Facebook MAU Fraud – Where’s the scrutiny???

Bottom line – Facebook’s MAU (monthly active user) definition criteria does not require going to facebook.com…just hitting like on an article or uploading a photo via mobile and never viewing ads meets the critera!

EASY METRICS: Some metrics are pretty straightforward: 85 points in a basketball game, 100 dollars, 42 nickels, 6 feet tall without shoes. They need no additional layer of information to interpret.

FUZZY METRICS: Some are a bit more complicated: 3 inches of rain (may not be uniformly distributed), ‘up to 32 miles per gallon’, 30% gross margin, avg savings of $336 by switching to xyz insurance company. Without additional context or a distribution of the metric under varying conditions…things are less clear-cut. Here’s an example:

insurance savings interpretation

View full article

Posted under Analytics Thought of Day, Facebook by Jeffrey James on 20th May, 2012

Leave a comment
Send to a friend

Warren Buffett on FB IPO – Analytics Thought of the Day

Warren Buffett on whether he’ll be participating in FB’s offering:

“We never buy into an offering” Buffett told CNN at Berkshire’s annual meeting in Omaha.

“The idea that something coming out…that’s being offered with significant commissions, all kinds of publicity, the seller electing the time to sell, is going to be the best single investment that I can make in the world among thousands of choices is mathematically impossible,” said Buffett.

buffett facebook valuation

View full article

Posted under Analytics Thought of Day, Finance, Risk by Jeffrey James on 6th May, 2012

1 Comment
Send to a friend

Google Analytics Advanced Segments vs SiteCatalyst Segmentation (data warehouse)

WARNING – REAL CONTENT AHEAD:

This might get a little complicated!  A segment is just a set pulled from the overall data. It may include everything or nothing but most likely contains something between the two. Talk about abstract!

visitor visit pageview segment example

View full article

Posted under Google Analytics, Google Analytics vs Omniture SiteCatalyst, Omniture SiteCatalyst by Jeffrey James on 6th May, 2012

1 Comment
Send to a friend

Critical Thinking and Messaging

First I’ll say that I have no problems with Chevrolet.

max towing capacity

View full article

Posted under Analytics Thought of Day by Jeffrey James on 3rd May, 2012

2 Comments
Send to a friend

iPad 2x more likely to be dropped in 2nd year? iPad 2 2x as likely to be dropped…

Confusing title eh?

Carrying around an iPad is a fairly consistent process right? I mean, you have your iPad…perhaps in a case, perhaps not. You bring it to work, to grab coffee or go wherever. Would you agree that the odds of dropping it are fairly consistent over time? Do you think the odds of you dropping it over time are higher, or lower? Seems like it wouldn’t make much of  a difference, at least to me.

Take  coin coss – let’s say you’ve had heads 9 times in a row and you had to bet on the next result. I’d bet on tails due to the recent assymmetry in the distribution but anyone who has studied probability knows that there’s a constant 50% of a head or tail, assuming a fair coin and a vigorous toss.

ipad drop probability

According to electronics aftermarket warranty provider ‘SquareTrade‘ – the accident claim rate doubles in the 2nd year of ownership. Think about that – given the condition that someone has owned their device for more than 1 year, the odds of them dropping the device doubles. Now – it would be idiotic to try and say that the longer you have something, the greater the odds of dropping it are…or would it? What are some obvious and less obvious factors that may drive this reality:

  1. After time people become more careless – this doesn’t account for the doubling of the accident rate for the iPad 2 over the iPad 1  in similar years however.
  2. Could people be giving the iPad to others as they purchase new ones? Given the recipient has no ‘skin in the game’, are they more careless?
  3. Are more younger people getting iPads now, explaining the doubling of accident rate between versions – said differently, teens don’t take good care of their stuff?
  4. The volume of iPad2′s sold relative to the original is much much higher and they appealed to a broader market. The broader the market, the more likely you’ll find the device in the hands of ‘less careful people’?
  5. The device drops multiple times but it takes longer than a year to break the camel’s back, so to speak?
  6. To point 5. the iPad 2′s display is ‘weaker’? Doubtful but worth considering.
  7. People get tired of their scoffed up screen so they destroy the device on purpose – they did purchase a warranty after all. Calls into question whether the accident rates for insured and uninsured iPads are equal…I suspect not.
  8. Borrowing from point 7, because they have the warranty they are more careless over time. Especially after the screen has become used up a bit.
I believe point 8 is likely the most true, perhaps with a tip of the hat to 5 as well – but without some deep introspection you could easily misconstrue what this statistic actually ‘means’. This is also true in business analytics. In any situation there are several variables or conditions which are basically ‘silent’. Meaning they don’t have a line on your chart, nor do they really belong on your P&L report because they are hidden by nature. Only through careful reflection can you sort of determine precisely what it is you cannot see…and your degree of precision will definitely vary!

Next time – Kindle Fire burn rates,

Jeff

View full article

Posted under Analytics Thought of Day by Jeffrey James on 28th April, 2012

2 Comments
Send to a friend

Could data ever be ‘that good’?

Imagine a world where you could present your business and marketing data for the past 24-36 months, along with your expected results over the next 12 months; then receive a quote for insurance that will pay the shortfall should you miss your forecast?

View full article

Posted under Analytics Thought of Day, Data Quality by Jeffrey James on 27th April, 2012

5 Comments
Send to a friend

Are regression models useful?

Lots of businesses try using regression models. The basic idea is great – use the value of some variable(s) which you may be able to control to predict the value of another variable [most likely a metric] for which you’d like to optimize (probably get as high as possible).

analytics - regression model scatter plot

View full article

Posted under Statistics by Jeffrey James on 25th April, 2012

5 Comments
Send to a friend

Metrics for their own sake (analytics thought of the day)

Journalism at its finest!

View full article

Posted under Analytics Thought of Day, Finance by Jeffrey James on 23rd April, 2012

4 Comments
Send to a friend

Data Visualization – A fun evolution

With a title like that I bet you’re expecting something pretty crazy! Not really…just seeing how basic visualization concepts can rapidly evolve into highly descriptive multidimensional presentations.

Cartesian Plane – “plane” old X and Y (horizontal, typically time and vertical, typically a measure of some sort)

Quite a bit can be done here…continuous functions are indeed integrable. Business data generally doesn’t trespass left of the origin, especially dealing with a time series (negative months?).

View full article

Posted under Data Visualization by Jeffrey James on 16th March, 2012

2 Comments
Send to a friend

Website Personalization Strategies for 2012

Imagine if you could see all the tabs your visitor had open while browsing your site. You’d see some names that would make you cringe:

tabbed browsing

Yes my friends, your site is NOT the only one open. Half the ‘time on page’ metric is really time on the page of another website. Sharing is caring, no? What’s the next level of differentiation after price and basic usability?

View full article

Posted under Personalization by Jeffrey James on 11th February, 2012

2 Comments
Send to a friend

Page 1 of 612345
<