Thursday, May 23, 2013

NFL Offensive Contribution Analysis - 1993-2012

It started with me looking over my wife's shoulder to see what she was up to as we sat on the couch.  She was busy on her laptop, and clearly wasn't giving Game of Thrones her undivided attention (but then, to her, it's just 'That Rape Show with Dragons')

But I noticed she was working with some really neat looking charts, and I asked her what she was using to visualize her data (she works in marketing analytics, so data is kind of her thing).

She explained it was a dashboard program called Tableau.

I filed that away, thinking it might come in handy for either my own professional or personal use.

Then, as I started to delve deeper into some NFL analysis, I finally found a good opportunity to start playing around with it.

My brother runs an Eagles blog,, and recently published a cool area chart showing the major contributors to the Eagles offense over the last few years.  It was a lot of fun to look at, mostly because the chart is a great illustration of the various eras the Eagles went through (although, it overlooked the Freddie Mitchell era)

That got me thinking, could I build something in Tableau that would allow anyone to look at that for their team and their favorite years???

After a couple hours fiddling around, I think I finally got the hang of it.

I pulled together every player season from Pro Football Reference from 1993, for any player who accumulated over 200 yards from scrimmage.  Now this means that it's mostly running backs, wide receivers, and tight ends.  I could have included total yards of offense as the metric, but it would have turned into a chart of quarterbacks and frankly, I thought this would be more fun.  After the 'fun' of teaching myself a little bit about how Tableau works, I think I got a pretty good start.

The dashboard below illustrates the yards from scrimmage for offensive skill players for each season from 1993 to the present.  Below the chart itself, you can select a specific team you want to see illustrated, and use the slider bars to see certain years (I apologize for the formatting of the slider bar numbers, that's something I haven't figured out yet).

But I thought it was a pretty fun way to look at a football team. Let me know your thoughts, and if people like it, maybe I can make some more. If I do, I'll be sure to publish the links on Twitter 

Thursday, May 9, 2013

TV Everywhere isn't an Authentication Problem, its a Use Case Problem

John Ourand's recent column in Sports Business Journal (which I unfortunately didn't pick up in my Twitter feed until today) talks about the disappointing results of TV Everywhere.

TV Everywhere has been the big push from cable companies for years.  A system of subscriber verification which would enable paying customers to access their programming outside of the TV and set-top box.

With continuing penetration of smartphones, and the expansion of tablet devices (which I don't think were even top-of-mind for execs when they launched TV Everywhere), execs figured this would be a big way to keep customers from cutting cords.

But customers aren't using it.  And while Ourand doesn't have hard data to back this up, the quotes from industry executives speak for themselves.  There appears to be broad acknowledgement that TV Everywhere isn't really big, anywhere.

But the execs in Ourand's article attribute it to one specific aspect of the interface, subscription verification, and that's where I disagree.  I really don't think that's the issue.

Don't get me wrong, user authentication for a cable subscription is incredibly painful.  Just ask anyone who tries to use HBO Go, which has a similar process.  Find your cable provider, click, and figure out what your login credentials are.  I have Comcast, and never use my login information outside of every six months when I have to call their customer retention group to renegotiate my pricing.  So it's a huge pain, agreed.

But I feel like that's not the biggest problem.  To me, there are two issues that keep me from watching TV channels on my devices:

1 - Horrible wireless network speeds
2 - No clear use case

The first one is obvious.  Whenever I walk home from my office, I try and stream audio on my iphone, typically from Stitcher or Pandora or Songza or whatever.  And it never, EVER, works.

That's because thousands of other people are doing the same thing.

Now I don't have the newest iphone, so I'm not on LTE, and maybe that would make it easier.  But the fact remains that I'm constantly in densely populated areas with everyone trying to stream content at once. I love when people send me funny youtube videos, and they almost always take forever to load.  Now you want me to watch streaming television programming???  Which is likely 10x as long as your average internet video?

That brings me to the second issue.  What's the use case for TV Everywhere?

It let's me watch TV wherever I am.  OK.  A fine theory.  Follow-up question - where do I want to be watching TV???

At home?  Sure, but I have a TV for that...a couple of them.  In all major rooms of the house.

But I guess I'm being obtuse.  TV Everywhere is a mobile solution, so naturally it's not going to be primarily used at home.  Fine, so where do I spend my time apart from my house?

At work?  Well, my clients probably wouldn't like it

In the car? Ha, we already have a Driving While Texting problem, I'm not going to compound that with a Driving While Watching SportsCenter problem

When I'm out with friends?  It's not like we're going to gather around to watch Homeland on my ipad

Tell me a place you would absolutely use TV Everywhere to stream live TV?

Maybe you would want to do it at an airport waiting for a flight that's delayed.  Maybe.

But I don't see a clear use case.  At all.  Now, if there were TV you absolutely had to see live, I could imagine it being used.  But DVRs have solved that problem.  No traditional content needs to be consumed as its broadcast anymore, and with massive HDTVs relatively affordable, who would choose to watch Game of Thrones on their iphone when they can wait until they get home?

The time I spend on my devices (iphone, ipad) outside of my house is marked by one major theme.  Whether I'm waiting in line, on the bus, in the elevator, it's always for brief periods.  I've become conditioned to use my devices to fit the time period and attention span required.  Checking email, checking Twitter, checking news stories...those all take less than 30 seconds.  I've never had the sudden urge to immerse myself in Westeros while I'm waiting for a burrito at Chipotle.

To me, that's the real reason TV Everywhere isn't getting used.

There's just no real good place to use it, anywhere.

Thursday, May 2, 2013

Why Self-Checkout Doesn't Work

There was an article in today's Wall Street Journal about Kroger, the major grocery chain, and their efforts to reduce waiting time at checkout.

Supermarket giant Kroger Co. is winning the war against lengthy checkout lines with a powerful weapon: infrared cameras long used by the military and law-enforcement to track people.

These cameras, which detect body heat, sit at the entrances and above cash registers at most of Kroger's roughly 2,400 stores. Paired with in-house software that determines the number of lanes that need to be open, the technology has reduced the customer's average wait time to 26 seconds. That compares with an average of four minutes before Kroger began installing the cameras in 2010.

Now, the article goes on to describe how Kroger uses that data, which tracks how customers move through the store, to predict how much staffing they'll need at the registers in half-hour increments.

I took particular interest in the article because I really really don't like waiting in lines.  It's the equivalent of a human traffic jam that'll take some time to clear, increasing the cost (in the form of my time) all so I can have a sandwich or buy toothpaste or get a question answered.

It's not all lines inherently bother me, necessarily.  If I go to a Chipotle at 12:30, I know it'll have a massive line and I know about how long it takes.  Plus, it's clear from the structure of the operations and the attitude of the employees that they're trying to get me through as fast as they humanly can (I'll bet that's even tracked back at HQ).

So, lines are OK if they meet two primary conditions:

 - The wait time is expected/known
 - The business/employees are actively working to make my wait shorter

Of course, for anyone who has shopped at a grocery store, you know that neither of those are necessarily true.

The second condition is rarely seen in practice.  Contrast the furious burrito-making at Chipotle (or the equally furious sandwich making at Subway) with most grocery store checkouts.  It's not always the fastest pace (although some chains have worked to track checkout clerk speed, but I think results have been mixed)

The absence of the first condition is a result of poor planning and inertia, at least with most supermarkets.  You usually have no idea how long you'll wait.

Any recent shopper knows the drill, walk around the checkout area, scanning the aisles for the best slot.  Who has a relatively empty cart?  Which line has the shortest queue?  Which clerk looks like the give the biggest sh*t?

Often these estimates are met with futility.  Like when the woman buying a solitary candy bar whips out a change purse and starts counting (Aside: If you look to make exact change for a purchase, please stop.  You are placing a larger inconvenience on everyone behind you, all so you can get the exact $1.09 for your Three Musketeers.  No one's life is improved by your choice)

The problem is a high degree of variation in service time.  Some customers will unpack their basket as fast as possible, pay with a credit card, and engage in zero extraneous communication.  Others seem to WANT to take as much time as possible.

This creates a problem when you have a dozen discrete lines like a traditional supermarket.  It creates a system that's inherently unfair.  Someone in one line has to wait much longer than someone in another.

Now, at one point supermarkets were pretty sure they figured this out.


The purported saving grace for the busy shopper, the self-checkout was designed to free us all of the tyranny of the lines (or, more accurately, free the supermarkets from the tyranny of hiring and paying employees)

I have to admit, I was smitten when they first rolled out the technology.  You see, having worked in a traditional supermarket in high school, I had a lot of experience bagging groceries.  And I knew that if I could just do it myself, I'd be all that much faster.

Of course, in reality, nothing could be further from the truth.

There are a series of practical realities that preclude self-checkout from ever being a reasonable solution (in its current form)

1 - Random weight items: Self-checkout is simple when the packaging is standardized, you just scan the bar code.  But with any random weight items (i.e., produce), they require a specific ID code, typically 4-5 alphanumeric characters.  Well, if you're an experienced clerk and know them all by heart, that's fine.  But for the rest of us, we're left to stumble through an alphabetical picture book that makes you feel about as smart as a 4 year old.  Find the apple.  Now find the onion.  Now find the broccoli.  It's literally a kindergarten exercise, but one that takes way more time than it otherwise would.

2 - Anti-theft nonsense: Of course, when you allow for 'self' checkout, you're implicitly allowing customers the opportunity the steal from you.  Like a slightly muted version of the honor system, you're giving customers the responsibility to manage their own purchases.  There's a social contract element to that, only the ingenious minds behind the self-checkout couldn't go that far, and put weight scanners in the bagging area, surely so they could ensure that one scan equals one package in the bagging area. 

Of course, whatever testing they had in place didn't do the job.  I have yet to go through a self-checkout that trusts me.  After I scan and bag my item, the machine will request me to put the item in the bagging area.  Caught in a weird catch 22, because I've already bagged it, I'll look confused for a second.  The machine asks again, seemingly growing impatient.  But what can I do machine?  I've already bagged it!!!  By then it's too late, the machine has signaled for an attendant (this is not a machine with compassion.  If it were on the parole board Red would still be wasting away at Shawshank)

This isn't to say that customers don't steal, here's proof.

They just do it in better ways!

3 - Bagging: Self-checkout also means you have to bag groceries yourself.  That's obvious to many people, but I've seen a number of folks roll through self-checkout without really understanding that they'll have to then corral all their stuff and get out the door.  Even for a bagging savant like me, it's no better than a push with the employee equivalent.

4 - Other absurd rules: To buy alcohol, I obviously can't use self-checkout, unless I have a 17 year old grocery clerk come over and verify the fact that I should've stopped buying malt beverages 10+ years ago

Add up all the ridiculous delays and poorly designed mechanisms, and you have a recipe for a LONGER wait time.  Some stores have caught on and are ripping out their self-checkout aisles (I'm sure others found some labor savings in making things more inconvenient for customers, and have stuck with it).

But if a chain really wants to reduce lines, it's honestly not all that complicated.

You need a mechanism that takes the key wait time driver (service time variance), and remove it as a driver.

How do you do that???

By having one line!

Just like airports, which have one line leading to multiple service desks, so too should supermarkets be designed.

Some places I've shopped at (Walgreens and Nordstrom Rack here in Chicago), have actually done this.  And it's much much better.  The one line is longer, but it moves much faster, and I never have to kick myself for getting in line behind someone who barely knows what planet their on.

All supermarkets should shift to this, right???

Well, sure, in a perfect world.  But unfortunately, that would take some investment in moving all the fixtures around.  Supermarkets would also lose out on a lot of valuable high-margin checkout aisle shelves.

And since there aren't really a ton of new supermarkets coming in to compete, there's a bit on an incentive issue...

So it might be a long time before I can comfortably take my place in one giant line at my local supermarket.  But then, I could just order the stuff online...