you, the info is pretty clear and actionable today. This customer has to be fired” (Yes, customers can and sometimes should be fired).
“Nah, their a great customer, the data must be wrong. I have a gut feeling about this things.” stated the sales manager
”I don’t know what that means but the data is clean, this customer is costing your org money.”
”Data must be wrong. I know this customer and their solid.”, says the sales manager as he gets up and walks away.
Fast forward to 2021, different clients, different people, same stories.
The businesses that need to be looking at their data the most and making informed decisions aren’t. The businesses that have made the leap and are attempting to use their data get stuck in KPI hell.
Is this familiar to you?
Do you receive a daily notice (could be an email, link to a dashboard, etc) that provides metrics on the business? If you do, go one step further; does that notice provide any actionable intelligence or is it a data dump from some team responsible for data?
Why? What is our obsession for doing a bunch of “math” and then slapping it into some output for maybe accidental consumption by a reader?
Follow this back and forth I had with a client during a review of their processes.
”Hey, can I ask you a question? Do you look at this KPI email?”
”Oh yea, I read the top line.”, states the manager.
”This email is pages long, it won’t even fully load in email you have to download it. Why just the top line?”
”I don’t know, I like to be informed.”
”Informed of what, what action does the top line help you to make?”
”Action?” Says the manager, “No action, I read the top line to see if we sold more yesterday than the day before and then I delete the email.”
No comparables. No context. No visuals. Would we expect anything more then a cursory glance and delete?
Over a decade later. Different companies, different people, and the same issues. So much data and so much invested in “doing math” on the data. But rarely are we asking if we even care or should care about what the data is telling us.
Embrace a better way
Reports and data tables are easy. Their easy because they are comfortable. We know how to make them. Others know how to consume them. It takes zero effort to slap some math on a report, dress it up in the company’s brand and boom, instant “thing”.
Yet, if you aren’t starting with asking questions; really trying to think around the problem you are trying to solve then reports like that provide surface information at best.
Ask hard questions. Questions that weed out report writers and leave you with true analysts that can critically think through problems and devise solutions that make meaningful impact on a business.
Did the sales of a widget go up? Who cares. Why did the sales of a widget go up? How do I make it go up more? Can it be recreated? What meaningful change can we make?
Start with asking questions, hard penetrating questions and invest real time into answering them. Silly outputs like did sales go up will take care of themselves after that.
Let me know, is your business getting stuck in this same trap?
Lies, Damned Lies & Statistics
This week’s topic falls into the kinda stats, kinda lies area. For those who enjoy stats, you are probably familiar with the R^2 term. This comes about when evaluating regressions, a very popular methodology in stats. R^2 is very popular in part due to its core tenant, it relays the level of explanation of a data set’s variation.
Huh you ask?
Think of it this way. You have some set of events that are occurring. You create a regression model to try to explain those events. R^2 helps you determine how well the model explains the events.
It’s interesting because regression has a variety of different tests for evaluating models. R^2 is likely the most popular and is easy in the corporate world. Again, easy to explain, easy to relate the core tenant. In recent discussions this topic came up because something that schools are really pushing hard in the analytics field is ensuring tests and models are significant.
Sadly, significance in this situation is aimed at the math. Not at the business. Now, I’m not writing that statistical significance is worthless or shouldn’t be done.
What I am writing is that if you can’t answer if the regression, the model, helps the business what does it matter if the R^2 value is high or there are statistically significant features.
I’ve had conversations with analysts that revolve entirely around the math. Whether that person could say if something was good for the business was a foreign concept. Zero help to the business.
Its time that analysts get outfitted with not only the ability to create a model but also skills in education and visual story telling. Teach your teams to do better, capitalize on finding data people who not only love working with data, but they understand how to tell the data’s story.
A look back…
So this section was originally a PSA. I wrote this back in 2021. Different times.. and yet not so much. I’ll tell ya one thing, diesel wasn’t over 5 bucks a gallon yet.
I thought about rewritting this but I have some upcoming deep dives into general health questions and didn’t think doing the double dip was worthwhile. However, I also felt keeping this in was interesting because it is a snapshot in time but the trends hold true.
**** original writing below ****
I thought it would be an interesting exercise to look at some data related to the push to mask kids due to the COVID-19 pandemic and its variants.
It probably doesn’t need to be written, but humans tend to be amazingly bad at assessing risk. We tend to use various biases to rationalize risky shit we do and we use other biases to fear monger things that aren’t risky.
So, here is the data in all its gory… (not a typo)
There are approximately 209,128,094 people aged 18 or older in the United States.
Which means there are approximately 72, 293,812 people under the age of 18.
According to the CDC, the demographic for 18 and older saw a death amount of 519,775 deaths due to Covid-19. For the sake of the written discussion, let’s say that is an accurate number. That is a rate of .249%. Scary stuff to be sure.
According to the CDC, 168 million plus people have been vaccinated. Now, we know this isn’t true but again for discussion, it wasn’t authorized for people below the age of 18 for some time and below 12 still. So we can safely assume that the 168,000,000 are 18 and older. Again, according to the CDC, there have been 1,829 deaths for people vaccinated, breakthrough deaths. That is a rate of .00109%.
The take away so far? Vaccines are freakin awesome (Moderna, Phizer, J&J kick ass). Go America!
Now, let’s do the math for under 18. Again, we have 72,293,812 people under the age of 18. According to the CDC, there have been 474 deaths due to Covid. That is a rate of .000656%.
So, if you are under the age of 18, you are less likely to die from Covid than if you are over 18 AND you’ve had the vaccine (which again is amazingly effective).
Risk assessment time! Here are some of the things that are more dangerous to people under the age of 18 than Covid-19.
Water - Drowning death rate: .000892%
Toys - Mechanical Suffocation rate: .00161%
Cars - Accident death rate: .00177%
Cheeseburgers - Heart Disease death rate: .000829%
So, consider before mandating or supporting the mandating of things like masking for kids are you also in favor of banning pools, small toys, cars and McDonalds. If masking kids makes sense, then the others above should as well.
And if we thinking of banning those things, these are scary times indeed.
******
Now I know that Covid-19 is a wildly polarizing topic. I’ve my theories on why, but we’ll leave that for another time. If you have thoughts, updated numbers or just want to rage (or agree, I’ll take that too), hit me up in the comments section.
Something not data for the non data people!
So the Johnny Depp, Amber Heard trial…. couldn’t write a romcom breakup better if I wanted to. Some highlights:
Hollywood people are absolute crazy town.
If you are wondering if your marriage or relationship is on the rocks… Consider if you would literally poop in your partner’s bed.
There are smart people and dumb people in every profession. If you don’t believe that, prepare to have your mind blown by #muffingate.
Can you imagine being the lawyer who gets remembered for #muffingate.
Shameless Plug - Fabled Fudge (www.fabledfudge.com) is getting ready to launch! That’s right, the delicious fudge that comes to some of your doors at Christmas will now be available year round with some amazing new flavors to come.
I am super excited to be able to make this a reality. Have had a lot of fun figuring out the logistics of this, testing new recipes and making something near and dear to the Wilcox Family.
Sign up for the mailing list over there, be the first to get the goods!
As awlways, If you think someone would enjoy this newsletter, please pass it on. Reminder, the newsletter will be moving to a new home. www.thepenumbral.com Be on the lookout for the new look and feel.
Jason, another spot on posting brother , and I catch the 'mirror' you are holding up to us ! Keep em coming man.
As our team is working to roll out some pretty intensive PowerBI reports, we have to keep reminding the end users that if you don't go into these reports with specific questions (like "what is our long term profitability on ITT deals ?", or "why is our service revenue down 35% from last year?") , then you will get overwhelmed with a lot of drill down data etc. Most of us are not trained in statistics or data analysis, so it would be rare us to open a report and have an epiphany moment from the data. Thanks for pushing us all to better understand that "just because you can measure something...it doesn't mean you should measure everything " ! **kiss your sister/brother kinda thing right?? :-)
Hit the nail on the head Mike! If you are going into a situation producing giant tables of numbers, consumers of that info are quickly going to start ignoring it.
Worthwhile data science is about answering questions asked by the business units not outputting tables of data that can't even be contained within an email.