Don’t Blame Personalization After Messing It Up

In late 2019, Gartner predicted “80% of marketers who have invested in personalization efforts will abandon them by 2025 because of lack of ROI, the peril of customer data, or both.” But before giving up because the first few rounds didn’t pay off, shouldn’t marketers stop and think about what could have gone wrong?

In late 2019, Gartner predicted “80% of marketers who have invested in personalization efforts will abandon them by 2025 because of lack of ROI, the peril of customer data, or both.” Interesting that I started my last article quoting only about 20% of analytics works are properly applied to businesses. What is this, some 80/20 hell for marketers?

Nonetheless, the stat that I shared here begs for further questioning, especially the ROI part. Why do so many marketers think that ROI isn’t there? Simply, ROI doesn’t look good when:

  1. You invested too much money (the denominator of the ROI equation), and
  2. The investment didn’t pay off (the numerator of the same).

Many companies must have spent large sums of money on teams of specialists and service providers, data platforms featuring customer 360, personalization software (on the delivery side), analytics work for developing segments and personas, third-party data, plus the maintenance cost of it all. To justify the cost, some marginal improvements here and there wouldn’t cut it.

Then, there are attribution challenges even when there are returns. Allocating credit among all the things that marketers do isn’t very simple, especially in multichannel environments. To knock CEOs and CFOs off their chairs – basically the bottom-line people, not math or data geeks – the “credited” results should look pretty darn good. Nothing succeeds like success.

After all, isn’t that why marketers jumped onto this personalization bandwagon in the first place? For some big payoff? Wasn’t it routinely quoted that, when done right, 1:1 personalization efforts could pay off 20 times over the investment?

Alas, the key phrase here was “when done right,” while most were fixated on the dollar signs. Furthermore, personalization is a team sport, and it’s a long-term game.  You will never see that 20x return just because you bought some personalization engine and turned the default setting on.

If history taught us anything, any game that could pay off so well can’t be that simple. There are lots of in-between steps that could go wrong. Too bad that yet another buzzword is about to go down as a failure, when marketers didn’t play the game right and the word was heavily abused.

But before giving it all up just because the first few rounds didn’t pay off so well, shouldn’t marketers stop and think about what could have gone so wrong with their personalization efforts?

Most Personalization Efforts Are Reactive

If you look at so-called “personalized” messages from the customer’s point of view, most of them are just annoying. You’d say, “Are they trying to annoy me personally?”

Unfortunately, successful personalization efforts of the present day is more about pushing products to customers, as in “If you bought this, you must want that too!” When you treat your customers as mere extensions of their last purchase, it doesn’t look very personal, does it?

Ok, I know that I coveted some expensive electric guitars last time I visited a site, but must I get reminded of that visit every little turn I make on the web, even “outside” the site in question?

I am the sum of many other behaviors and interests – and you have all the clues in your database – not a hollow representation of the last click or the last purchase.  In my opinion, such one-dimensional personalization efforts ruined the term.

Personalization must be about the person, not product, brands, or channels.

Personalization Tactics Are Often Done Sporadically, Not Consistently

Reactive personalization can only be done when there is a trigger, such as someone visiting a site, browsing an item for a while, putting it in a basket without checking out, clicking some link, etc. Other than the annoyance factor I’ve already mentioned, such reactive personalization is quite limited in scale. Basically, you can’t do a damn thing if there is no trigger data coming in.

The result? You end up annoying the heck out of the poor souls who left any trail – not the vast majority for sure – and leave the rest outside the personalization universe.

Now, a 1:1 marketing effort is a number’s game. If you don’t have a large base to reach, you cannot make significant differences even with a great response rate.

So, how would you get out of that “known-data-only” trap? Venture into the worlds of “unknowns,” and convert them into “high potential opportunities” using modeling techniques. We may not know for sure if a particular target is interested in purchasing high-end home electronics, but we can certainly calculate the probability of it using all the data that we have on him.

This practice alone will increase the target base from a few percentage points to 100% coverage, as model scores can be put on every record. Now you can consistently personalize messages at a much larger scale. That will certainly help with your bottom-line, as more will see your personalized messages in the first place.

But It’s Too Creepy

Privacy concerns are for real. Many consumers are scared of know-it-all marketers, on top of being annoyed by incessant bombardments of impersonal messages; yet another undesirable side effect of heavy reliance on “known” data. Because to know for sure, you have to monitor every breath they take and every move they make.

Now, there is another added bonus of sharing data in the form of model scores. Even the most aggressive users (i.e., marketers) wouldn’t act like they actually “know” the target when all they have is a probability. When the information is given to them, like “This target is 70% likely to be interested in children’s education products,” no one would come out and say “I know you are interested in children’s education products. So, buy this!”

The key in modern day marketing is a gentle nudge, not a hard sell. Build many personas – because consumers are interested in many different things – and kindly usher them to categories that they are “highly likely” to be interested in.

Too Many Initiatives Are Set on Auto-Pilot

People can smell machines from miles away. I think humans will be able to smell the coldness of a machine even when most AIs will have passed the famous Turing Test (Definition: a test of a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human).

In the present day, detecting a machine pushing particular products is even easier than detecting a call-center operator sitting in a foreign country (not that there is anything wrong about that).

On top of that, machines are only as versatile as we set them up to be. So, don’t fall for some sales pitch that a machine can automatically personalize every message utilizing all available data. You may end up with some rudimentary personalization efforts barely superior to basic collaborative filtering, mindlessly listing all related products to what the target just clicked, viewed, or purchased.

Such efforts, of course, would be better than nothing.  For some time.  But remember that the goal is to “wow” your target customers and your bosses. Do not settle for some default settings of campaign or analytics toolsets.

Important Factors Are Ignored

When most investments are sunk in platforms, engines, and toolsets, only a little are left for tweaking, maintenance, and expansion. As all businesses are unique (even in similar industries), the last mile effort for custom fitting often makes or breaks the project. At times, unfortunately, even big items such as analytics and content libraries for digital asset management get to be ignored.

Even through a state-of-the-art AI engine, refined data works better than raw data. Your personalization efforts will fail if there aren’t enough digital assets to rotate through, even with a long list of personas and segments for everyone in the database. Basically, can you show different contents for different personas at different occasions through different media?

Data, analytics, contents, and display technologies must work harmoniously for high level personalization to work.

So What Now?

It would be a real shame if marketers hastily move away from personalization efforts when sophistication level is still elementary for the most.

Maybe we need a new word to describe the effort to pamper customers with suitable products, services and offers. Regardless of what we would call it, staying relevant to your customer is not just an option anymore. Because if you don’t, your message will categorically be dismissed as yet another annoying marketing message.

 

For Measurement-Oriented Marketers: The Best of ‘Here’s What Counts,’ 2019

Over the past year, “Here’s What Counts” opined on several topics. But the ones that gained the most traction involved Gen Z’s views on privacy, social media data collection, and 1:1 marketing.

Over the past year, “Here’s What Counts” opined on several topics. But the ones that gained the most traction involved Gen Z’s views on privacy, social media data collection, and 1:1 marketing.

The most popular post, “Have We Ruined 1:1 Marketing? How the Corner Grocer Became a Creepy Intruder,” was reposted on LinkedIn by Don Peppers, co-author of the book, “1:1 Marketing.”  The idea grew out of an assignment I gave my students at Rutgers School of Business in Camden, N.J. The students had to compare the 1996 version of database marketing, as described by Arthur Hughes in the introduction to his watershed book, “The Complete Database Marketer,” with the current state of online direct/database marketing. Hughes likened a marketing database to the Corner Grocer, who kept mental notes on his customers’ names, personal preferences, and family connections. Specifically, the students had to tell me how marketing technology innovations have enhanced database marketing since 1996.

The Takeaway:

While they concede that the targeted ads they experience are usually relevant, several of them noted that they don’t feel they have been marketed to as individuals; but rather, as a member of a group that was assigned to receive a specific digital advertisement by an algorithm. They felt that the idealized world of database marketing that Hughes described in 1996 was actually more personal than the advanced algorithmic targeting that delivers ads to their social media feeds.

It’s not surprising that Gen Zers expect a more personalized marketing experience. As I wrote in “Gen Z College Students Weigh-in on Personal Data Collection — Privacy Advocates Should Worry.”

Some Gen Zers don’t mind giving up their personal data in exchange for the convenience of targeted ads and discounts; others are uneasy, but all are resigned to the inevitability of it.

Student comments included:

Resignation

“I do not feel it is ethical for companies to distribute our activities to others. Despite my feelings on the situation, it will continue — so I must accept the reality of the situation.”

 Rationalization

“… I feel as though consumers gain the most from this value exchange. Marketers can do pretty much whatever they want with the information that they collect, but they do not really ‘gain’ from this exchange, until people actually purchase their products …  Even if this exchange allows marketers to play with people’s vulnerabilities, it is ultimately consumers’ choice on whether or not they want to buy something.”

 And, in response to a New York Times article about Smart TVs spying on people, one student expressed:

Disgust

“Marketers are gaining money and information through various means and have the ability to do so without risk, because consumers are not going to read [a] 6,000-word privacy policy just to be able to work a television.”

Lest we think that the younger generation is alone in eschewing concerns about privacy, take a look at “Getting Facebook Sober: What Marketers Should Know About Consumers’ Attitudes and Social Data.”

While people claim to be concerned about privacy, they’re not willing to pay for it.  A Survey Monkey poll done for the news site Axios earlier this month shows that three-fourths of people are willing to pay less than $1 per month in exchange for a company not tracking their data while using their product — 54% of them are not willing to pay anything.

As we charge into 2020, we need to carefully consider how the data we give up so willingly is used to manipulate not only our purchasing behavior, but our beliefs and values. In the post, “A Question for Marketers: Is it Social or Is it Media?” I recount Sasha Baron Cohen’s speech at the Anti-Defamation League (ADL) calling Facebook “the greatest propaganda machine in history.”

I sent The Guardian’s publication of Cohen’s speech to my children, two of whom have given up their Facebook accounts. My daughter replied, “Did you learn about this on Facebook? If so, irony is dead.”

Actually, I did. RIP, irony.

Don’t Ruin Good Models by Abusing Them, Marketers

Models may be built, but the work is not nearly done until they are properly applied and deployed in live campaigns. When in doubt, always consult with the analyst in charge; hopefully, before the drop date.

Modern-day 1:1 marketing is all about precision targeting, using all available data. And precision targeting is not possible with a few popular variables selected based on human intuition.

If human intuition takes over the process, every targeting logic would start with income and age. But let me ask you this: Do you really think that the differences between Mercedes and Lexus buyers are just income and age? If that is too tricky, how about the differences between travelers of luxury cruise lines and buyers of luxury cars? Would that be explained by income and age?

I’m sorry to break it to you bluntly, but all of those targets are rich. To come up with more effective targeting logic, you must dig deeper through data for other clues. And that’s where algorithmic solutions come into play.

I’ve worked with many smart people over the years, but I’ve never met a human who is capable of seeing through interactions among complex data variables without a computer. Some may understand two- or even three-dimensional interactions, when presented in a graphic format, but never more than that. Conversely, a simple regression model routinely incorporates 10 to 20 variables, and provides us with rank orders in forms of simple scores. Forget the next generation AI algorithms; humans have been solidly beaten by computers for decades when it comes to precision targeting.

So, when you have a dire need for more accurate targeting (i.e., you want to be mostly right, not mostly wrong); and have an ample amount of data (i.e., more data variables than you can easily handle); don’t even hesitate to go with statistical models. Resistance is simply futile. In the age of abundant data, we need models more than ever, as they convert mounds of data into digestible answers to questions. (For an extended list of benefits, refer to one of my early articles “Why Model?”)

But today, I am not writing this article to convince non-believers to become believers in statistical models. Quite frankly, I just don’t care if someone still is a non-believer in this day and age. It’s his loss, not mine. This not-so-short article is for existing users of models, who may have ruined them by abusing them from time to time.

As a data and analytics consultant, I get called in when campaign results are less than satisfactory; even when statistical models were actively employed in the target selection process. The most common expression I hear in such cases is, “The model didn’t work.” But when I dig through the whole process, I often find that the model algorithm is the only error-free item. How ironic.

I’ve talked about “analytics-readiness” so many times already. And, yes, inadequate sets of input data can definitely ruin models. So allow me to summarize ways users wreck perfectly adequate models “after” they were developed and validated. And there are many ways you can do that, unfortunately. Allow me to introduce a few major ones.

Using the Model in a Wrong Universe

Without a doubt, setting a wrong target will lead to an unusable model. Now, an equally important factor as the “target definition” is the “comparison universe.” If you are building a response model, for example, responders (i.e., targets) will be compared to non-responders (i.e., non-targets). If you are off in one of those, the whole model will be wrong — because a model is nothing but a mathematical expression of differences between the two dichotomous groups. This is why setting a proper comparison universe — generally, a sample out of the pool of names that you are using for the campaign — is equally as important as setting the right target.

Further, let’s say that you want to use models within preset universes, based on region, age, gender, income, past spending level, certain number of clicks, or any other segmentation rules. Such universe definitions — mostly about exclusion of obvious non-targets — should be determined “before” the model development phase. When such divisions are made, applying the model built for one universe (e.g., a regional model for the Mid-Atlantic) to another universe (e.g., the Pacific Northwest region) will not provide good results, other than with some dumb luck.

Ignoring the Design Principle of the Model

Like buildings or cars, models are built for specific purposes. If I may list a few examples:

  • “Future customer value estimation, in dollars”
  • “Propensity to purchase in response to discount offers via email”
  • “Product affinity for a certain product category”
  • “Loyalty vs. churn prediction”
  • “Likelihood to be a bargain-seeker”
  • Etc.

This list could be as long as what you want as a marketer.

However, things start to go wrong when the user starts ignoring (or forgetting) the original purpose of the model. Years back, my team built a model for a luxury cruise line for a very specific purpose. The brand was very reputable, so it had no trouble filling in staterooms with balconies at a higher price point. But it did have some challenges filling in inside staterooms at a relatively high price of entry, which was equivalent to a window room on a less fancy ship. So, the goal was to find cruisers who would take up inside staterooms, for the brand value, on Europe-bound ships that depart U.S. ports between Thanksgiving and Christmas. A very specific target? You bet.

Troubles arose because it worked all too well for the cruise line. So, without any further consultation with any analysts, they started using that model for other purposes. We got phone calls only after the attempt failed miserably. Now, is that really the fault of the model? Sure, you can heat up your house with a kitchen oven, but don’t blame the manufacturer when it breaks down by abusing it like that. I really don’t think the warranty applies there.

Playing With Selection Rules

Some marketers are compelled to add more rules after the fact, probably out of sheer enthusiasm for success. For instance, a person in charge of a campaign may come up with an idea at the last minute, and add a few rules on top of the model selection, as in “Let’s send mails only to male prospects in the high-score group.” What this means is that he just added the strongest variable on top of a good model, which may include 15 to 20 variables, all carefully weighted by a seasoned statistician. This type of practice may not lead to a total disaster, but the effectiveness of the model in question is definitely diluted by the post-selection rules.

When the bad results start to come in, again, don’t blame the modeler for it. Because “you” essentially redesigned the model by adding new variables on top of existing predictors. Unfortunately, this type of last-minute meddling is quite common. If you have a good reason to do any “post-selection,” please talk to the analyst before the model is built, so that she can incorporate the rule as a “pre-selection” logic. She may give you multiple models fitted for multiple universes, too.

Realigning the Model Groups in Arbitrary Ways

Model scores are just long numbers — with eight or nine decimal places, in general. It is hard to use sheer numeric values like that, so kind modelers generally break the scored universe into 10 or 20 equal groups. (We call them decile or demi-decile groups.)

For instance, each decile group would represent 10% of the development and validation samples. When applied to the campaign universe, resultant score groups should not deviate too much from that 10% mark.

If you see big bumps in model group sizes, it is a clear sign that something went wrong in scoring, there were significant changes in the input variables, or the model is losing its effectiveness, over time.

I’ve seen cases where users just realigned the model score groups after the fact, simply because groups were not showing an equal 10% break anymore. That is like covering serious wounds with a make-up. Did the model work after that? Please take a wild guess.

Using Expired Models

Models do have limited shelf-lives. Models lose their predictive power over time, as market conditions, business models, data sources, data procurement methods, and target profiles all inevitably go through changes.

If you detect signs of lagging results or wide fluctuations in model group distribution (i.e., showing only 3% in the top decile, which is supposed to be around 10%), it is time to review the model. In mild cases, modelers may be able to refit the model. But in this day and age of fast computers and automation, I recommend full redevelopment of the model in question at the first sign of trouble.

Ignoring the ‘Level’ of Prediction

A model for target marketing is to rank the records from high to low scores, according to the design principle. If you built an affinity model for “Likely to be an early adopter,” high score means the target is more likely to be an early adopter, and low score means she’s less likely to be one. Now, the level of the record matters here. What are you really ranking, anyway?

The most common ones are individual and household levels. It is possible to build a model on an email level, as one individual may have multiple email addresses. If you are in a telecom business, you may not even care for the household-level identity, as the “house” may be the target, regardless of who lives in there.

In the application stage, matching the “level” of prediction is important. For household models, it is safe to assume that almost all predictors in the model are on a household level. Applying such models on a different level may negatively affect the model performance. A definite “no” is using household-level score for an address, not knowing who lives there. One may think “How different will the new mover be from the old resident?” But considering a wide variety of demographic variables commonly used in models, it is something that no modeler would recommend. If the model employed any transaction or behavioral data, don’t even think about switching levels like that. You’d be better off building a regional model (such as ZIP model) only using geo-demographic data.

Applying Average Scores to Non-Matches or Non-Scorable Records

Sometimes, scores are missing because of non-matches in the data append process, or strict universe definition using pre-selection rules. It can be tempting to apply some “average” score to cover the missing ones, but that is a big no-no, as well. Statisticians may perform such imputation on a variable level to fill missing values, but not with model scores.

If you really have to have a score for every record, build separate models for non-match or non-select universes, using any available data (if there are any to be used). In CRM models, no one should just drop non-matches into demographic files, as the main drivers of such models would be transaction and behavioral data. Let missing values play out in the model (refer to “Missing Data Can Be Meaningful”).

For prospecting, once you set up a pre-selection universe (hopefully, after some profile analysis), don’t look back and just go with a “scored” universe. Records with missing scores are generally not salvageable, in practice.

Go Forth and Do Good Business With Models, Marketers

As you can see, there are many ways to mess up a good model. A model is not an extension of rudimentary selection rules, so please do NOT treat it that way. Basically, do not put diesel fuel in a gasoline car, and hope to God that the engine will run smoothly. And when — not if — the engine stalls, don’t blame the engineer.

Models may be built, but the work is not nearly done until they are properly applied and deployed in live campaigns. When in doubt, always consult with the analyst in charge; hopefully, before the drop date.

Have We Ruined 1:1 Marketing? How the Corner Grocer Became a Creepy Intruder

When Don Peppers and Martha Rogers wrote “The One to One Future: Building Relationships One Customer at a Time” in 1993, the Internet was a mere twinkle in Al Gore’s eye. But direct marketers felt excited about 1:1 marketing, and even vindicated.

When Don Peppers and Martha Rogers wrote “The One to One Future: Building Relationships One Customer at a Time” in 1993, the Internet was a mere twinkle in Al Gore’s eye. But direct marketers felt excited, and even vindicated, about the promise of a future where data-driven personalization would deliver the right message to the right customer at the right time.

But now that it’s here, are consumers happy with it?

Recently, I had the students in my direct marketing course at Rutgers School of Business read the introduction to “The Complete Database Marketer” by Arthur Hughes, which was published in 1996 when only 22% of people in the U.S. had Internet access. In the intro entitled “The Corner Grocer,” Hughes explains how database marketing can connect marketers with their customers with the same personal touch that the corner grocer had by knowing all of his customers’ names, family members, and usual purchases.

The students then had to compare the 1996 version of database marketing, as described by Hughes, with the current state of online direct/database marketing, where data collection has been enabled by e-commerce, social media, and search engine marketing.

  • What marketing innovations has technology enabled that didn’t exist before?
  • How has online marketing enhanced the concept of database marketing?
  • How have new marketing techniques and technologies changed consumer behavior?
  • How has social media affected direct/data-driven marketing for the marketer and the consumer?
  • What are some of the fundamental differences between the challenges and opportunities that today’s online marketers face vs. those that the 1996 database marketer faced?

Most of these digital natives were born after Hughes’s book was published. The students experience digital marketing every day, and they’ve seen it evolve over their lifetimes. While they concede that the targeted ads they experience are usually relevant, several of them noted that they don’t feel they have been marketed to as individuals; but rather, as a member of a group that was assigned to receive a specific digital advertisement by an algorithm. They felt that the idealized world of database marketing that Hughes described in 1996 was actually more personal than the advanced algorithmic targeting that delivers ads to their social media feeds. Hughes told the tale of Sally Warner and her relationship with the St. Paul’s Luggage Company that started with returning a warranty card and progressed with a series of direct mail and telemarketing. For example, knowing that Sally Warner had a college-bound son, St. Paul’s sent a letter suggesting luggage as a graduation gift. Hughes describes the concept of database marketing:

“Every contact with the customer will be an opportunity to collect more data about the customer. This data will be used to build knowledge about the customer. The knowledge will be used to drive strategy leading to practical, directly personal, long-term relationships, which produce sales. The sales, in turn, will yield more data which will start the process all over again.”

But Arthur couldn’t foresee the data collection capabilities of Google, Facebook, Instagram, and Amazon. Instead of the friendly corner grocer, database marketers have become a creepy intruder. How else could an ad for a product my wife had searched for at Amazon on her laptop generate an ad for the same product in my Instagram feed? (Alright, I will concede that we use the same Amazon Prime membership, but really?) We don’t have a smart speaker in the house, and I dread to think about how much creepier it could become if we did.

Recently, while visiting someone who has a Google Home assistant, I asked about the level of spying they experienced in exchange for the convenience of having voice-activated control over their household lights and appliances. They responded by asking, “Google, are you spying on us?”

The smart speaker replied, “I don’t know how to answer that question.”

Have we ruined 1:1 marketing?

Do you know how to answer that question? Tell me.

Factors for Marketers to Consider in Attribution Rules

At the end of each campaign effort, a good database marketer is supposed to study “what worked, and what didn’t,” using attribution rules. Call it “Back-end Analysis” or “Campaign Analytics.” Old-timers may use terms like “Match-back.” Regardless, it is one of the most important steps in 1:1 marketing that is synonymous with what we used to call “Closed-loop Marketing.”

At the end of each campaign effort, a good database marketer is supposed to study “what worked, and what didn’t,” using attribution rules. Call it “Back-end Analysis” or “Campaign Analytics.” Old-timers may use terms like “Match-back.” Regardless, it is one of the most important steps in 1:1 marketing that is synonymous with what we used to call “Closed-loop Marketing.” (refer to my first article on Target Marketing from 11 years ago, “Close the Loop Properly”).

In fact, this back-end analysis is so vital that if one skips this part of analytics, I can argue that the offending marketer ceases to be a 1:1 or database marketer. What good are all those databases and data collection mechanisms, if we don’t even examine campaign results? If we are not to learn from the past, how would we be able to improve results, even in the immediate future? Just wild guesses and gut feelings? I’ve said it many times, but let me say it again: Gut-feelings are overrated. Way more overrated than any cheesy buzzword that summarizes complex ideas into one or two catchy words.

Anyhow, when there were just a few dominant channels, it wasn’t so difficult to do it. For non-direct channel efforts, we may need some attribution modeling to assign credit for each channel. We may call that a “top-down” approach for attribution. For direct channels, where we would know precisely who received the offers, we would do a match-back (i.e., responders matched to the campaign list by personally identifiable information, such as name, address, email, etc.), and give credit to the effort that immediately preceded the response. We may call that a “bottom-up” method.

So far, not so bad. We may have some holes here and there, as collecting PII from all responders may not be feasible (especially in retail stores). But when there was just direct mailing as “the” direct channel, finding out what elements worked wasn’t very difficult. Lack of it was more of a commitment issue.

Sure, it may cost a little extra, and we had to allocate those “unknown” responders through some allocation rules, but backend analysis used to be a relatively straightforward process. Find matches between the mailing (or contact) list and the responders, append campaign information — through what we used to call “Source Code” — to each responder, and run reports by list source, segment, selection mechanism, creative, offer, drop date and other campaign attributes. If you were prudent to have no-mail control cells in the mix, then you could even measure live metrics against them. Then figure out what worked and what didn’t. Some mailers were very organized, and codified all important elements in those source codes “before” they dropped any campaigns.

Now we are living in a multi-channel environment, so things are much more complicated. Alas, allocating each of those coveted responses to “a” channel isn’t just technical work; it became a very sensitive political issue among channel managers. In the world where marketing organizations are divided by key marketing channels (as in, Email Division vs. Direct Mail Division), attribution became a matter of survival. Getting “more” credit for sales isn’t just a matter of scientific research, but a zero-sum game to many. But should it be?

Attribution Rules Should Give Credit Where Credit’s Due

I’ve seen some predominantly digital organizations giving credit to their own direct marketing division “after” all digital channels took all available credit first. That means the DM division must cover its expenses only with “incremental” sales (i.e., direct-mailing-only responses, which would be as rare as the Dodo bird in the age of email marketing). Granted that DM is a relatively more expensive channel than email, I wish lots of luck to those poor direct marketers to get a decent budget for next year. Or maybe they should look for new jobs when they lose that attribution battle?

Then again, I’ve seen totally opposite situations, too. In primarily direct marketing companies, catalog divisions would take all the credit for any buyer who ever received “a” catalog six months prior to the purchase, and only residual credit would go to digital channels.

Now, can we at least agree that either of these cases is far from ideal? When the game is rigged from the get-go, what is the point of all the backend analyses? Just a façade of being a “data-based” organization? That sounds more like a so-called “free” election in North Korea, where there are two ballot boxes visibly displayed in the middle of the room; one for the Communist Party of the Dear Leader, and another box for all others. Good luck making it back home if you put any ballot in the “wrong” box.

Attribution among different channels, in all fairness, is a game. And there is no “one” good way to do it, either. That means an organization can set up rules any way it wants them to be. And as a rule I, as a consultant, tend not to meddle with internal politics. Who am I to dictate what is the best attribution rule for each company anyway?

Here’s How I Set Up Attribution Rules

Now that I am a chief product guy for an automated CDP (Customer Data Platform) company, I got to think about the best practices for attribution in a different way. Basically, we had to decide what options we needed to provide to the users to make up attribution rules as they see fit. Of course, some will totally abuse such flexibility and rig the game. But we can at least “guide” the users to think about the attribution rules in more holistic ways.

Such consideration can only happen when all of the elements that marketers must consider are lined up in front of them. It becomes difficult to push through just one criterion — generally, for the benefit of “his” or “her” channel — when all factors are nicely displayed in a boardroom.

So allow me to share key factors that make up attribution rules. You may have some “A-ha” moments, but you may also have “What the … ” moments, too. But in the interest of guiding marketers to unbiased directions, here is the list:

Media Channel

This is an obvious one for “channel” attribution. Let’s list all channels employed by the organization, first.

  • Email
  • Direct Mail (or different types of DM, such as catalog, First Class mail, postcards, etc.)
  • Social Media (and specific subsets, such as Facebook, Instagram, etc.)
  • Display Ads
  • Referrals/Affiliates
  • Organic Search/Paid Search
  • Direct to Website (and/or search engines that led the buyers there)
  • General Media (or further broken down into TV, Radio, Print, Inserts, etc.)
  • Other Offline Promotions
  • Etc.

In case there are overlaps, which channel would take the credit first? Or, should “all” of the responsive channels “share” the credit somehow?

Credit Share

If the credit — in the form of conversions and dollars — is to be shared, how would we go about it?

  • Double Credit: All responsible channels (within the set duration by each channel) would get full credit
  • Equal Split: All contributing channels would get 1/N of the credit
  • Weighted Split: Credit divided by weight factors set by users (e.g., 50% DM, 30% EM, 20% General Media, etc.)

There is no absolutely fair way to do this, but someone in the leadership position should make some hard decisions. Personally, I like the first option, as each channel gets to be evaluated in pseudo-isolation mode. If there was no other channel in the mix, how would a direct marketing campaign, for example, have worked? Examine each channel and campaign this way, from the channel-centric point of view, to justify their existence in the full media mix.

Allocation Method

How will the credit be given out with all of those touch data from various tags? There are a few popular ways:

  • Last Touch: This is somewhat reasonable, but what about earlier touches that may have created the demand in the first place?
  • First Touch: We may go all of the way back to the first touch of the responder, but could that be irrelevant by the time of the purchase? Who cares about a Christmas catalog sent out in November for purchases made in May of the next year?
  • Direct Attribution: Or should we only count direct paths leading to conversions (i.e., traceable opens, clicks and conversions, on an individual level)? But that can be very limiting, as there will be many untraceable transactions, even in the digital world.
  • Stoppage: In the journey through open, click and conversion, do we only count conversions, or should the channel that led to opens and clicks get partial credit?

All of these are tricky decisions, but marketers should not just follow “what has been done so far” methods. As more channels are added to the mix, these methods should be reevaluated once in a while.

Time Duration (by Channel)

Some channels have longer sustaining power than others. A catalog kept in a household may lead to a purchase a few months later. Conversely, who would dig out a promotional email from three weeks ago? This credit duration also depends on the type of products in question. Products with long purchase cycles — such as automobiles, furniture, major appliances, etc. — would have more lasting effects in comparison to commodity or consumable items.

  • Email: 3-day, 7-day, 15-day, 30-day, etc.
  • Direct Mail — Catalog: 30-day, 60-day, 90-day, etc.
  • Direct Mail — Non-catalog: 7-day, 14-day, 30-day, 60-day, etc.
  • Social: 3-day, 7-day, 15-day, etc.
  • Direct Visit: No time limit necessary for direct landing on websites or retail stores.
  • General Media: Time limit would be set based on subchannels, depending on campaign duration.

Closing Thoughts

The bottom line is to be aware of response curves by each channel, and be reasonable. That extra 30-day credit period on the tail end may only give a channel manager a couple extra conversions after all of the political struggles.

There is really no “1” good way to combine all of these factors. These are just attribution factors to consider, and the guideline must be set by each organization, depending on its business model, product composition and, most importantly, channel usages (i.e., how much money bled into each channel?).

Nevertheless, in the interest of creating a “fair” ground for attributions, someone in a leadership position must set the priority on an organizational level. Otherwise, the outcome will always favor what are considered to be traditionally popular channels. If the status quo is the goal, then I would say skip all of the headaches and go home early. You may be rigging the system — knowingly or unknowingly — anyway, and there is no need to use a word like “attribution” in a situation like that.

Use People-Oriented Marketing: Because Products Change, But People Rarely Do

In 1:1 marketing, product-level targeting is “almost” taken for granted. I say almost, because most so-called personalized messages are product-based, rarely people-oriented marketing. Even from mighty Amazon, we see rudimentary product recommendations as soon as we buy something. As in: “Oh, you just bought a yoga mat! We will send you absolutely everything that is related to yoga on a weekly basis until you opt out of email promotions completely. Because we won’t quit first.”

In 1:1 marketing, product-level targeting is “almost” taken for granted. I say almost, because most so-called personalized messages are product-based, rarely people-oriented marketing. Even from mighty Amazon, we see rudimentary product recommendations as soon as we buy something. As in: “Oh, you just bought a yoga mat! We will send you absolutely everything that is related to yoga on a weekly basis until you opt out of email promotions completely. Because we won’t quit first.”

How nice of them. Taking care of my needs so thoroughly.

Annoying as they may be, both marketers and consumers tolerate such practices. For marketers, the money talks. Even rudimentary product recommendations — all in the name of personalization — work much better than no targeting at all. Ain’t the bar really low here, in the age of abundant data and technologies? Yes, such a product recommendation is a hit-or-miss, but who cares? Those “hits” will still generate revenue.

For consumers, aren’t we all well-trained to ignore annoying commercials when we want to? And who knows? I may end up buying a decent set of yoga mat cleaners with a touch of lavender scent because of such emails. Though we all know purchase of that item will start a whole new series of product offerings.

Now, marketers may want to call this type of collaborative filtering an active form of personalization, but it isn’t. It is still a very reactive form of marketing, at the tail end of another purchase. It may not be as passive as waiting for someone to type in keywords, but product recommendations are mixture of reactive and active (because you may send out a series of emails) forms of marketing.

And I’m not devaluing such endeavors, either. After all, it works, and it generates revenue. All I am saying is that marketers should recognize that a reactive product recommendation is only a part of personalization efforts.

As I have been writing for five years now, 1:1 marketing is about effectively deciding:

  1. whom to contact, and
  2. what to offer.

Part One is good old targeting for outbound efforts, and there are a wide variety of techniques for it, starting with rules that marketers made up, basic segmentation, and all of the way to sophisticated modeling.

The second part is a little tricky; not because we don’t know how to list relevant products based on past purchases, but because it is not easy to support multiple versions of creatives when there is no immediate shopping basket to copy (like cases for recent purchases or abandoned carts).

In between unlimited product choices and relevant offers, we must walk the fine lines among:

  1. dynamic display technology,
  2. content and creative library,
  3. data (hopefully clean and refined), and
  4. analytics in forms of segments, models or personas (refer to “Key Elements of Complete Personalization”).

If specific product categories are not available (i.e., a real indicator that a buyer is interested in certain items), we must get the category correct at the minimum, using modeling techniques. I call it personas, and some may call it architypes. (But they are NOT segments. Refer to “Segments vs. Personas”).

Using the personas, it is not too difficult to map proper products to potential buyers. In fact, marketers are free to use their imaginations when they do such mapping. Plus, while inferred, these model scores are never missing, unlike those hard-to-get “real” data. No need to worry about targeting only a small part of potential buyers.

What should a marketer offer to fashionistas? To trendsetters? To bargain seekers? To active, on-the-go types? To seasonal buyers? To big spenders? Even for a niche brand, we can create 10 to 20 personas that represent key product categories and behavioral types, and the deployment of personalized messages become much simpler.

And it gets better. Imagine a situation where you have to launch a new product or a product line. It gets tricky for the fashion industry, and even trickier for tech companies that are bold enough to launch something that didn’t exist before, such as a new line of really expensive smartphones. Who among the fans of cutting-edge technologies would actually shell out over a grand for a “phone”? This kind of question applies not just to manufacturers, but every merchant who sells peripherals for such phones.

Let’s imagine that a marketer would go with an old marketing plan for “similar” products that were introduced in the past. They could be similar in terms of “newness” and some basic features, but what if they differ in terms of specific functionality, look-and-feel, price point and even the way users would use them? Trying to copy some old targeting methods may lead to big misses, as even consumers hear about them from time to time.

Such mishaps happen because marketers see consumers as simple extensions of products. Pulling out old tricks may work in some cases, but even if just a small bit of product attributes are different, it won’t work.

Luckily for geeks like us, an individual’s behavior does not change so fast. Sure, we all age a bit every year; but in comparison to products in the market, humans do not transform so suddenly. Simply, early adapters will remain early adapters, and bargain seekers will continue to be bargain seekers. Spending level on certain product categories won’t change drastically, either.

Our interests and hobbies do change; but again, not so fast. It took me about two to three years to turn from an avid golfer to a non-golfer. And all golf retailers caught up with my inactivity and stopped sending golf offers.

So, if marketers set up personas that “they” need to push their products, and update them periodically (say once a year), they can gain tremendous momentum in reaching out to customers and prospects more proactively. If they just rely on specific product purchases to trigger a series of product recommendations, outreach programs will remain at the level of general promotions.

Further, even inbound visits can be personalized better (granted that you identified the visitor) using the personas and set of rules in terms of what product goes well with what persona.

The reason why models work well — man-made or machine-built — is because human behavior is predictable with reasonable consistency. We are all extensions of our past behaviors to a greater degree than the evolution rate of products and technologies.

Years ago, we’ve had a heated internal discussion about whether we should create a new series of product categories from VHS to DVD. I argued that such new formats would not change human behavior that much. In fact, genres matter more than video format for the prediction of future purchases. “Godfather” fans will buy the movie again on DVD, and then again in Blu-ray. Now some type of ultra-high-definition download from some cloud somewhere. Through all of this, movie collectors remain movie collectors for their favorite types of movies. In other words, products changed, but not human attributes.

That was what I argued then, and I still stand by it. So, all the analytical efforts must be geared toward humans, not products. In coming days, that may be the shortest path to fake human friendliness using AI and machine-made models.

 

Election Polls and the Price of Being Wrong 

The thing about predictive analytics is that the quality of a prediction is eventually exposed — clearly cut as right or wrong. There are casually incorrect outcomes, like a weather report failing to accurately declare at what time the rain will start, and then there are total shockers, like the outcome of the 2016 presidential election.

screen-shot-2016-11-17-at-1-03-34-pmThe thing about predictive analytics is that the quality of a prediction is eventually exposed — clearly cut as right or wrong. There are casually incorrect outcomes, like a weather report failing to accurately declare the time it will start raining, and then there are total shockers, like the outcome of the 2016 presidential election.

In my opinion, the biggest losers in this election cycle are pollsters, analysts, statisticians and, most of all, so-called pundits.

I am saying this from a concerned analyst’s point of view. We are talking about colossal and utter failure of prediction on every level here. Except for one or two publications, practically every source missed the mark by more than a mile — not just a couple points off here and there. Even the ones who achieved “guru” status by predicting the 2012 election outcome perfectly called for the wrong winner this time, boldly posting a confidence level of more than 70 percent just a few days before the election.

What Went Wrong? 

The losing party, pollsters and analysts must be in the middle of some deep soul-searching now. In all fairness, let’s keep in mind that no prediction can overcome serious sampling errors and data collection problems. Especially when we deal with sparsely populated areas, where the winner was decisively determined in the end, we must be really careful with the raw numbers of respondents, as errors easily get magnified by incomplete data.

Some of us saw that type of over- or under-projection when the Census Bureau cut the sampling size for budgetary reasons during the last survey cycle. For example, in a sparsely populated area, a few migrants from Asia may affect simple projections like “percent Asians” rather drastically. In large cities, conversely, the size of such errors are generally within more manageable ranges, thanks to large sample sizes.

Then there are human inconsistency elements that many pundits are talking about. Basically everyone got so sick of all of these survey calls about the election, many started to ignore them completely. I think pollsters must learn that at times, less is more. I don’t even live in a swing state, and I started to hang up on unknown callers long before Election Day. Can you imagine what the folks in swing states must have gone through?

Many are also claiming that respondents were not honest about how they were going to vote. But if that were the case, there are other techniques that surveyors and analysts could have used to project the answer based on “indirect” questions. Instead of simply asking “Whom are you voting for?”, how about asking what their major concerns were? Combined with modeling techniques, a few innocuous probing questions regarding specific issues — such as environment, gun control, immigration, foreign policy, entitlement programs, etc. — could have led us to much more accurate predictions, reducing the shock factor.

In the middle of all this, I’ve read that artificial intelligence without any human intervention predicted the election outcome correctly, by using abundant data coming out of social media. That means machines are already outperforming human analysts. It helps that machines have no opinions or feelings about the outcome one way or another.

Dystopian Future?

Maybe machine learning will start replacing human analysts and other decision-making professions sooner than expected. That means a disenfranchised population will grow even further, dipping into highly educated demographics. The future, regardless of politics, doesn’t look all that bright for the human collective, if that trend continues.

In the predictive business, there is a price to pay for being wrong. Maybe that is why in some countries, there are complete bans on posting poll numbers and result projections days — sometimes weeks — before the election. Sometimes observation and prediction change behaviors of human subjects, as anthropologists have been documenting for years.