Complex Customer Cohort Retention Using PRODUCTX

Post by Rob Collie

First, a Few Quick Updates…

1) Just a reminder that enrollments are open for the live, in-person two-day classes taught by me (Rob) in Indianapolis and Washington DC (in October and November, respectively.)  Sign up now – those dates will be upon us sooner than we all realize!

2) The pre-order for DAX 2nd Edition is going Great!  Thanks to everyone who has pre-ordered and/or contributed.  PRE ORDER HERE if you have not already, and don’t forget the exclusive swag rewards like the sticker, poster, and t-shirt.

3) Ola Freaking Rosling is now using Power BI, and threw us a “shout-out!”  OMG, check out this tweet.  I nearly peed myself:


If you don’t recognize the name, his dad is Hans Rosling, the speaker in the first TED talk I ever watched:


Click Image to View the Talk

…In which he did the animated bubble chart demo that I’m 100% positive was the inspiration for Amir Netz adding animated bubble charts in the next two product releases Smile

Here’s Ola and Hans co-presenting on the state of world health and social justice, via data of course:


Again, Click to View the Talk

A celebrity in that Google-y, Apple-y, Silicon Valley-y “doing good with data in the public interest” space using Power BI and saying that Power BI is awesome, and that he’d previously been scared off by all the anti-MS propaganda…  well that is simply awesome.

But I won’t lie – that he gave US a mention in that tweet – I *do* find that to be even cooler.  I’m human.  I can’t help it.  I smile every time I think about it.

Slaying the “White Whale” of Variable-Rate Forecasting with PRODUCTX


“Arrrrr!  We Be Meeting Again!”

Read the Rest

Counting Overlapping/Shared Twitter, Facebook, Instagram, etc. Followers

Post by Rob Collie

From Last Week’s Client Work

Last week a client asked us to solve a somewhat unusual problem:  given any two lists of Twitter followers, tell us how many followers “overlap” between the two lists.

Two Lists of Twitter Followers:  How Do We Find the Overlap Using Power Pivot / Power BI / DAX?

How Many of Han Solo’s Followers Also Follow Leia Organa, and Vice Versa?
(Randomly-generated Twitter handles are funny.  I particularly like “@Gommo” and “@Xxfok”)

Loading the Data:  Using Power Query

Let’s use Power Query to perform the import this time, both because we’re using PQ a lot more around here now that we have Power Update, and because we’re gonna need PQ for the more complex steps later.

Note that all of the steps below are performed using Excel 2013.  (I find Power Query to be a bit too clumsy in Excel 2010.)

Power Query, aka Power BI Data Import

Importing from a Table Using Power Query:  Step 1
(Unchecked “has headers” because of the “Han Solo’s Followers” Row)

Read the Rest

Net Promoter Score: Fiendishly Simple in PowerPivot! (Caution: Post Contains 26 Movie Quotes)


Net Promoter Score in Power Pivot

Net Promoter Scores Are Fiendishly Simple to Calculate in Power Pivot

What is “Net Promoter Score?”

Fundamentally, it’s a measure of how many of your customers love you, minus how many of them dislike you.  Hence the name – Net Promoter Score.

WARNING:  I am personally no expert here.  I am doing my usual thing:  take a small amount of knowledge and wield it like a battle axe.  I was helping a client today (Monday) with this, and am writing about it a mere three hours later.  But I figure there are lots of people out there who need to do this sort of thing, and THEY get what it all means.  So allow me to share how EASY these calcs are in Power Pivot. describes NPS as:

Read the Rest

Modeling Viral and Marketing Growth, Part 3 of 3

Why am I doing this in PowerPivot?  Primarily as a challenge.

This is a question I should have answered before I even started down this road.

To be honest, I did it primarily as a challenge – to stretch my brain a little bit.  If I were faced with this exact same task in my daily work, undoubtedly I would just use normal Excel formulas.  In some ways, this modeling exercise has been a deliberate misuse of PowerPivot.  A handful of parameters with no source data whatsoever – this is NOT what the PowerPivot engine was built for, which explains why the PowerPivot solution is actually significantly more difficult than the Excel solution.

“So you’ve been deliberately wasting our time??”

No, I do think there is real value in this exercise, for two reasons:

  1. Brain-stretching with new techniques always comes in handy later.  For instance, on the first post Sergey commented that he’d been thinking about loan amortization measures and this could be applied to that.
  2. I can see this technique being added, as a supplement, to a broader PowerPivot model.  For instance, a model containing lots of real customer data over time, and then a [Projected Customers] measure that forecasts future customer populations based on various assumptions and/or marketing investments.

So with that in mind, here it is:  the final installment of viral/marketing modeling in PowerPivot.

Read the Rest

Modeling Viral and Marketing Growth, Part Two

Picking up from last week’s post, the first thing I want to show is that I kinda cheated last time.  To see what I mean, let’s look at Rahul’s original chart:

Viral Marketing Growth in PowerPivot:  Customers Flatten Out Over Time

In Rahul’s Viral Model, Total Customers “Goes Flat” Quickly

In Rahul’s model, if we start With 5,000 initial customers and a viral factor of 0.2, we end up with 6,250 customers and we never get any more!

But in my model from last week, if I use 5,000 and 0.2, customers keep piling up exponentially:

Exponential Ongoing Viral Growth in PowerPivot

In My Model from Last Week, Customers Never Go Flat –
They Just Keep Growing Exponentially

So why the difference?

Read the Rest

Modeling Viral Growth and Marketing in PowerPivot

A Tale of Two Charts

Let’s say you operate a business that relies heavily on “word of mouth” – customers recommending your product/service to their friends and colleagues. Or at least, you THINK it relies heavily on that sort of thing.

You need to decide how much to spend on traditional advertising – to supplement the social/viral marketing that your customers do on your behalf.  Take a look at each of these two charts – the captions for each attempt to capture the knee-jerk conclusions you might draw:

Modeling Viral Growth versus Traditional Direct Advertising in PowerPivot

“Advertising?  We Don’t Need No Stinking Advertising!
That is SO Yesterday!  We’re Viral Baby!”

Modeling Viral Growth versus Traditional Direct Advertising in PowerPivot

“All These Youngsters and Their ‘Viral This’ and ‘Social Media That’ – That’s All Just Fancy Excuses to Be Lazy – You Clearly Need to BRING Your Message to the Customer”

If chart 1 reflected reality, you may opt to spend very little on traditional advertising.  But in a chart 2 world, you’d be silly to rely on viral growth.  But which one (if either of them) describes your situation?

Back in October, Rahul Vohra (CEO of Rapportive) wrote a two-part blog series on this topic, posted here on LinkedIn.  I took a note, at the time, to revisit his work and “convert” it to PowerPivot.

It’s a very different kind of problem from what I normally do in PowerPivot – this isn’t about analyzing data I already have, but about calculating future outcomes based on a handful of parameters.  And that leads to some different kinds of thinking, as you will see.


Read the Rest