Big Data, Plainly Spoken (aka Numbers Rule Your World)

Fooled by the big bad digital marketing data

My latest piece in HBR (link) deals with the uncomfortable truth that digital marketing is just as wasteful as off-line advertising. This waste is a result of the prevalence of fake data. Because so much money is channeled into digital these days--all the big Web companies are essentially advertising agencies, scammers are everywhere siphoning off the advertisers' dollars. As I demonstrate in the piece, these scammers can make a good living. New technologies like cloud services and programmatic/algorithmic buying have unintended consequences of making the system more opaque and more in-the-moment, abetting the scammers. The industry is ignoring the problem because it benefits from fake traffic and fake clicks. Data analysts are churning out meaningless reports built off of fake numbers.

See the entire article here.

Theranos and ethics

I have to admit I missed the hype about Elizabeth Holmes, and only learned about her from the Wall Street Journal hit job that came out last week (link). It feels like I arrived to the party after the police has shut the fun down!

Holmes had been profiled in all the important places--no doubt many of her interviewers are feeling a bit red-faced. She dropped out of Stanford to start a company that sells a full range of diagnostic tests based on tiny drops of blood. There is purportedly an underlying algorithm behind this product, known as "Edison". I use the word purportedly because few outside the company have seen the algorithm. Theranos said that is a trade secret.

The company is a big hit among the investment community. It is currently valued at $9 billion. Within the medical community, reaction has ranged from skepticism to cautious optimism. The secrecy simply means no expert in the field can judge whether this is vaporware or not. The WSJ report rounds up a number of former employees who claim that Theranos is all sizzle and no substance.

***

Like others, I have no evidence on which to judge whether the purported Theranos technology works. The other allegation by WSJ is that the vast majority of the test results provided by Theranos came from traditional methods, and thus are no different from other existing providers. This shines the light on the issue of business ethics.

Imagine that I am starting NewCo. NewCo is marketing some new invention, say, a new chair design that is better for your back. In fact, the NewCo chair is not better than other ergonomic chairs in the market. However, the chair is not any worse either. It is simply a re-branded chair of one of the chairs already in the market. NewCo sells this chair at a lower price because it has venture-capital money to subsidize the cost. In short, consumers are getting the same product at a lower price while thinking that they are getting a better product.

One point of view is that if this is legal, it is okay. No one forces a consumer to buy this chair; nor does anyone force an investor to put money in NewCo.

But something doesn't feel right here. Just because something may not be illegal, should you do it?

While consumers are not harmed, there are potential losers in this game: (a) the other competitors who lose market share because NewCo is selling its chair below cost, and (b) the future investors who are left holding the bag after the original NewCo investors cash out their investment.

You can of course argue that consumers are losing because they are being deceived. This view says money is not the only consideration.

***

Looking for precedents in the marketplace, I find that our legal system has no consistency in how we deal with these situations.

Tap water is being re-branded and sold to consumers. This is considered legal.

By contrast, one isn't allowed to sell a sugar pill as medicine, even if the pill does no harm, and even if the sugar pill might yield a clinical benefit through the placebo effect.

However, one can sell nutritional supplements which do not have strong science behind them.

For me, these issues cannot be solved in the legal system. We need ethical standards.

***

Since the initial expose, the Journal has published a couple of updates on Theranos:

FDA comes knocking (link)

An interview with Elizabeth Holmes (link)

Walgreens rethinks its relationship with Theranos (link)

The analytics talent gap: a hiring manager's perspective

We all know that data science/business analytics is booming. Lots of companies are creating lots of jobs. However, the barriers to entry into this field are substantial. In an article for rSQUAREedge, I summarized my own experience as a hiring manager.

A data analyst needs intuition about data, which requires practical experience. Certain skills, like web analytics and A/B testing, are not academic research areas, and are not unusually taught at schools. Other skills, like survey research and statistical testing, have theoretical underpinning but the secret is how theories are adapted to the real world.

The key issue is risk. As a hiring manager, it is difficult for me to hire someone who is not exposed to most of the requirements. Sure, I can train the new hires. But that takes a lot of time, a lot of patience, and may not bring the right rewards--such as when the newly-trained employee jumps ship.

In the article, I describe the types of resumes that make me think twice about pulling the trigger. (I advise rsQUAREedge and help put together their training program.) Let me know if you have similar experiences.

 

 

Statbusters: Simple computations mislead, from selfies to colleges

In this week's Statbusters (link), we discuss two recent widely-shared articles, one on deaths while taking selfies, and the other on the gender gap in income among graduates of top-tier universities.

The common element between these two pieces is a reductionist analysis that looks at the correlation between a single variable X and an outcome Y when the outcome Y is affected by a multitude of variables. For example, it is reported that female graduates of Princeton or Harvard or Stanford earns $50,000 less annually than male graduates ten years after graduation. The two groups being compared do not differ just by gender, despite what that statement implies.

Note, however, that there is nothing wrong with the computation. Similarly, the deaths linked to taking selfies indeed outnumber deaths by shark attacks. The problem is that the analysis is misleading. It causes readers to come to the wrong conclusions.

Read the full article here.