Category Archives: Data

The Marketer’s Case Against Cambridge Analytica

By now, you’ve probably read about how digital analytics firm Cambridge Analytica acquired Facebook profile information from about 50 million people without their express permission.  Not to put too fine a point on it, that’s stealing.  And marketers should be outraged about it.

Since Cambridge Analytica used the information to benefit Donald Trump’s Presidential campaign, the news has taken on a political cast.  For once, I will not get political.  Nevertheless, marketers should recognize this news as a teachable moment about why data privacy matters.

At first read, it may seem hard to understand what happened.  Here’s how it worked, according to the Times:

Researchers there [a center at Cambridge University that worked for Cambridge Analytica] had developed a technique to map personality traits based on what people had liked on Facebook. The researchers paid users small sums to take a personality quiz and download an app, which would scrape some private information from their profiles and those of their friends, activity that Facebook permitted at the time.

In other words, the researchers gave a few people–270,000–some money to get access to their information.  They used those 270,000 to scrape profile information from a total of 50 million people, who did not give any permission whatsoever.

Since Cambridge Analytica scraped the information unseen and their clients used the information to power opaque campaigns in social media, it may seem harmless.  After all, no one got a credit card bill with thousands of dollars in fraudulent charges, which happens when hackers breach stores and banks.

It comes down to fairness.

Specifically, I think of something I learned years ago when I had a timeshare company as a client.  I spoke with owners (i.e. the people who owned shares, not the people who owned the properties), who genuinely liked their vacation ownership (per the preferred term).  They advocated timeshare properties to their friends.  And that’s where the trouble started.

My client offered shareowners a bonus for referrals, a common industry practice.  I don’t recall whether they paid the bonus for names and phone numbers alone or based on actual sales, nor do I recall the amount.  What I do recall is the satisfaction one shareowner felt when he told me that he sent my client his church’s congregational phone directory for them to mine for prospects.

“Friends, heaven is beautiful.  But have you seen Myrtle Beach?”

In my mind, the situation looks the same as Cambridge Analytica’s.  One person got paid, but he compromised hundreds of others who a) did not give permission and b) probably didn’t know why they got calls from a timeshare company.  Clearly, these kind of tactics will erode trust in and favorability for the brand.

Certainly, three minutes trying to get a timeshare salesman off the phone does not rise to the level of damage from a campaign to manipulate an election.  However, in terms of trying to show how marketers should not try to gather information, I think it fits the bill.

Making the Case for Platforms (Warning: Explicit Math)

In some recent posts, I’ve discussed the values of platforms over campaigns.   For those of you just joining us, platforms, or brand-owned spaces that foster long-term engagement with customers, offer an economical alternative to acquiring and re-acquiring customers via Facebook and Google.

Today, I’d like to discuss the economics.  There will be math.  I am not above putting pictures of puppies in this post to keep you engaged.

I will stop at nothing to get you to pay attention to numbers.

The basic equation shouldn’t hurt; you need to compare costs to build and maintain the platform against expected customer value over a relevant time period.  I have no hard-and-fast figures for any of the above.  As always, the devil is in the details.

Costs to build and maintain

Costs fall into two categories: costs to create the platform and costs to drive consumers to it.  Creative costs vary as widely as the forms they take.  These costs depend mostly on what the marketer intends the platform to do, which in turn depends on what’s right for the audience.  Figuring out platform functionality–and hence creative costs–will probably take up the lion’s share of strategy development time due to the open-ended nature of building a platform.

Driving customers to the platform, aka acquisition, also represents a substantial cost.  Yes, despite all I’ve written about turning away from campaigns, I will now discuss why platforms require campaigns.

Puppy reminds you to think about a cost per action model!

A platform won’t grow customers all by itself.  Just as with any other marketing tactic, marketers need to make sure their audience knows about it for it to succeed.  That said, bear in mind that the platform will have lower acquisition costs than, say, a campaign designed to sell something.  A relevant platform offers something useful or entertaining to the consumer, something of immediate value.  A timely article or a fun mini-game will have broader interest than a straight sell.

Expected customer value

Acquisition costs nothwithstanding, platforms ultimately must drive some measurable value such as sales.  Direct brands and retailers have no problem here since they can usually connect a visitor to a sale easily.  Other brands in sectors such as consumer packaged goods, automotive or consumer durables have some more math to do, especially in terms of attribution.  The same goes for direct or B2B brands that don’t rely on digital channels to close the sale.

In short, how the hell do you correlate activity on a cleaning tips app to sales of a washing machine?

It’s not “attribute,” it’s “attri-cute!”

As Bob Dylan would have said had he gone into marketing, “the answer my friend, is proxies.”  Well, he might have said it, at any rate.  Marketers need to find good proxies for purchase behavior.  Let’s use the cleaning tips app and imagine our client is GE appliances (disclosure: they were a client of mine before GE sold them off).  GE appliances could push a coupon through the app or encourage new purchasers to register their new appliances via the app.  From this proxy, GE appliances could use the cost of the appliance or some derivation thereof as the yardstick for value.

Relevant time period

Marketers can’t measure a platform’s success in the same relatively short time period that they might use for a typical campaign.  Platforms engage consumers at different points of their journey, so results may not happen in the near-instant time frames associated with digital campaigns.

As a starting point, the time period for measuring platform success should correspond to the customer journey in some way.  On a recent platform project for an automotive brand, for instance, I used three years as a period of measurement because three years represents a typical (if short) ownership period for a car.  For the appliance example above, ownership periods represent too long a period to wait; people hang onto major appliances for more than a decade!  Instead, it might help to look at a length of time related to the purchase cycle.  A typical CPG, on the other hand, has the opposite problem.  People replenish their pantries and supply closets weekly.  As a result, the measurement period might represent a typical timeline for a customer to go from new customer to brand-loyal customer.

That wasn’t so bad, was it?  Now, let’s talk long-term lease depreciation.

I’ve sketched out the math for platforms in very broad terms.  Hopefully, you can use this math as a framework for evaluating ideas that will allow you to break your brand’s dependence on the digital duopoly.  If not, I hope you liked the puppy pictures.

Data. And Research. And Fried Chicken.

When it comes down to it, marketing strategists have only three tools at their disposal: data, research and their own opinions.  Make that two things; opinions don’t count because everyone else has them, too.  Strangely, most strategists willingly ignore 50% of the tools remaining and focus on just research or just data.  I’ve learned that not only do research and data strengthen each other, they also cover each other’s blind spots.  If you consider yourself a strategist, you should use both.

The case for research informing data

Imagine a clothing brand that sells a sweater in both black and white in its online store.  Any data analyst can pull an astounding amount of information about the sales that will inform marketing strategy.  At its simplest, these data can show what percentage of people bought black vs. white out to as many decimal places as you please.  Without too much trouble, an analyst could also find interesting trends such as which color sold better at which time of the day or which one resulted in the larger average order size.  Add in data about the buyers and the analyst can tell you where black sold better than white or which sold better to longtime customers vs. first-time buyers.

Know what the data can’t tell you?  How many people wanted a blue sweater instead?

While data partisans prize data for their irrefutability, data tend to look backwards at what people did and not forwards at what they might do.  Looking forward means more than adding another sweater color.  In a larger sense, it means seeing opportunities that data simply can’t predict.

The case for data informing research

Research, on the other hand, has something to learn from data.  One need only look at polls where Americans report interest in healthier foods and compare that interest to actual sales of, say, Doritos to see the limits of research..  The discrepancy between what people say and what they do doesn’t invalidate research, of course.  Research gets strategists into the heads of consumers in a way that data simply can’t.  Like a CT scan, good research shows us not simply what consumers think, but how and why they think that way.

That said, data enhance research’s usefulness by giving scale and weight to findings or even by ensuring that the strategists answer the right questions in the first place.  In a famous example from the 1960s, one of the corn oil companies conducted research that proved, to them at any rate, that people preferred the taste of home-fried chicken over a new product, Shake ‘N Bake.  Correct answer, but to the wrong question.  Had the corn oil company recognized that people DID buy Shake ‘N Bake, perhaps they would have asked those purchasers WHY they bought the product and taken action to maintain sales of their product.

Come to think of it, I’d like to research a nice piece of fried chicken

Think “both/and” not “either/or”

A strategist who employs both data and research together can expect to provide stronger rationales for her recommendations.  Data give research a foot in the here-and-now while research gives data more understanding of why the numbers are what they are and how marketers can identify new opportunities.  Try using them to support one another and then maybe, just maybe, your opinions might count after all.

Data Laziness

Ridership on the New York City subways declined last year because, well, they’re not sure, really:

The [Metropolitan Transit] authority’s acting chairman, Fernando Ferrer, said on Thursday that several factors could be contributing to the decline: rising subway delays, the popularity of Uber and other apps, and weekend maintenance work that disrupts service.

“It may be all of the above,” Mr. Ferrer told reporters after an authority board meeting. “I’m very glad that our ridership is at historic highs. If it declines a little bit — and I’ve seen those numbers, and it’s a little bit — there is no reason for alarm.”

You want “reason for alarm?”  I’ll give you reason for alarm: the MTA’s chairman can’t be bothered to run a simple Excel spreadsheet.  Let’s call this “data laziness” and show you how easy it would be to get a more definitive answer.

Actually, Excel is a much more useful tool here

Continue reading

Free Advice for the ACLU

On the heels of my post yesterday about 84 Lumber’s liberal-leaning Superbowl ad, I felt the need to share that liberal does not necessarily equal smart.

Witness, this email from the American Civil Liberties Union, of which I am a proud member of long standing (OK, since last Saturday):

Please don’t call me Benjamin, BTW

Note the call to action: Tell your senator today about DeVos’ disturbing positions before they vote

Both my of my senators, Chuck Schumer and Kristen Gillibrand, have gone on record against Ms. DeVos.  The ACLU has my address; they know there’s no point in my calling my own senators.  Yet here we are.

Listen, one marketer to another, here’s what you could have done:

  1. Kept your powder dry.  You could have suppressed anyone from a state where both senators oppose Ms. DeVos and thus saved the opportunity to email me at a time when I can actually be of help in New York.
  2. Given me an alternative.  They could have asked anyone in a state with two Democratic senators to badger friends in other states, especially the Badger State, where Ron Johnson supports Ms. DeVos.  They could have asked me to share the organization’s point of view on Ms. DeVos via social media.  Hell, they could have asked for a few more bucks.

If we want to protect our sacred civil liberties, we will need courage.  We will need faith in the fundamental decency of the American public.  We will need legislators and jurists to do their jobs in acting as a check on executive power.

We will also need better email marketing than this.

Strategy in the Era of Brute Force Marketing

Will marketing strategists become the horse grooms of the 21st century?

horse-535760_1280

Nice work if you can get it

Grooming horses probably seemed like a nice job in the 19th century.  After all, you got plenty of exercise and got to work with animals.  What’s not to like?

Well, in a word, Buicks.

Just as one form of technology destroyed the jobs of hundreds of thousands of horse grooms, another may lay waste to the jobs of thousands of marketing and advertising strategists.  As more and more digital marketing tools adopt optimization features, some of the core functions of the marketing strategist may begin to seem redundant.  However, I think that smart strategists will regard these tools not the way that John Henry regarded the steam drill but rather the way the first taxicab driver regarded a Model T.  That is, technology doesn’t take jobs away; rather, it makes them bigger.

Continue reading

Who Cares if Email Marketing Pays for Itself?

Let me suggest a modest proposal: run your email marketing program at a loss.  As in, if you’re a retailer, stop worrying how much each email nets in incremental sales.  As in, if you’re a B2B marketer, stop worrying about how many leads each email generates.  As in, if you’re some other kind of marketer, double your email marketing budget and hang the cost.

After 10 years of articles about email marketing’s superior ROI, throwing fiscal caution to the wind seems like the worst idea since rolling coal.  Naturally, I don’t suggest taking this step for its own sake.  Rather, I suggest adjusting the way we evaluate email marketing to serve a purpose that serves the broader enterprise: research.

old-money-454458_1280

Go ahead. Rip it up.

OK, I’ve exaggerated my point of view in a shameless attempt to get your attention.  However, I strongly endorse using email as an inexpensive, flexible and fast research tool.  Let’s look at what you could achieve by integrating research into your email marketing.

Continue reading

Behind the Numbers: Market Research Isn’t a Tightrope Walk

Unlike tightrope walking, this post will dispel the suspense right away: market research differs from tightrope walking in that it’s actually a good idea to look down.

bydgoszcz-700132_1920

Not brave enough to conduct market research

As I’ve said previously, I enjoy reading the eMarketer newsletter every morning because it usually has an interesting chart or two.  Usually, the headline summarizes the charts like so:

Most Mobile Banking Users Check Balances, Statements

Indeed, according to the survey, 85% check balances and/or statements.  End of story.

Except that’s not where the story ends.  Enter the tyranny of the top two box.

Continue reading

Pre-Emptive Cringing

How soon before advertisers worm their way into Goals in Google Calendar?

donkeys-1130681_1920

Pardon us, but do you have any oats?

If you didn’t read the announcement, Google has added a feature to its popular calendar that makes it easier for users find time to reach specific goals.  You want to work on your Spanish twice a week?  Tell GCal and it’ll schedule two sessions each work para aprender Español.

Given that Google earns a tad of 90% of its income from advertising (PDF), you must excuse me for cringing in advance.

Step 1: Advertisers start buying keywords in your Calendar

Continue reading

Models Can Be Dumb

I once worked in a building that housed a photo studio.  The good news was that we often rode in the elevator with pretty, pretty models.  The bad news was that they often seemed flummoxed by the elevator button panel and would hover over it, looking for the 12th floor.

Yes, human models have a reputation for acting dumb.  By the same token, computer models often act dumb, too.  As a result, marketers need to take caution when employing models.

model-600222_1920

It would be wrong of me to use a picture of a pretty young lady just to get clicks.

Just ask the Australian Army.

Continue reading