The legitimate non-profit trends that Millennials are a part of, but don’t define

Yes, it’s a long title, but I promised nuance.

The most compelling part of the case for Millennials focus is time.  Existing donors are aging to a point when for various reasons they are no longer able to give.  I’ve argued this isn’t a Millennial thing, except insofar as they are the more extreme (so far) leaders of some important trends in nonprofit marketing.  The big ones:

Addressable media.  I remember when I saw my first cost-per-click advertising campaign.  The heavens opened and a choir of angels came down and sang “Behold! You shall not pay to advertise to nonresponders ever again!”

Time may have caused me to exaggerate this slightly, but there simply was no reason to advertise online in CPM form again.

Addressable media will make a similar sea change.  Now, there need only be four types of people you pay to advertise to online:

  • Those who are interested in your particular offering as evidenced by their searches.
  • Those who have expressed an interest in your Web site by coming and perhaps taking an action you are interested in.
  • A custom audience of people you define.
  • People who model similarly to the people above.

You’ve heard about how people seek out news that agrees with whatever viewpoints they have.  We are going to come to an age when you are going to have ads that are similar — your ads will be targeted to things the Al Gore rhythms know you want.

And as an advertiser, you need to coordinate these with your other direct marketing activities, as this will go from novel to expected to required faster than you think.

Personalization.  Millennials especially, but really all of us, are becoming more and more immune to broad brush approaches (hence why addressable media are important).  This is why books like The Cluetrain Manifesto and Permission Marketing sound current and relevant now even though they were written 15-plus years ago.

With the multiplication of media, we are simply not going to have time to pay attention to things that don’t pay attention to use.  That used to mean getting your name right instead of saying Dear Occupant.  However, as my personal law dictates, it’s going to be more and more important to know more and play back what you know about a person.  To learn more about personalization techniques, try this.   In particular, playing back people’s connection to the cause is important.

Impact. It’s often said that Millennials want to see what their gift does.  Doesn’t everyone? People want to see that they are making a difference.  Not that your organization is making a difference, but that they are making a difference.  The best thing that you can do is tell them the story of that difference related to the why that makes them give.


Requiescat in pace

Content marketing.  In the NonProfit Pro piece, I also said that “Content marketing was highly effective before it was Content Marketing and will continue to be effective long after it becomes lower case again.”  I would work to debunk the hype around that this is a new idea, but the podcast This Old Marketing does a better job than I ever could, showing content marketing schemes that go back to Poor Richard’s Almanack and before.  I’d also love to explain what content marketing is in simple words, free from hype, but Sorry for Marketing did it better than I ever could as well.  And it’s funny; here’s a sample image of the Jargon Monster:

jargonmonsterNote that Millennials is one of those jargon words…

So I’ll simply add that bringing people in through content not only acquires new constituents; it helps you learn about what those specific constituents want so you can deliver it.  

Mobile.  If you aren’t optimizing for mobile already, do so.  It’s now, for many if not most, the primary way that people are looking at your Web site, donation form, content marketing, etc.

Cultivation — valuing people over institution and connections over transaction.  FOTB (Friend of the Blog) Angela Struebing did a nice 2016 intro here talking about getting to real donorcentricity and talking about impact, rather than our usual talking about programs and studies and such.

But this is frequently talked about in the context of Millennials — they want a relationship, not a transaction; they want to fund causes and impacts, not organizations.  Like so many things, this isn’t just Millennial phenomenon, but something we will have to wrestle will from now until I don’t know when.

So, in summary, when someone says they want to target Millennials, start by trying to improve your messaging to humans.  I assure you, regardless of what you read to the contrary, Millennials are human.

The legitimate non-profit trends that Millennials are a part of, but don’t define

Nessies, Bigfeet, and Large Numbers of Millennial Donors

First, thank you to the Golden Globes for honoring my blog post about correlation not equalling causation featuring Matt Damon as Best Blog Post Comedy.

This week, I’m going into detail on my thought from my NonProfit Pro piece that said

We should regard a nonprofit that courts a Millennial audience at the expense of their core like the person who dyes their hair and takes off their ring to hit on people at a college bar: unfaithful to those who love them, uncomfortable with who they are, and ill-equipped to succeed even if success were desirable.

Let me first say that if your nonprofit wants to be around for the long-term, you will have to address younger people.

Because birth, death, and math.

And as we’ve said that past few days, you shouldn’t not target Millennials.  There really is no such generation and, even if you are looking just at the age group, there are enough intragenerational variations that there will be quality prospects in any age group.

My argument is just you can probably ignore all of the “how to talk to Millennials” think pieces you have seen and will see, and refocus on telling your nonprofit’s story well to individuals who will react well to it.

The maxim that we should be following is his:


That’s Willie Sutton and he allegedly said, when asked why he robbed banks, that it’s where the money is.

The height of Millennial absurdity, expressed to me by an otherwise sensical colleague, was the argument that we needed a Millennial-focused planned giving strategy.

So where is the money?  Blackbaud’s The Next Generation of American Giving says it’s:

  • 43% Baby Boomers
  • 26% Maturers (or Civics)
  • 20% Gen X
  • 10% Gen Y

As Blackbaud summarizes “In short, the odds are strong that for the vast majority of causes, your next donor will be over age 50.”

And that donor will be more profitable.  The average gift from a Boomer or a Mature is$454 and $478 respectively.  For Gen Y, it’s $272.

I know, I talked big against Blackbaud sometimes, but they really do good reports.  This one is here and I’m sorry you will have to give them your email to get it, but it’s worth it.

So, in summary:

  • Millennials often don’t have the unique attributes often attributed to them.
  • In fact, the whole generation system is pretty flawed.
  • They are not one coherent group available for targeting; in fact, other attributes like race and gender are far more predictive.
  • Even if they were available for targeting, they aren’t worth targeting for donations right now.

And here I said this was going to be a more nuanced look.

However, we did see that there were some things that are changing over time.  While these are commonly attributed to Millennials, they are likely trends that will change the way we do direct marketing over the long term.  That’s what I’ll take on tomorrow.


Nessies, Bigfeet, and Large Numbers of Millennial Donors

The intragenerational dynamics of Millennials

Monday and yesterday, I argued that many of the so-called Millennial attributes aren’t unique to Millennials and, in fact, that the dynamics among generations are overblown.

I should stipulate here that it seems obvious that people who have significant events at formative times in their lives may have similar reactions.  Those who lived through the Great Depression were more likely to save as a result.  Similarly, many from this generation don’t like to have extended long-distance phone calls because they used to be very expensive.

Ironically, for me, it’s this belief in formative events that makes me less likely to buy into generational dynamics.  It seems odd to me that for whom 9/11 happened while in college will likely think the same way about security issues on average than someone for whom 9/11 happened while in utero.  Further, to say that their reactions would be preordained seems even more implausible.

All of this could be excused, perhaps, if it led to a usable schema.  After all, if something works in practice but not in theory, it simply begs for better theory.

However, looking at Millennials and saying they act one way or the other as a group is not reliable.  In fact, it would likely be better to look at any other factor than age to get an idea of a person.

This sounds controversial, but let’s take this chart as an example.  While a sdt-next-america-03-07-2014-0-13simple example, President Obama’s appeal among younger voters was a significant part of the narrative in the 2008 election.  

As you can see, white Millennial’s approval rating of President Obama is between white Gen Xers and white Boomers.  Non-white Millennial approval is slightly higher than non-white Boomers, but within the margin of error.

So if you wanted to predict whether someone supports President Obama, it would be far more instructive to know someone’s race than their age.  Or, put another way, a 25-year-old white person is more likely to be like a 65-year-old white person than a 25-year-old non-white person.

Let’s look at more actionable variables for us as direct marketers.  One thing we do know for sure is that Millennials own social media, right?


Sort of.  They use social media more than other age groups.  However, 11% have no Facebook accounts and 27% use it less than once per week.  And that’s the most used social network.

And, not a surprise, it’s not the same by sex:

sillsgraphic1 Hat tip here 

And there’s significant age variation within Millennials.  About a quarter (27%) of 31-35 year olds use Snapchat, compared with almost two-thirds (65%) of 21-25 year olds.  I should mention that some of the more enlightened generational theorists of my acquaintance talk about how people on the border of generational categories are tweeners and these are spectra, rather than hard dividing lines.  This warms my heart in part because I’m an Xer and my wife is a Millennial despite only a two-year age gap.

This is something for-profit marketers have caught on to.  The Hotwire PR study of communications trends proclaimed the end of trying to talk to Millennials as a monolithic group and more toward addressable media and direct marketing (including print!) to address as individuals.

So the big question I would have is why would you want a strategy for Millennials, when you could have strategies to acquire online advocates as warm leads, renew lapsed donors, and everything else that is actually related to your organization.  I think you’ll find that your walkers look very much like your walkers, your advocates like your advocates, and so on, than your Millennials like your Millennials.

This brings up another question: is it worthwhile to target far younger constituents as a way to get gifts?  My answer is no, with caveats, and I’ll hit the details tomorrow.

Agree?  Disagree?  Let me know in the comments.

The intragenerational dynamics of Millennials

Mythbusting the millennial mythos

Yesterday, I ranted a bit about the intellectual lassitude of people who talk about the unique attributes of millennials.  Let’s put some of these to the test.

First, a background on where the idea of generations comes from.  Authors William Strauss and Neil Howe created the Strauss-Howe generational theory — that every 20 or so years, there is a new generation.  They further posit that there are four generational patterns in rotation: prophets, nomads, heroes, and artists.  So, for example, according to them, the Silent Generation are prophets, the idealists that helped create the post-war establishments that Baby Boomers, as nomads, rebelled against.  Gen Xers are the heroes, who grow up increasingly protected, but mature into self-reliance.  Millennials are artists, who “grow up overprotected by adults preoccupied with the crisis, come of age as the sensitive young adults of a post-crisis world.”

Of course, it won’t surprise you if you read yesterday’s post that I am switching these around.  According to Strauss and Howe, Silent Generation members are artists, Baby Boomers are prophets, Gen Xers are nomads, and Millennials are prophets.  

In the end, the whole thing reads like the Chinese zodiac readings on a restaurant placemat: vague enough to apply to anything or to nothing.  The generational theory is non-falsifiable.  There are no hypotheses to test.  And thus, there is no science behind it and thus belief in the system is as valid as Roswell or Bigfoot.

In fact, the writing often (to me) smacks of what is called a Jacques statement in cold reading (aka faking that you are psychic).  The Jacques statement is named for the character who gives the Seven Ages of Man speech in “As You Like It”; it means tailoring your prediction to the age of the subject.  Take a look at the very very different things that happen to each generation as they get older:

  • Prophet: “tend to be remembered for their coming-of-age passion and their principled elder stewardship”
  • Nomad: “tend to be remembered for their rising-adult years of hell-raising and for their midlife years of hands-on, get-it-done leadership”
  • Hero: “tend to be remembered for their collective coming-of-age triumphs and their hubristic elder achievements”
  • Artist: “tend to be remembered for their quiet years of rising adulthood and their midlife years of flexible, consensus-building leadership”

Thus, other than the artist, everything rebels when they are young.  And everyone matures when they get older.  These are the alleged “differences” among the generations.

Additionally, in the generational dynamics world, the future is already written.  By definition, a baby born today is a hero and their child will be a prophet.  I don’t have a scientific basis for this, but I find this level of determinism unsettling, especially when there isn’t a compelling reason to believe in it.

So those are the underpinnings of the theory, such as they are.  Now let’s look at some of the more commonly asserted attributes of Millennials.  In order to be a truly Millennial trait, it would have to be something that does not happen to every generation that is this age (because then you can target all 20-somethings similarly across time without generational embroidery or Jacques statements) and something that does not continue over time (because that’s a trend and not a generational commonality.

Take, for example, the technological savvy of the Millennial generation.  All of the data do point toward greater use of social media, greater use of the Internet, greater mobile use, etc.  But this trend seems to be going in one direction: up.  Not only are all age groups showing greater adaptation among all age groups, but there is no sign that the 15 and under set (the to-be-named generation after the Millennials) will not be even more digitally native than Millennials.  For me, then, the statement “Millennials are the most tech savvy generation” has the same meaning as “the youngest adults are Millennials” — something that will eventually be supplanted (perhaps by the Singularity).

Other general trends that you may have heard of as uniquely Millennial:

  • Millennials are the most educated generation.  Same thing as with technology: why would this trend stop with Millennials?
  • Millennials prefer cities to suburbs.  Actually, a 25-30-year-old today is less likely to live in a city today than one in 2000.  This is something that is unique to young people, not to Millennials. 
  • Millennials job hop.  FiveThirtyEight myth busted this one for me here.   Young worker job switching is actually down from both one and two decades ago.
  • Millennials want to see the impacts of their gifts.  Do you think this is not common among other age groups?
  • Millennials want a trophy for every little thing they do.  IBM did a good study of generations in the workplace here.  It found that Millennials were only slightly more likely than Gen Xers to want recognition from their boss and less likely than Baby Boomers to want their views solicited by their boss.  Gen Xers, not Millennials, were the most likely to think that everyone on a team should be recognized.
  • Millennials are uniquely socially conscious.  That same study found that Millennials were less likely than their Gen X and Boomer counterparts to want to leave a job to follow their heart or save the world.  Oxford Economics found the same thing here; only a fifth of Millennials said making a difference is important to their job satisfaction.

So this debunks some of the more common attributes that Millennials are commonly cited to have.  

But this would be all academic if there were a good way to create messages that worked for Millennials generally.  Unfortunately, there isn’t, because of significant intra-generational differences.  We’ll discuss that tomorrow.

Agree?  Disagree?  Let me know in the comments.

Mythbusting the millennial mythos

Millennial Myth Busting: Attributes of Gen Y

I had the privilege of having a Corner Office article published in December’s NonProfit Pro. (I’d link to it, but it doesn’t yet appear to be online.)  One sentence in that piece triggered more reaction than all three months of my blogging combined:

We should regard a nonprofit that courts a Millennial audience at the expense of their core like the person who dyes their hair and takes off their ring to hit on people at a college bar: unfaithful to those who love them, uncomfortable with who they are, and ill-equipped to succeed even if success were desirable.

What I had not realized is that it appears that there are cultural warriors on both sides of a debate that summarizes to “Millennials are awesome and the future” versus “Millennials are horrible and the Earth is doomed.” And I had come down in the “get off my lawn” Gran Torino camp.


His next movie was focused on how angry he was with that empty chair.

So this week, I wanted to add a bit of nuance to this statement and to the strategy discussion of millennials and non-profits.  I add emphasis to strategy here.  My central point is the Corner Office article was to highlight that sometimes trends are used instead of strategies.

Nowhere does this seem clearer to me than in the discussion about generational dynamics, especially as it concerns the unique snowflakes called millennials.  The discussions remind me of the introduction to the Duck tours of Wisconsin Dells we went on growing up, where the tour guide would tell you that what you were about to hear was about one-third the truth, one-third Native American legends, and one-third out-and-out lies.

Since the Confirmed/Plausible/Busted trichotomy is likely copyrighted (copywrote?) by people who bust myths far better than I, I’ll use this truth/legend/lie way of breaking things down.  

Let’s review some of the attributes that millennials purportedly possess:

[They] have radically different life experiences than those in generations before them.”

[T]hey distrust hierarchy. They prefer more informal arrangements. They prefer to judge on merit rather than on status. They are far less loyal to their companies. … They know computers inside and out. They like money, but they also say they want balance in their lives.

They are “more collaborative”, “less hierarchical,” “more altruistic”, “more tech-savvy,” “balanced”, “candid in their communications,” and “rule-shy.”  

Most of children seem to be taking so long to grow up, at least by conventional measures. Therituals that once marked adulthood – graduation, the first job, marriage, children – have been delayed, eliminated or extended.

They are a “highly educated, pampered group, their numbers are small but their impact is great.”

By now, you have probably guessed the conceit here – all of these things weren’t said about millennials.  They are contemporaneous accounts about Generation X and Yuppie Baby Boomers.

The truth is that many of the things said about millennials are the things said about kids since time immemorial.  So much of what you hear about kids today with their smartphones, their social networks, and their texting are echoes through the ages of hearing about kids today with their fire, their pointed sticks, and their paintings inside the caves.

I’d highly recommend a Mental Floss list of some of these throughout the ages for the humor in this.  As you see surveys about how gender and sexuality are more fluid among the young (also here)), hopefully this 1771 broadside against the feminization of the then-current set of men sounds familiar:

Whither are the manly vigor and athletic appearance of our forefathers flown? Can these be their legitimate heirs? Surely, no; a race of effeminate, self-admiring, emaciated fribbles can never have descended in a direct line from the heroes of Potiers and Agincourt.

Can’t you just hear the ”harrumph” that must have followed this statement, possibly followed by a feverish polishing of a monocle or some such?  Plato himself wrote that kids are rude and don’t respect authority.  I would have put that in the list above, but the fact that it’s in ancient Greek might have been a giveaway that it wasn’t talking about millennials.

Yet these same “insights” are being repackaged for this current generation and will likely be repackaged for the next generation.  So tomorrow, we’ll put these to the test.

Agree?  Disagree?  Let me know in the comments.

Millennial Myth Busting: Attributes of Gen Y

Understanding and using Facebook’s algorithm

Facebook is the nexus of a lot of debate as to how best to incorporate social media into other marketing efforts.  My argument will be there is a twofold Facebook strategy: 1) using organic content to engage your superfans and 2) using addressable media to reach everyone else.


500 million users.  How quaint.

Like Google, the base of the Facebook algorithm (EdgeRank) is fairly easy:

  • Affinity: How close the person creating the content is to the person receiving it.
  • Weight: How much the post has been interacted with it, with deeper interactions counting more
  • Time decay: How long it has been since it has been posted.

These interactions are multiplied together and summed, roughly.

Like Google, however, it has been altered over time significantly.  There are now significant machine learning components baked in that help with spam detection and bias toward quality content.  Additionally, now users can prioritize their News Feeds themselves.  Finally, because of the sheer amount of content available, the organic reach of an average post is single digit percentages or below, meaning that if you have 100,000 likes, maybe 2,000 people will see your average post.

The implications of this base algorithms are stark:


  • Organic reach on Facebook is for the people who really love you.  Many people think of Facebook as a new constituent acquisition system.  However, people who come in dry will almost never see your posts.
  • Consequently, only things that connect with your core will have any broader distribution.  Think of who is in the top two percent of your constituents: employees, top volunteers, board members, and that may be about it.  If those people don’t give the post weight, no one outside of this group will see it.
  • What you have done for them lately has outsized weight.  Research into Facebook interactions shows that Facebook gives outsized weight to what a person as interacted with in their last 50 interactions.
  • Facebook is not for logorrhea like Twitter.  Think of your posts as a currency you spend each time.  If your post gets above average interactions, you will move your average up and interact with more people; if not, your reach will lose.  Posting too many times (which varies from organization to organization) will diminish your audience as average reach will decline).  Additionally, all of the things you have to post for organizational reasons (e.g., sponsor thank yous) are spending your audience and you have to assess how much you are willing to spend to fulfill those objectives.
  • This all adds up to the uber-rule: Facebook is for things your core supporters will interact with quickly.  If they don’t, it won’t reach your more distant supporters and it will lessen the likelihood that your next post will reach them as well.
  • It also relates to the second uber-rule: because Facebook can change its algorithm as it wishes, you should not build your house on rented land.  The best thing you can do with your interactions is to direct them to your site, to engage your content and sign up for your list.

This all sounds a bit dire, so I should also highlight how to reach the other 98%(ish) of your Facebook audience as well as some of your non-Facebook audience on Facebook: addressable media.

Facebook allows you to upload a list of your supporters and target advertising to them specifically whether or not they are current Facebook likers of you.  You can learn more about this on my CPC ads post here.  This also goes into lookalike audiences, a way of getting people who aren’t who you talk to currently, but look a lot like them, a nifty acquisition trick.  Since organic reach won’t get you to these loosely and non-affiliated people, this is the only way to achieve that reach.  And, since it is cost-per-click, you can control your investment and your results.

But like discussed above, these campaigns should be to build your relationship to people outside of Facebook.  For the same reason companies advertising on CBS don’t work to build a greater relationship to CBS, but rather to the advertising companies, your advertising on Facebook shouldn’t be aimed at getting Mark Zuckerberg et al more friends — they have over a billion of them already.

Understanding and using Facebook’s algorithm

How to use Google’s algorithm in your direct marketing

You may say a search engine optimization strategy is not direct marketing. I humbly disagree.  In fact, working with Google and other search engines (but mostly Google) can help you with your warm lead generation, helping you get your direct marketing program started for free as I’ve advocated in the past.  In addition, by knowing what a warm lead same to you for and about, you can customize your approach to that person in interesting ways.

So, how does the Google algorithm work?  It’s been through approximately a googol different versions throughout the years (there’s a good basic list here), but some of the underlying thinking behind it has been largely unchanged.

It’s instructive to think about search engines pre-Google.  There were two different models: directories that were maintained by hand, by either a company (e.g., Yahoo) or by a community (dmoz) and search engines that used textual analysis to determine how applicable a page was to your search (e.g., Altavista, Lycos).  The first model has obvious problems with the scale of the Web.  The second has problems with determining quality. People who would spam every possible keyword for a page at the bottom of the page or create 100,000 pages each focused on optimizing for its own set of terms performed well in these engines, but probably should not.

The fundamental question was how do you have a computer determine reputation?

The basic insight that the Google founders had was from the world of academia, where a research paper’s quality can be estimated by how many papers cite it.  They realized that when someone links to a page, they are voting for that page’s quality.  Looking at the initial linking pattern, you can get a basic view of what important sites are.  Then, you can factor in the quality of the linking sites to alter the quality rankings.  After all, getting one link from, for example, the White House is more important that 100 different links from Jim Bob’s Big House of Internet.


This is the core of the original Google algorithm called PageRank (named after Larry Page, not Web pages, oddly enough).

The changes over the years since have made this influence important, but not the sole criterion as it used to be.  Other factors now include:

  • Machine learning based on what people actually click on (a different type of “voting”)
  • Weighting toward mobile-friendly sites
  • Personalization of search engine listings
  • De-spamming algorithms
  • Devaluation of ads above the fold
  • Incorporation of social signals
  • Situational reputation (e.g., if my blog linked to you, it would help you more for direct marketing terms than with your hummingbird mating pattern blog)

And it’s constantly evolving.  So there are a few implications to this:

The easiest way to get good search engine listings isn’t to optimize for Google; it’s to create quality content.  I know.  This is a bummer.  Or not, if you have quality content.  The goal of Google and other search engines is to evolve to make searching a true meritocracy.  In the beginning, you had a chance of gaming the system.  You don’t have that chance now.

That does include things like not having ad-based content, making it mobile friendly, and prompting social media interactions.

There’s an important corollary to this, which is that anyone who tells you that they have a special sauce either is lying or won’t have their tactics last out the year.  That said, there are a few that you can do that will help both your content quality and your search engine listings.

Make sure you have the terms you want to be found for in your articles.  Not even Google will find the best possible page for “is James Bond a Time Lord?” (hint: it’s this one) if it doesn’t have the words James Bond and Time Lord on it.  Ideally, these will be prominently placed (e.g., in the title or header tags) and frequent (but not spammy frequent).

Check your bounce rates. With machine learning incorporated into the algorithm, you want to make sure people are getting what they came for when they come to your page.  This makes continual testing and improvement of your content will pay dividends.

Create content for the searches you want to dominate. Let’s say you are (or want to be) the premier early childhood education nonprofit in Missoula, Montana.  You find through your keyword research that people don’t necessarily look for “early childhood education”; they look for conditions (e.g., “autism services”, “Down’s Syndrome”) or symptoms (e.g., “child not speaking”, “when starting crawling”, “development milestones”).  Look the volume of search terms, which you can do with Google’s free keyword suggest tool once you have your AdWords account and Google Grant.

You do have a Google Grant, don’t you?  If not, get one ASAP here.  

So, let’s say you want to focus on autism to start — you should be creating content that helps parents in your area learn about autism, what it is, and how you can serve them.  Lather, rinse, and repeat with your other areas of content.  Not only will this help with the Google algorithm (in terms of keyword density and in terms of more people linking to quality content), but it will also help with conversions (as people get content that fills an established need) and in knowledge past conversion (if someone comes in on an autism search term to autism content, you can market to them differently than someone looking for Down’s Syndrome content).

Finally, ask your partners to link to your specific content.  This isn’t link spamming, but rather you linking to people who have good content for your constituents and vice versa.  This will help lift both of your boats.

I’m sorry that there are no magic beans to sell you here from the algorithm. But hopefully this will help you avoid buying someone else’s.

With Facebook, however, there are a few more lessons for organic content that we will cover tomorrow.

How to use Google’s algorithm in your direct marketing

Regression analysis in direct marketing

If you don’t know what a linear regression analysis is or how it is measured, I recommend you start with my post on running regressions in Excel here.

OK, now that you’re back, you’ll notice I did an OK job of saying what a linear regression analysis is and what it means, but I didn’t mention why these would be valuable.  Today, we rectify this error.

In yesterday’s post on correlations, I mentioned that they only work for two variables at a time. This is extremely limiting, in that most of your systems are more complex with this.  Additionally, because of interactions between multiple variables, it’s difficult to determine what is causing what.  I’ve discussed before how the failure of the US housing market was related to people assuming variables that are independent were actually correlated with each other.  

Linear regression analysis allows you to look at the intercorrelations between and among various variables.  As a result, regression analysis is the primary basic modeling algorithm.  In fact, it’s often used as a baseline for other approaches — if you can’t beat the regression analysis, it’s back to the drawing board.

Two side notes here:

First, if you are interested in learning to do this yourself, I strongly recommend Kaggle competitions.  Kaggle is where people compete for money to produce the best models for various things — right now, for example, they are running a $200,000 competition on diagnosing heart disease, a $50,000 competition for stock market modeling, and a $10,000 competition to identify endangered whales from photography.

It’s some pretty cool data stuff and the best part is that they have tutorial competitions for people like me (and perhaps you; I would hate to assume).  One sample is to model what passengers would survive the sinking of the Titanic from variables like age, sex, class ticket, fare, etc.  They walk you through correlation, regression, and some more advanced modeling techniques we’ll discuss later in the week.  Here, as ever, they look for improvement on regression as the goal of more advanced models.

Second, it’s tempting to view regression as a Mendoza line* of modeling: a lowered hurdle that shouldn’t be bothered with.  But regression can give you fairly powerful results and, unlike many of the other more advanced modeling we’re going to discuss, you can do it and interpret it yourself.

That said, like correlation, it doesn’t know what to do with non-linear variables.  For example, you have probably noticed that your response rate falls off significantly after a donor hasn’t donated in 12 months (plus or minus).  A regression model that looks at number of months since last gift will ignore this and assume that the difference between 10 and 11 months is the same as the difference between 12 and 13 months.  And it isn’t.  It also will choke on our ask string test in the same way as correlations will.

So here are some things worth testing with regression analyses:

Demographic variables: you may know the composition of your donor file (and if you are like most non-profits, it’s probably female skewed).  But have you looked at which sex ends up becoming the better donor over time?  It may be with a regression analysis that the men on your file donate more or more often (or not), which could change your list selects (I know I have been known to put a gender select on an outside file rental to improve its performance).

Lapsed modeling your file: Using RFM analysis, you know what segments perform best for you and which go into your lapsed program (if not, use RFM analysis to figure out what segments perform best for you).  However, there may be hidden gems in your file that missed a gift (according to you) and would react well if approached again hidden in your lapsed files.  Taking your appended data like wealth, demographics, and other variables alongside your standard RFM analysis can help find some of these folks to reach out to.

Content analysis: In the early regression article, I show a (bad) example of using regression analysis to find out what blog posts work best.  This can be applied to Facebook or other content as well.

What I didn’t mention is that once you have this data, it probably applies across media.  What works on Facebook and in your blog are probably good topics for your enewsletters, email appeals, and possibly paper newsletters as well.  Through this type of topic analysis, you will figure out what your constituents react to, then give them more of it.

This, however, looks at your audience monolithically.  In future posts, I’ll talk about both some ways to cluster/segment your file like k-means clustering and some ways on improving on regression analysis with techniques like Bayesian analysis.  For now, though, it’s time to look at some formulae that rule our worlds even beyond direct marketing: what do Google and Facebook use?


* A baseball term coming from Mario Mendoza, a weak-hitting shortstop who usually averaged around .200 batting average.  Anyone below Mendoza in the batting average category was considered to be hitting below the Mendoza line or very poorly.  (He made up for this for several years with strong fielding).  And now you know the rest of the story.


Regression analysis in direct marketing

Correlations in direct marketing II: The Wrath of Khan Academy

Yesterday, we saw how to run and interpret correlations.  Today, we’re going to look at the implications of the way correlations are set up for direct marketers.

First and foremost, I must stipulate that correlation does not equal causation.  I did a good job of discussing this in a previous post talking about how attractive Matt Damon is in his movies.  Rather than go into a lot of detail on this, I’ll link over to that post here.  Looking back at that post, I forgot to put in a picture of Matt Damon, which I will rectify here:



Intelligence and attractiveness correlate;
I wish I could have explained this to people in high school.

This is fairly intuitive, given our discussion of height and weight earlier.  With exceptions for malnutrition and the like, it really doesn’t make sense to say someone’s weight causes them to be taller or height causes them to be heavier.

There’s a great Khan Academy video that covers a lot of this here. The Khan Academy video also gives me an excuse for the name of the blog post that I really couldn’t pass up.

Back to correlations, they only predict linear (one-way) relationships.  Given the renewal rates above, a correlation is not the ideal tool for describing this relationship, as it will give you a rubric that says “for every decrease in decile number, you will have an X% increase in renewal rate.”  We can see looking at the data that this isn’t the case — moving from 2 to 1 has a huge impact, whereas moving from 9 to 6 has an impact that is muddled at base.

Another example is the study on ask strings we covered here.  When looking at one-time donors, asking for less performed better than asking for the same as their previous gift.  Asking for more also performed better than asking for the same as the person’s previous gift.  However, if you were to run a correlation, it would say there is no relationship because the data isn’t in a line (graphically, you are looking at a U shape).  We know there is a relationship, but not one that can be described with a correlation.

You’ll also note that they only work between two variables.  Most systems of your acquaintance will be more complex than this and we’ll have to use other tools for this.  That said…

Correlations are a good way of creating a simple heuristic.  SERPIQ just did an analysis of content length and search engine listings that I learned about here. They found a nice positive correlation:


Hat tip to CoSchedule for the graph and SERPIQ for the data.

As you read further in the blog post, you’ll see that there is messiness there.  It’s highly dependent on the nature of the search terms, the data are not necessarily linear, and non-written media like video are complicating.  However, the data lend themselves to a simple rule of “longer form content generally works a little bit better for search engine listings” or, in a lot of cases, “your ban on longer-form content may not be a good idea.”  While these come with some hemming and hawing, being able to have simplicity in your rules is a good thing, making them easier to follow.

But refer back to the original point: correlation isn’t causation.  Even in the example above of word count being related to search engine listings, more work is required to find out what type of causal relationship, if any, there is between word count and search engine listings.

Hope this helps.  Tomorrow, we’ll talk about regression analysis, which will take you all the way back to your childhood to look for memories that will…

Um, actually, it will be about statistical regression analysis.  Never mind.

Correlations in direct marketing II: The Wrath of Khan Academy

Correlations in direct marketing: an intro

This week, I’d like to take a look at some of the formulae and algorithms that run our lives in direct marketing.

Al Gore giving his global warming talk in Mountain View, CA on 7

The algorithm was invented by Al Gore and named after his dance moves
(hence, Al Gore rhythm).
Here is a video of him dancing.

Before you run in fear, my goal is not to make you capable of running these algorithms — some of the ones we’ll talk about this week I haven’t yet run myself.  Rather, my goal is to create some understanding of what these do so you can interpret results and see implications.

And the first big one is correlation.

But, Nick, you say, you covered correlation in your Bloggie-Award-winning post Semi-Advanced Direct Marketing Excel Statistics; will this really be new?

My answers:

  1. Thank you for reading the back catalog so intensely.
  2. The Bloggies, like 99.999998% of the Internet, do not actually know this blog exists.
  3. In that post, I talked about how to run them, but not what they mean.  I’m looking to rectify this.

So, correlation simply means how much two variables move together (aka covariance).  This is measured by a correlation coefficient that statisticians call r.  R ranges from 1 (perfect positive relationship) to 0 (no relationship whatsoever) to -1 (perfect negative relationship).  The farther from zero the number is, the stronger the relationship.

A classic example of this is height and weight.  Let’s say that everyone on earth weighed 3 pounds for every inch of height.  So if you were 70 inches tall (5’10”), you would weigh 210 pounds; at 60 inches (5’0”), you would weigh 180 pounds.  This is a perfect correlation with no variation, for an R of 1.

Clearly, this isn’t the case.  If you are like me, after the holidays, your weight has increased but you haven’t grown in height.  Babies aren’t born 9 inches long and 27 pounds (thank goodness).  And the Miss America pageant/scholarship competition isn’t nearly this body positive.  So we know this isn’t a correlation of one.

That said, we also know that the relationship isn’t zero.  If you hear that someone is a jockey at 5’2”, you naturally assume they do not moonlight as a 300-pound sumo wrestler on the weekend.  Likewise, you can assume that most NBA players have a weight that would be unhealthy on me or (I’m making assumptions based on the base rate heights of the word with this statement) you.

So the correlation between height and weight is probably closer to .6.

There’s a neat trick with r: you can square it and get something called coefficient of determination.  This number will get you the amount of one variable that is predicted by the other.  So, in our height-weight example, 36 percent (.6 squared) of height is explained by its relationship with weight and vice versa.  It also means that there’s 64 percent of other in there (which we’ll get to tomorrow when we talk about regression analysis.

You can get some of this intuitively without the math.  Here’s a direct marketing example I was working on a couple of weeks ago.  An external modeling vendor had separated our file into deciles in terms of what they felt was their likelihood of renewing their support.  Here’s what the data looked like by decile:

Decile Retention rates over six months
1 50%
2 40%
3 35%
4 32%
5 30%
6 25%
7 27%
8 24%
9 28%
10 21%

No, this isn’t the real data; it’s been anonymized to protect the innocent and guilty.

You need only look at this data to see that there is a negative correlation between decline and retention rate — the higher (worse) decile, the lower the retention rate.  It also illustrates that it’s not a perfect linear relationship — clearly, this model does a better job of skimming the cream off the top of the file than predicting among the bottom half of the file.  

Tomorrow, we’ll talk about the implications of these correlations for direct marketing.

Correlations in direct marketing: an intro