Tag: machine learning

Posted on

Machines Won’t Take Over…But A Few AI Titans Might

Lately, if you’re like me and enjoy following the AI narrative (even if just for grins & giggles), you’re inevitably sucked into philosophical wormholes that always seem to pop you out at the same place – a world where machines rule all.

AI titans
Tech Titans

Strangely, though, we rarely encounter future scenarios that follow a path we’re already on, where machines are but tools used to assist us. If we project this scene forward, some interesting questions to ask are, “What does that world look like, and who are its haves and have-nots?  Are AI titans forming?”

AI, for all its hype and promise, is still very much in its infancy.  Far from being able to get up, put on its clothes, and take your job, AI today is less of a super scary robot, and more like a smart washing machine (funny you should ask, as there is one of those).  It can help us conserve resources and do specialized tasks more efficiently, like getting clothes clean using fewer resources, but it really can’t do higher order thinking we take for granted like abstract judgement and reasoning. However, that super smart washing machine (and all its other specialized variants) has an owner, and together they can wield tremendous influence.  And anti-trust laws (put in place over 100 years ago to prevent corporate behemoths from controlling entire markets) may be full of loop holes in the digital age.

Using a singularity argument where machines alone rule provides a convenient escape from a more complex debate about a future where various human and machine forces collide and collapse together.  In this scenario, a select set of firms use walled garden data to feed their AI, and as such, seize unprecedented levels of control, influence, and power.

Here’s an example.  We’re already seeing a massive rationalization of power and influence collapsing into AI titans like Google, Facebook, Apple, Microsoft, and Amazon (controlled by surprisingly few individuals); not pure machines, but formidable entities nonetheless, fueled by AI, and directed by small pools of mighty people already circling their wagons around a plethora of data.

In the short run, we (the consumers) seem to benefit, getting innovative little features and conveniences such as travel guidance and digital yellow pages, but unbeknownst to most, to get these we sacrifice gobs of data and hence privacy.  Each time we travel with GPS on, our whereabouts are tracked and stored.  Each time we search, we provide preference footprints.  Meanwhile, the behemoths rack the data up, building behavior and preference repositories on each of us.

So what’s the rub?

First, it’s our data.  Thus, it would be nice to be able to view it, and if it’s wrong, correct it.  The European Union passed a law recently that goes into effect in May 2018 called GDPR – General Data Protection Regulation.  Its intent is to give consumers more rights and transparency with their digital data.  Other consumers outside the EU could use similar privacy protection laws.

Second, to some extent, without being cognizant of it, our choices are already being limited.  For example, when you search in digital maps, perform online comparison-shopping, or ask a voice pod for restaurant recommendations, the top options returned may not be calculated objectively.  Ranking algorithms already place higher emphasis on businesses that pay more to play, and search conglomerates, like Google, rank their interests (including businesses they have a stake in) higher.

Each time we purchase something, we’re casting a vote.  When we go through a buying cycle, we are creating implied demand, and when we purchase we’re reinforcing that the supply is meeting the demand we created. When this cycle is cornered, choice becomes an illusion.  To illustrate, on June 27, 2017 the EU slapped Google with a record-breaking $2.7 billion fine, charging the tech titan with doctoring search results giving an “illegal advantage” to its interests while harming its rivals.

Third, firms can and will use your data for their benefit, and not necessarily yours.  Prior to the digital age, people stereotyped others by their physical choices such as their house, car, job, shopping habits, and clothes.  Although today those choices still factor in, we also project digital personas: where we surf, what we share and like on Facebook and Instagram, what devices and channels we use, how we interact online, and so forth.  When these behaviors are crunched and codified, they become rich fuel for algorithms that can manipulate, discriminate, or even do harm, without the algorithm’s owners having any concerns for side or after effects.  Show preference for fast cars and thrill-seeking vacations, and not only will you receive more of those offers, but you might also receive higher insurance premiums.  Share enough medical history, and an insurer’s algorithm may score you at high risk for a chronic disease, even when there’s no medical diagnosis, and there’s no certainty you’ll ever develop that condition.  That might make it very hard to get medical coverage.

Admittedly, not all of the use cases lead to undesirable outcomes.  In late 2016, American Banker ran an article on next-gen biometrics detailing how banks use consumer digital behavior signatures to detect fraud and protect consumers from its effects.  And although consumers initially do benefit from such a service, what’s interesting (and concerning) is the nature of the behavior data fed to the fraud detection algorithm:  the angle at which the operator typically holds the smartphone, pressure levels on the touch screen, and cadence of keystrokes.

Unquestionably, the bank’s primary goal is predicting whether an imposter is behind the device in question.  Nonetheless, what’s stopping this same bank from using that data to predict a consumer’s likely mental state, such as likelihood of inebriation, legal or otherwise?  Moreover, whether that prediction is ultimately accurate is irrelevant to the immediate recommended action and the subsequent consequences.  We have little protection from the effects of algorithmic false positives, and today, except for credit scores, few brands have any accountability for model scoring accuracy.

Here’s a scenario.  An algorithm thinks you’ve been drinking based on your smartphone behavior and flags you as too drunk to drive and disables your car, forcing you to find another way home.  That’s one thing, but think about this – that same data might also be available to prospective employers, who use it to forecast your job performance, scoring you lower than other candidates based on its dubious drug use prediction.

Who owns and manages your digital behavior data?  Are they subject to use restrictions? The answer is (although the data is about your profile and your behavior) – you don’t own it and your rights are limited. And although some of the more inconsequential data is scattered about (such as name, address, date of birth, and so on), the deeper behavioral insights are amassed, stored, and crunched by the AI titans, with seemingly no limits or full transparency, and with little insight into where its shipped, and who else might eventually use it.  They suggest we simply trust them.

Those that ignore history are doomed to repeat it

History is always an amazing teacher.  In the 19th century, railroads consolidated into monopolies that controlled the fate of other expanding industries, such as iron, steel, and oil.  They dominated the distribution infrastructure – just as today’s AI titans, in many respects, control the lifeblood of modern day companies – their prospect and customer traffic.  And those other expanding industries (iron, steel, oil) were no different.  They too controlled the fate of other expanding industries, which all needed their materials.

Soon after their start, Google’s founders adopted a mantra, “Don’t be evil.”  In October 2015, under the new parent company Alphabet, that changed to “Do the right thing.”  Although the revised phrase still rings with the implication of justice, it raises the question of who benefits from that justice, and if there’s a disguised internal trust forming.

Everyone knows that business, by its very nature, is profit driven.  There’s nothing wrong with that, yet history teaches us that we need checks and balances to promote a level playing field for other competitors or potential entrants, and for consumers.

History Lesson

In his 1998 book “The Meaning of it All,” Richard Feynman, a famous scientist, tells a story of entering a Buddhist temple and encountering a man giving sage advice.  He said, “To every man is given the key to the gates of heaven. The same key opens the gates of hell.”  Unpacked and applied to AI today:

  1. The term “every man” can imply an individual, or organization made of people, or humankind as a whole.
  2. Science, technology, data, and artificial intelligence are but tools. As history shows, humans use them for good and evil purposes.
  3. AI’s impact on the future isn’t pre-determined. Each of us can play a role in shaping how it turns out.

Let’s ensure we live in a world where many (not a select few) benefit from AI’s capacity and ability to improve lives, and that those responsible for its development, evolution, and application are held to fair and ethical standards.

Can AI be the rising tide that lifts all boats?

The power and potential of artificial intelligence technologies is clear, yet our ability to control it, and deploy it sustainably is not.  Who should regulate and control it (and its fuel- our data) is an evolving and ongoing debate.

Used responsibly and applied democratically, we all stand to benefit from AI.  Paradoxically, while it renders some of our old jobs obsolete, it retrains us for a new world where it and we play new and more rewarding roles – where living standards rise and mortality rates fall.

What’s our guarantee we’re marching toward that future?

Honestly, there are no guarantees – our world is devoid of certainty.  However, we can influence likely outcomes by advocating for practical checks and balances.  Call me a dreamer, but I envision a world where our privacy is valued and respected.  Where we better understand the value of our data and get a reasonable exchange in return when we share it. Where we appreciate what happens when we release it, and can hold those accountable that illegally mangle or pawn it; and a world where we have assurance that when we share data, others uphold their end of the agreement, and we have recourse if they don’t.

If you would like to continue contemplating some of the top ethical implications of AI’s evolving story, click on this link:

https://www.weforum.org/agenda/2016/10/top-10-ethical-issues-in-artificial-intelligence/

Here’s my favorite quote from it:

“If we succeed with the transition, one day we might look back and think that it was barbaric that human beings were required to sell the majority of their waking time just to be able to live.”

Peace

Posted on

Customer Engagement – From BI Guesswork to Prescriptive AI

Customer Engagement approaches, and the technology used to enable them, have evolved immensely over the last 25 years.  Two distinct eras define this period, as well as a major technological shift to real-time systems with AI feedback loops.

Prescriptive AI

The BI Guesswork Era

During the advent of the Business Intelligence (BI), Marketing Technology and Campaign Management era (circa 1990), marketers had limited predictive powers.  In many cases, when it came to what individuals really needed, they resorted to guesswork.  They channeled their energy to perfect efficiencies in targeting and automation.  Their main emphasis was finding an approximate audience for products so they designed promotions for large segments of the population. They fixated on finding segments that fit into certain “likelihood to respond” buckets, and then repeatedly tested timing, messages, and creative content by peppering those segments with treatments.  In other words, they identified massive groups, matched offers to these groups, and then used technology to systematize their marketing.

Although some of those marketers drew on basic models (such as RFM – Recency, Frequency, Monetary), which provided rough guidance on how deep to mail into a file, most didn’t even do this.  Typical response rates were 0.5% at best.  During this period, the average adult was receiving about 50 pounds of junk mail a year – coined junk mail because the promotions were irrelevant 99.5% of the time.  Thus, the majority viewed this activity as frivolous, mocking it with nicknames and jokes.  Regardless, marketers were unrelenting as they continually carpet-bombed until consumers either responded or learned how to opt-out.

Their tools of choice were crude in nature.  They were slow, not fine-grained, and certainly not customer-centric.  Usually, the campaign flowcharts they devised utilized basic analytics where deterministic queries ran against databases returning huge customer lists called segments.  If there was any further segment refinement, they relied on business intelligence technologies like OLAP (Online Analytical Processing) and dashboards to support their intuition.  Even as some of the more sophisticated marketers attempted predictions, providing those models with feedback was nearly impossible due to the batch processing nature of the flows and platforms they employed.  As shown in Figure 1, although some crept up the analytics value chain toward being predictive and answering the question “What will happen?” most fell short.

Figure 1:

business intelligence

Source: http://www.bi-bestpractices.com/view-articles/5642

Using a backward approach, engineers pre-developed the product, and marketers wrangled the packaging, promotions, and messaging to the audience – again using more guesswork than analytics.  It was difficult to react contextually, at scale, to actual individual needs, so instead they focused on groups of customers.

And so they executed bulk outbound communications at scale. With promotional ammunition in hand, readily available data afforded them reasonable targeting coordinates, and computers and devices served as the delivery mechanisms. The marketplace and emerging technology supported a numbers game and rewarded short-term economic gains.  Longer-term loyalty and longitudinal effects took a back seat.

By the turn of the century, direct marketers were plodding ahead using ever-richer consumer profiles that enabled them to focus promotions on increasingly smaller segments.  And even though in 1995, Peppers & Rogers had coined the term “1:1 marketing,” enterprise marketers were no where near direct conversations with individual consumers.  Still constrained by scale, they were stuck communicating to segments, albeit smaller and smaller ones.  What they didn’t realize was they were about to hit a wall (Figure 2)

Figure 2:

Real-Time Evolution

By 2005, marketers had the tools to perform hyper-targeting.  They aggressively tested different incentives, creative elements, and fine tuned things based on response metrics.   Scoring models were refined, though the expense was large, and the iterations long.  The results didn’t so much alter someone’s behavior, but more provided alternatives to consider, often ones that still had borderline relevance to a current need.

Often the goal, instead of steadfast loyalty, was simply to increase immediate purchases with minimal marketing waste.  In theory, if targets responded and steadily purchased, no matter the purchase, more purchases should follow.  Supposedly then, over the long haul, the business accomplished its goal of capturing more share of wallet.

Around 2010, some leading edge marketers who realized the value of a real-time approach, began hitting that wall.  The foundation of the system they had spent 15 years building was the wrong foundation.  It was a platform built for segmentation, and it supported the wrong approach. They needed a “Real-time 1:1” platform, customer-centric prescriptions, and a more dynamic feedback loop.

Enter the Prescriptive AI Era

Good marketers have always been similar to psychologists in that they study consumer behavior. With today’s data and technology, it’s possible to take engagements one-step further – diagnosing, and treating those customers to alter their behavior methodically over time.  Stealing a page from the broadcast advertisers’ playbook – who use “subliminal seduction” – many marketers are marching toward implementing systems that use incremental and proactive drip therapy to persuade inner minds toward brand myopia.

The only piece missing from the puzzle is a real-time platform.  Traces of this began appearing in 2010, as big data systems, parallel computing, solid-state storage, and other technology advances drove computing costs radically down, and speeds up.

Today the pieces are in place, and more are climbing aboard, as real-time platforms have fully emerged and are cheaper and more reliable.  It’s now feasible to use customer-centric prescriptive tactics at scale and get huge lift over baseline approaches.  Models can predict behavior to an amazing degree of accuracy.  The artificial intelligence (AI) models both diagnose and – using Decision Management – proactively prescribe next-best-action engagement treatments.

Figure 3:

next-best-action

Everyone knows engagement professionals today have more channels.  They’re no longer constrained to broadcast media delivery systems (that lack dynamic feedback loops), and can now use digital response media and even physical surveillance.  And with this plethora of channels, they can administer and perfect personalized, contiguous, and hypersonic stimuli-response strategies.  Essentially, they can employ an always-on brain, powered by rich consumer data, advanced machine learning algorithms, and a 24 x 7 continuous learning loop.

What’s more, these machine learning technologies and embedded predictive algorithms can work in a very deliberate and intelligent way, dynamically creating conditional content and promotions, each time consumers reengage on a digital channel.  Incremental repeated responses (or lack thereof) allow these models to learn, tune themselves, and in essence direct and alter the future – programming individual behavior.  Customers are enticed to reveal ever-increasing amounts of personal information, in exchange for points or some privilege, trusting the exchange is amenable, and the information use one-dimensional.

All of this behavioral activity – social, purchase, demographic, and so forth – is recorded, with the aim of feeding it back into those same algorithms that iterate to find new patterns, refine predictions, and subsequently inform Decision Strategies that recommend the next series of treatments.  In some cases, these systems can even run autonomously, using advanced data science techniques such as genetic algorithms, game theory, and reinforcement learning.  System designers seed the rules of the game, configure the objective function and constraints, and then push “Go.”  The designers and their business counterparts peer in on occasion to monitor whether goals, such as higher loyalty and profit, are trending in the right direction.

Figure 4:

AI Learning Loop

Although this suggests overt manipulation, it’s not necessarily malevolent.  Provided customers have choice (and are well informed and discriminate), and businesses operate ethically (on a level playing field), the economic scales can still balance, and brands that provide products and experiences with the best value can still prevail, and consumers get a fair exchange of value.  You may have noticed, however, a few important “ifs” in this last statement.

Whether we like it or not, we now live in the Prescriptive Era, where the mission of brands is to get to know us, maybe even better than we actually know ourselves. That might sound crazy, but consider this statement from a recent article, “The Rise of the Weaponized AI Propaganda Machine” [i] where an analytics firm compiled data on Facebook likes and built millions of consumer behavior profiles, subsequently fed into an AI political campaigning machine:

“With 300 likes, Kosinski’s machine could predict a subject’s behavior better than their partner. With even more likes it could exceed what a person thinks they know about themselves.”

Whether you buy this or not, the fact remains that consumer profiles are becoming richer and consumer behavior predictions more accurate.  Data are exploding, as are the algorithms voraciously feeding on them.

Brands compiling this data and wielding their algorithms do it because they say they want to know us better.  Presumably, this enables them to continuously add value, deliver insights, help automate our lives, and make attractive recommendations.

Ostensibly then, for consumers, it comes down to a few simple questions:

  • How much is our data worth to us?
  • What’s the value of the insights that brands provide when they use our data?
  • Are we getting an equitable exchange?
  • Can we trust brands to honor their commitments regarding the use of our data?
  • Do we understand the fine print in those agreements?

Consider the mission statement for Datacoup, a data company based in New York, who have gone one step further and are trying to make a marketplace where consumer’s have a more direct exchange of value for their data:

“Our mission is to help people unlock the value of their personal data. Almost every link in the economic chain has their hand in our collective data pocket. Data brokers in the US alone account for a $15bn industry, yet they have zero relationship with the consumers whose data they harvest and sell. They offer no discernible benefit back to the producers of this great data asset – you.”[ii]

So are you getting value for the data you’re giving up?  Are the “Prescriptions” you get in return an equitable exchange?  Are you aware of what happens to your data after you release it?

A Day in the Life of Your Data

We all joke about the eye-glazing 56 page “Terms and Conditions” from Apple that we always accept and never read.  We want the free software, and don’t worry about the consequences. However, if you use that approach for everything you do online, that mindset is dangerous.

Consider this for a moment.  Most firms have language that allows them to send your data to affiliates, which is a fancy word for other companies. Once floating in the ecosystem, it’s grinded, distilled, and appended to other copies, until records of your preferences, habits, and behavioral are expressed in 5,000 or more different ways.  If it’s wrong, it doesn’t matter, because you don’t own it, don’t have access to it, and can’t change it.  In many ways, it’s another version of you, right or wrong.

Is Prescriptive AI Working?

So back to the question of whether it’s helping.   It’s fair to say there are cases where it adds value.  Here are some examples:

  • You decide you aren’t satisfied with your telecommunication services. You’ve made it obvious (with various signals) you’re considering other alternatives.  Your current provider prescribes an attractive bundle that satisfies your needs. You get a better bundle of services, and your provider retains you.  The bundle is custom tailored for you, using AI.
  • You have investments with a firm. You provide additional data on your financial goals, risk tolerance, and other investments, and they provide advice (prescriptions) on how to achieve your goals over time, within the parameters you set.  They provide various alternatives and education that prove useful to your financial planning.   Presumably, some of those alternatives include additional investments with them, and turn out to be good choices.
  • Your health plan suggests meaningful diet, exercise, and other tips that promote a healthy lifestyle. They are custom tailored to you, based on your family history, age, and other personal data you provide.   They reward you with lower premiums or credits.

These are just a few examples, and many more exist across industries such as travel and leisure, automotive, insurance, and retail.  And while good exchanges do exist, there are plenty of examples where the prescription doesn’t justify the information surrendered because the value exchange is unbalanced, or the prescriptions are ineffective.

Final Thoughts

In her book, “Weapons of Math Destruction[iii],” Cathy O’Neil writes:

“Many of these models, like some of the WMDs we’ve discussed, will arrive with the best intentions.  But they must also deliver transparency, disclosing the input data they’re using as well as the results of their targeting. And they must be open to audits. These are powerful engines, after all.  We must keep our eyes on them.”

She highlights important considerations we must heed.  I’m not convinced we’re spiraling toward a dystopian society regarding the use of prescriptive AI for customer engagement, but I do believe a balance is necessary between efficacy of these systems and fairness.  As responsible marketers, we should be mindful of the ramifications of the models we use for prescriptive purposes, and as consumers, it’s our job to demand transparency, choice, and a level playing field.

[i] Anderson And Horvath, https://scout.ai/story/the-rise-of-the-weaponized-ai-propaganda-machine, January 2017

[ii] Datacoup, https://datacoup.com/docs#faq, February 2017

[iii] Cathy O’Neil, 1st edition, Weapons of Math Destruction (New York: Crown), 2016.

Posted on

AI in CX: Real or Superficial Intelligence?

Artificial Intelligence

By all accounts, 2017 has ushered in the dawn of the newest Artificial Intelligence (AI) era. Most technology hype cycles follow typical paths, quickly shooting up, often followed predictably thereafter by a meteoric reentry to reality.  Typically, the entire flight takes place over a decade or so, as the fuel of inflated hype burns out, and the gravity of commercial application pulls down on its excitement to test its true value.

AI, however, seems different. It has appeared, drew much fanfare, and then disappeared several times already – more akin to a comet, flaring a tail of excitement with each new orbit.  As it reemerges, nearing the heat of expectation once again, it lights up with a spectacular plume, flung into space for another long dark hiatus.

AI history suggests five such orbits already – so is it destined for cold dark space soon?

Superficial AI

Regardless of the metaphor du jour, what we must inspect is the true value returned today, not the imagined expectations of tomorrow. The best test of commercial viability is not an intelligence test; it’s whether consumers are getting more value, and if the business offering the products & services are using AI technology as leverage, providing those things with higher margins.

For example, my mobile device is now my phone, my Garmin, my camera, my alarm clock, my digital assistant, my video recorder, my dictation device, my virtual reality device, and so forth.  20 years ago, it might have cost me $5,000 for these services.  Today, I get it all for $500 – $700.

We’re all under pressure to do more in the same amount of time.  To that end, these devices have become indispensable – they are essential to modern day survival – adapt to them, use them efficiently, or you’re passed by.

Therefore, by some measures and definitions, AI has delivered this time around.  Personally, I don’t care when a big company announces their sixth AI acquisition, or what their advertisements or creative animations say.  In my view, the proof is if customers are buying, are satisfied with those purchases, and are reporting their lives are easier, more productive, and more enjoyable.

Businesspeople must apply the same tests.  Can they deliver better customer experience with AI?  Are their product & services measurably smarter and more efficient?

If they aren’t passing those tests, then it’s just superficial AI.

Real AI Value in CX

AI – Automated Intelligence

As we all admire the latest bright tail of inflated expectation, let’s study what AI has really contributed to delivering better customer experience (CX) this time around.

For starters, look again at that magical device, the smartphone.  It streams location data, activity levels, browsing preferences, timing behavior, and the like.  Businesses consume this contextual data, and use decision hubs infused with AI algorithms that in less than a second calculate a next best action or insight.  That’s real!   Big banks, telecommunication / technology firms, and retailers are doing this today to improve acquisition, on-boarding, cross selling, and retention rates.

For consumers, the insights automatically delivered include recommended products, drive time estimates, calendar reminders, and service alarms. Alerts & notifications remind when bills are due, when fraud occurs, or when more exercise is required to meet goals.  Cars drive & park themselves, thermostats learn, and media services understand consumer preferences.   Customers can interact with machines by simply speaking to them.

For the marketers responsible for engagement strategies, AI now recognizes buying patterns, automatically performs A/B and multi-variate tests, which ranks the best content & promotions for the right individuals, and even suggests the best timing for those recommendations.  For salespeople, AI predicts the best contacts, opportunities, and accounts to spend energy on to maximize close rates.  For service workers, AI deflects simple service requests, and guides agents on complex service processes to improve time to resolution, ultimately improving customer satisfaction.

Simply put, there can be little argument that AI has delivered value during this orbit, much of it in the form of automation as opposed to higher-level intelligence.  Fewer marketers deliver more relevant and better-timed tactics.  AI assisted sales means higher quality pipeline with sharper close rates. Contact center managers relish shorter handle times and more efficient call resolution with less staff, and consumers enjoy shorter wait times and voice / bot-assisted service. For those using AI, NPS and customer satisfaction scores are on the rise.

All of these outcomes are commercially feasible.  Every business (not just the avant-garde) must rapidly incorporate these proven technological capabilities.  Hesitate, and the likely result will be eventual irrelevance.

What’s next – In my lifetime?

With all this said it’s back to our question.  Can AI keep delivering, or is it bound to let us down soon?

As humans, we love to dream.  That’s important.  In fact, regardless of how fast machines move forward, it’s still something that separates us from them.   We envision a fanciful future, and plot our course toward it.  Along the way, we stumble, get humbled, get up, and plot again.  This is our nature.  Each step along this evolutionary path, we create and refine machines that help us achieve our dreams.

Our vision seems unchanged.  We long to make life easier and more enjoyable for more of us.  To do this, we must continue to refine our existing tools, and invent new ones that assist us, and make up for our physical and human limitations.  No different from our first instruments, modern day smart tools take over tasks we were never very good at, or simply couldn’t do. They help feed us, optimize our resource consumption, and make our very survival possible.  We are already dependent on them, and there is no turning back.

This is also true for customer experience tools.  Our expectations are high and climbing.  We expect to interact with brands that listen, understand our preferences, react accordingly, and when something goes wrong, can turn on a dime and make things right instantly.

When I enter a website, I expect the search to be intelligent, the user experience to be delightful, and the checkout process to be flawless.  If I chose to do all this while mobile, I expect the same experience on my smartphone.  If I need help, my first reaction is, “why did things go wrong in the first place…how could this have been prevented,” and then I test if resolution comes fast with low effort – and does the business learn from the mishap.

This is the new normal.  Unfortunately, many brands today are not delivering on this type of customer experience.  The bar is high, but the elevation of game is not so much a demand from technology as from organizational re-tooling and reorganization to accommodate for technologies already commercially available.

Technological advancements will continue to accelerate.  Smarts will show up in more devices. We will demand our machines become more human, especially in delivering customer service and better experiences.  As humans, we love a personal touch, a social exchange, a sense of community and belonging.  So far, machines have not been able to deliver on any of these aspects.  That’s changing.

Presently, there is very interesting research going on to bring more human-like aspects to machine interactions. Google’s DeepMind research lab has made impressive gains in speech synthesis (text-to-speech) in a project known as “WaveNet” where robotic voices are becoming a lot less robotic.  Similar advances in Chabot research is leading to smarter bots able to remember details, learn right from wrong answers, and hold basic conversations.  You can try one of the better ones at http://www.mitsuku.com/

These developments are exciting.  The possibilities are enormous.  Yet until these become commercially viable and noticeably better with true customer engagements, you should train your eyes on what is real in AI today.  For now, focus your investments and efforts on delivering real CRM value from AI tech today in the form of things like simple service request deflections, intelligent routing to the right agent, relevant product recommendations & next best offers (based on individual behavior profiles), and guiding salespeople with next best activities.

Meanwhile, keep close tabs on these other AI CX innovations as they progress, take some calculated risks on a few promising areas, and prepare for the next revolution of AI.  The AI comet will be back shortly.

Posted on

Rise of Machine Marketers – Transforming CX

Machine marketers are smarter marketers, always using machines for advantage.  But this isn’t new.

Direct marketing was born out of the ability to exploit addressable media as the way to garner feedback on whether their enticements were working.  Catalogs and snail mail with reply forms, evolved to email, telemarketing, and other mechanisms – smarter marketers understood guesswork would never win over using data, technology, and the scientific method.

Machine Marketers

Database Marketing

In the 90’s, the ability to more massively codify and share customer data, and use it to steer marketing campaigns drove a revolution. It sparked a major shift of media spending away from general advertising using TV & Radio, to addressable programs.  Database Marketers, the offspring of Catalog Marketers and ancestors of Machine Marketers, scraped for individualized customer information to power personalized treatments – where direct response open rates, response rates, and conversion rates kept score.

They loved data because when they used it to drive targeting in their programs, the patient responded. Realizing their treatments were working, they wanted more data, wanted it fast, and wanted it in pure forms. Native sources worked well, but they sought alternate supplies in the forms of public, compiled, and modeled data – anything to test for a slight edge.

A new market formed with a vast array of players, arising to meet the growing demand for customer data.

The 2nd Coming of Big Data

Then, a number of things happened. Even more individualized data poured onto the market.  Consumers shopped and bought online.  Consumers went mobile.  Consumer devices of all kinds started streaming behavior data.  Consumers readily traded personal information for points and promises.

Hardware continued to plummet in price and better software meant cleaner and more accessible data.  Data compilers flourished, with data as their raw material, and database & data science technology their assembly line, and the internet their logistics network.

Database marketers had struck oil again, but this time it was BIG – and IoT data was the source of their new bubbling crude.  Data refineries appeared everywhere.

Internal IT had competition – their 90’s data warehouses rendered obsolete by a Big Data revival.  Open source databases like Hadoop, were faster and ran on commodity hardware. SaaS providers offered a variety of big data subscription services, and agencies used bigger and faster hosted databases.

There was but one small problem. Insights weren’t leaping out of these primordial big data reserves.

Data Science and Data Mining Come of Age

Meanwhile, mad “Data Scientist” marketers continued to manipulate and tune their statistical models to improve lift. Early on, they realized that algorithms devised hundreds of years prior could now be fine-tuned and fully unleashed to predict which customers were more likely to respond and buy their products.

Less sampling with faster machines and more data meant better results.  Suddenly, more people became interested in what they were doing.  People were peering over their shoulders. The press told stories of firms predicting a pregnancy before grandparents even knew.  Adding fuel, the biggest brands on the planet (Google, Facebook, Amazon, et al) got into the game, doing big reveals, seemingly weekly, on the methods to their data science madness.

It was time to give this a makeover, market it, and commercialize it.  “Geez,” said the creative marketer. “We can do that!”

AI and Machine Learning – The re-launch

Our story takes us to circa 2012.  The time was right.  Cars were beginning to drive themselves; IBM’s Watson had won Jeopardy; Google was predicting our search terms and winning at the game Go.  Our iPhone was conversing with us, and Amazon & Netflix were courting us with recommended products to buy and movies to watch.

Honestly, no new science unexpectedly sprang forth, but as happens old science around for decades (decision trees, neural nets, Bayesian learning), became an overnight – well let’s call it an over 5 year – sensation.

What happened was how technology revolutions occur.   Attention begot investment, huge investment bought more innovation, and marketable innovations caught more attention – and the virtuous loop was in motion – adequately fed by a rich venture capital environment.

Marketers assembled the pieces into cost effective working solutions. They collected and compiled consumer data sources, cleansed and filtered them, fed them into pattern recognition and self-learning systems, detected opportunities and alerted touchpoint systems, automated waved campaign schedules, and connected their outputs to fulfillment systems.  They did all this via an interconnected stack of private and public clouds, transferring data and insights in seconds.

Michelangelo meets Newton – When Content met Context

CX AI

By 2016, another phenomenon unfolded.  Creative & scientific minds more closely collaborated.

Deep Learning, the science of neural networks, commoditized language and image processing, changing how we interfaced and worked with machines. Clunky interface paradigms gave way to elegant ones that were responsive and rewarding.  Design thinkers (those artsy fartsy types) were no longer an afterthought. Au contraire, they were now a strategic advantage.  Consumers dictated the definition of great customer experience: Relevance, value, simplicity, and visual beauty.

Machine Marketers, ever the opportunistic breed, seized the moment, further refining their targeting and personalizing creative treatments across available channels. Machines further assisted their agency suppliers, assisting them in turning out better, faster, and cheaper creative.  Technology further assisted marketers, auto generating optimal SEO terms, email subject lines, and even catchy tweets.  Machines advised on the optimal time to execute campaigns.  Next best recommendation rankings used statistical probability to find relevant products & services for more refined targets.

Beautiful creative no longer took months to produce.  In many cases, consumers produced content for brands – and the content bottlenecks holding back visual personalization broke lose.

Science and technology glued yet another critical piece into place.  Touchpoint systems where customers interacted could now understand natural language, and instantly fed back contextual data (location, last behavior, weather conditions, intent, mood, and so forth) straight through to systems primed with algorithms that learned in real-time, recalculating next best actions in a conversational mode.

“Computer, find me the closest coffee shop.”…”Ok, I found one 2.5 miles away, do you want directions?”

“Computer, I need a highly rated case for my X phone for under $25.”…”Ok, I found four with 5 star reviews that fit your X phone for under $25, do you want to hear about them?”

Fronted by Natural Language Processing (NLP), personalization engines married conditional & appealing content with contextual recommendations – spawning audio & visual personal assistants.  The result: off the charts lift and conversions.

These were contextual, conversational, and relevant interactions.  This was transformational.

artificial intelligence evolution

Machine Marketers Rise Up

In the end, let’s face it.  Marketers want to do one thing more than anything – sell more stuff.  Yet the smart ones know that the best means to that end is relentless focus on the customer.  Ensure each is a happy camper via an individualized relationship, and satisfaction and profits increase.

Thus, today more than ever, ALL marketers had better face one important fact.  They can’t achieve customer centricity at scale using the tools, data, or organization of yesterday.

Like any profession, winners constantly seek a new competitive edge using the latest technological advances in equipment, repeatedly testing innovations, measuring for improvement, and fine-tuning.

Artificial intelligent interfaces are changing the ways consumers interact with their devices, provide data, and interact with brands.  Data is flowing freely, and although privacy laws seem to ebb and flow, the trend has been toward more data sharing and the ability for the crafty to gain a deeper understanding of consumer behavior.

Technology – cheaper, smarter, more portable, and easier to use, continues to translate into the potential to deliver more relevant and convenient customer experience.  Those that get this, and execute on it, will win.

Machine marketers are those who master using the latest data & technology to their advantage – rising to that challenge, they rise to the top of their craft.

Note:  These views are my own, and not that of my employer

Posted on

Adtech Martech Convergence – Episode #7

In this 7th short video in my Machine Marketing Series, I give my views on the “Adtech Martech Convergence”  specifically as it relates to using machine learning.

 

I cover four main layers of technology to consider as this conversion takes place:

  • Customer Behavior Data – Why the Adtech Martech convergence may force a better coordination of this data as its compiled  along the customer decision life cycle.
  • Basic Analytics & Insights – I give some examples and why this area isn’t a huge concern or risk area.
  • Advanced Analytics (Machine Learning) – I explain why integration here is key, and give some marketing use case examples.
  • Programmatic Real-Time Automation – I outline key aspects of automation & workflow, and why these areas are essential to combine for a coherent Adtech Martech solution stack.
Posted on

Machine Learning Ecosystem for Marketing – Episode #6

In this 6th short video in my Machine Marketing Series, I explain the Machine Learning Ecosystem for Marketing and review about 50 major players.

I cover the 6 categories of Machine Learning vendors in the landscape I created with Tier 1 & Tier 2 players:

  • Category 1 – Paid Media / Digital Recommendation Vendors
  • Category 2 – Big Data Cloud Platforms with Machine Learning Services
  • Category 3 – Open Source Machine Learning Tools
  • Category 4 – Business Intelligence Vendors with Predictive Analytics
  • Category 5 – B2B Marketing / Account-Based-Marketing (ABM) / Lead Management
  • Category 6 – Real-Time (Multi-Channel) Interaction Management (RTIM)

In each category I discuss the Tier 1 (largest players) and the Tier 2 (contenders / firms to watch).

Posted on

Artificial Intelligence 2017 -5 things NOT to underestimate

Here are 5 things you will undoubtedly underestimate about Artificial Intelligence (AI) in 2017

artificial intelligence

“We are Now Controlling the Transmission”

If you aren’t familiar with the 60’s TV series “The Outer Limits” you need to watch this intro (its 58 seconds long).

Artificial Intelligence is controlling more than you realize, and in 2017, it’s going to accelerate.   AI algorithms are already affecting which products & services you see when you perform a search – regardless of the device or interface.  AI is pushing messages, offers, & advice to you that you may never have asked for. Artificial Intelligence is deciding what shows up in your news feeds, on the sites you frequent, and in the apps you use.

Sometimes those results will be relevant and useful.  Yet often, they won’t be that relevant and will have bias.   So beware, be careful, and try to understand better what is showing up and why.  Be an activist.   Consumerism is empowering.  Use that right to force companies into making AI better, in essence taking back some control.

“We know who you are…and where you are…Well kind of”

A little over 2 years ago, I wrote a blog post entitled “Customer Data & Decisions – Reflections of Me.”

The forces I cover in this piece related to the collection of consumer data will continue to accelerate in 2017.

My advice is work to get your own data house in order next year.  Go beyond the obvious, like pulling your credit report, to being prudent about how you safeguard and when you share any of your data.  Its valuable, but its also vulnerable.

When you install apps, be thoughtful about which ones you grant permission to track your location.  Did you know that Uber now tracks your location from the time you open the app to 5 minutes after you arrive at your destination?  Uber values that data, and can leverage it in a variety of ways – but do you want them doing that?  What have you received in return?  How will they share that data?

Think about these factors when you decide how much data to share, and spend time to understand how firms are using and compiling YOUR data.   In some cases, you may not have direct control, but nonetheless it’s important to understand what’s happening since after all, it is your data.

I’m not opposed to companies collecting relevant data to feed into Artificial Intelligence systems to use in a responsible way to add value to my consumer experience.  Yet I expect value in return, and I expect a firm to respect my privacy.

You have a vote in Artificial Intelligence evolution.  Don’t let it run amuck

artificial intelligence evolution

You vote with your wallet & purse everyday.  Intelligent devices are popping up everywhere and dropping in price, and firms are dropping Artificial Intelligence into almost everything we buy.  From voice activated appliances, to connected cars, we now live in a world where having a conversation with a computer is commonplace.

Again, what you buy helps decide which AI infused products thrive and which ones die.  Buy wisely.   Perform some reviews.  Render your opinion.   Your opinion on a frying pan may mean a little less to society then your review of how an interaction went with a Chat Bot.

More money than you can imagine will be spent on it

How much will be spent on AI?  No one really knows.   This post has some interesting estimates.  Let’s just say that all of us will underestimate the amount of energy that will be put into AI related research & technologies in 2017.   It’s going to be HUGE – and it’s not just in 2017. This is a revolution happening.

AI will have bias because of its creators & the data it feeds on

Artificial Intelligence actions & outcomes must be monitored & governed by its carbon-based creators.   Yet ironically, its creators are the ones that introduce bias into the AI brain.

From biased data (think about how a hiring bot might make decisions about which candidates are most likely to succeed in a job….it will use past employee performance data…which is biased by the hiring practices of the past), to the bias rules & code from the inventors.

Its incumbent on us all to be mindful of these tendencies, and advocate aggressively against bias in the machine.

Note:  These views are my own, and not that of my employer

Posted on

Machine Learning Measurement: Episode #5

In this 5th short video in my Machine Marketing Series, I explain Machine Learning Measurement and monitoring techniques.

I cover the concept of a Model Factory and Dashboard where you can:

  • Measure Machine Learning Lift employed in Marketing and CRM programs
  • Monitor Model Features and Model Data use
  • Pit models built for Customer Experience (CX) against each other (Champion / Challenger)
  • View KPIs (Key Performance Indicators) on a business Dashboard

Use these tips & best practices to benchmark against your efforts, and compare how your platform stacks up in using machine learning in your marketing and Customer Experience programs.

 

Posted on

Marketing Results: Machine Learning Episode #4

Getting positive marketing results for your customer engagement efforts is what it’s all about.

In this episode, I explore what to expect in terms of outcomes when you effectively employ machine learning (ML) and artificial intelligence (AI) in your marketing efforts.

I cover performance ranges I’ve seen actual companies achieve including:

  • Churn & Attrition reduction
  • Response Rates for your marketing execution & tactics
  • Sales Lift increases which means revenue to the bottom line
  • ROMI – Return on Marketing Investment
  • NPS – Net Promoter Score improvements
  • AHT – Average Handle Time reduction

You can use these figures to benchmark against your efforts, or even help you build a business case for embarking on using machine learning in your marketing and Customer Experience programs.

Posted on

Marketing Timing & Content: Machine Learning – Episode #3

Great timing & content lead to great marketing tactics and performance.

In this episode, I explore how you can use machine learning (ML) and artificial intelligence (AI)  to improve your message timing and the content you employ – further improving experiences for your customers.

I explore “2 Cool Areas” of Machine Learning & Artificial Intelligence applications – to make your marketing smarter:

  • Timing Optimizing for your Marketing Execution & Tactics
  • Automated Content Generation and Predictive Content Recommendations

My tips are aimed at improving your marketing efficiency & effectiveness.

Posted on

Machine Learning & AI Use Cases for CRM

Updated: May 6, 2020

Below you’ll find a good organization of classes of AI use cases for CRM (Customer Relationship Management) & CX (Customer Experience).  Presently, there are countless examples of applying Artificial Intelligence (AI) and Machine Learning (ML) to solve a broad class of problems beyond CX, from cancer diagnosis, robots in manufacturing operations, streamlining product development, detecting anti-money laundering, spam filtering, improving cybersecurity, to self-driving cars.

machine learning

Great marketers and CX experts have always been change agents.   As new-age change agents, they focus on AI use cases that enhance CRM, and have either already been proven to lead to successful outcomes, or that show significant commercial promise.

But what is AI and how can it be used to take customer engagement to new levels?  To make it simple, think of AI as the application of software technologies using standard hardware for:

CRM AI purposes –

  • Task automation:
    • Doing something in less time with fewer steps
    • Accomplishing something with less total effort
    • Doing something with less or even zero human intervention
  • Detecting, classification, and alerting:
    • Sensing and understanding a current problem (customer is dissatisfied or struggling)
    • Sensing current customer intent (wants help; wants more information; trying to buy)
    • Placing similar things together (customers, products, content)
    • Informing someone or some other system once a certain confidence threshold is reached
  • Predicting something is likely to happen:
    • Customer likely to buy more (or less; or never buy again)
    • Customer is likely to call for service
    • Contract is likely to close
  • Suggesting (in reaction or pro-action) a course of action leading to more optimal outcomes:
    • Doing X will resolve the problem
    • Recommending Z will satisfy the requirement
    • Offering Y will increase customer lifetime value and/or maximize profit

Machines are capable of doing these in a faster, more accurate, and more cost-effective way versus humans doing them manually.  When that happens, the result is a superior customer experience and better business results.

CRM is a technology that improves customer experience (either unassisted or assisted by an employee) as it relates to customer’s interactions with a brand’s marketing, sales, and service, and the fulfillment of commitments by those three functions.

Artificial Intelligence (AI) technologies –

Regarding different sub-areas of Artificial Intelligence and its class of technologies, consider this organization:

(Note: these categories are not mutually exclusive in that a given application may benefit from one or more of these technologies)

  • Machine learning – Algorithms using statistical methods designed to predict or forecast something. The learning is either supervised (i.e., assisted) because we either know the outcome trying to be achieved (or we have a teacher to inform us) and can determine if the model is right or wrong – or unsupervised (i.e., unassisted) because we have no idea of the patterns or outcomes that may be best).  The system “learns” if over time its predictions and performance improve.
  • Text, Video, Audio, and Image Analytics – this code scans unstructured data, and finds entities/objects, and classifies them (and/or extracts them). Deep learning is an advanced form of image analytics – using a technique with layers of brain-like neurons (Recurrent Neural Networks (RNN) or Convolutional Neural Networks (CNN)).
  • Complex Event (Monitoring) Processing (CEP) – an engine fed streaming data from one or more source, which detects certain patterns and then initiates follow on actions or processes.
  • Deep Q&A – using an AI system that has access to previously compiled knowledge sources; it scans, filters, and presents the best likely answers to questions.
  • Natural Language Processing – this code translates spoken voice to text, or to some other form useful as input to another system. It can also do the reverse, translating some other system’s output to spoken voice.  It also can translate from one language into another, or just detect the language.
  • Natural Language Generation – this code takes input (images (still or moving), text, etc..), and generates text descriptions of what its seen or found.
  • Numeric Analytics – this code uses commonplace mathematics (simple formulas) to surface insights, focusing the receipting’s attention to aid their future decisions.
  • Robotic Process Automation – this code repeats tasks that it’s instructed to repeat.
  • Deterministic rules – if/then/else statements that don’t change unless a programmer changes the rule.

Click here for an AI diagram of how I’ve organized these into building blocks so you better understand where each belongs in terms of its contribution to automation or true machine intelligence

AI Automated Intelligence

CRM AI Vehicles –

Regarding implementing AI use cases in CRM for improving customer engagement (e.g., using any code and a combination of the above algorithms), consider these main vehicles:

  • Virtual Assistants – a module designed to aid and assist a human with decision making, problem resolution or task completion including:
    • Chatbots (such as Facebook Messenger)
    • Conversational interfaces (such as Alexa, Google Home)
    • Office assistants built into desktop software or mobile apps (e.g., automatically schedule meetings)
  • Digital Recommendation Engines – ranks and bundles content recommendations, product recommendations, offer recommendations, and service recommendations and uses containers/spots on web, mobile, agent desktops, email, kiosk, and other digital devices to serve these to:
    • An employee or agent to sell a product or solve a service case
    • Engage directly with a consumer through any digital device

CRM AI Use Case Categories –

bigdata_value2

Given these areas, there are hundreds of potential valuable AI use cases for CRM.  Here are some examples:

Marketing

Profiling and Tactic Execution:

  • Predicting missing or outdated customer data values

~Forecast data value using time series, or simple inflation adjustments
~Predict data value from an image

  • Segmentation, clustering, targeting
  • Next best product / service recommendation
  • Next best content recommendation
  • Next best promotional recommendation
  • Next best channel to engage on
  • Next best time to send
  • Next likely transaction (and timing of it)
  • Next best search keyword

Predicting Customer Behavior and Value:

  • Models to predict customer churn
  • Models to predict customer preferences (including a preference for a particular product/service)
  • Models to predict customer lifetime value
  • Models to predict current customer value
  • Models to predict customer wallet share
  • Models to predict customer re-activation

Dynamic Price Optimization

  • Optimize retail price of products/services
  • Optimize incentive levels (discounts) of products/services
  • Optimize price paid for ad buys

Marketing Operations / MRM / Marketing Planning:

  • Tagging/filtering content – Recognize an object in image or video
  • Tagging/filtering content – Match an audio track to known copyright material
  • Tagging/filtering content – Classify image content type
  • Content generation
  • View trending topics
  • Forecast market size, marketing revenue/costs
  • Budget allocation planning/media mix optimization

Influencer Marketing:

  • Content identification

Sales Automation

  • Next best lead
  • Next best sales activity
  • Next best sale rep
  • Sales contracts
  • Sales and demand forecasting
  • Competitive intelligence

Customer Service

Customer Service / Support

  • Next best service, support, training action to resolve a complaint/problem
  • Intelligent routing: Find the next best expert/agent to resolve a complaint/problem
  • Text and images uploaded to support resolving cases
  • Tag and classify support tickets
  • Support process automation
  • Estimating service volumes, wait times
  • Optimize scheduling and support resource utilization
  • Contact center volume forecasting
  • Estimate wait time
  • Optimize scheduling and support utilization optimization

Predicting Customer Service Behavior:

  • Models to predict levels of customer frustration or satisfaction
  • Sentiment analysis with all forms of data.
  • Reasons behind high churn likelihood
  • Models to predict intent (likelihood of a customer calling and why)
  • View trending support topics
Posted on

Smarter CRM: Machine Learning Episode #2

You want Smarter CRM, don’t you?

In this “Mini-Cast” on machine learning (ML) and artificial intelligence (AI) for marketing & CRM, I outline the numerous applications you can use to improve customer experience.

AI for Smarter CRM
AI for Smarter CRM

In this second episode, I delve into the various areas of AI & ML applied to Marketing, Sales Automation, and Customer Service where ML & AI play huge roles in taking those functions to new levels of insight and intelligence, enhancing productivity, effectiveness, and delivering better customer engagement.

Episode #2 on YouTube:

Posted on

Machine Learning Marketing Series – Episode #1

Introduction & History of Machine Learning:

In this inaugural episode, I explore some useful history on the various areas and aspects of machine learning and artificial intelligence, and its evolution leading up to today.

Specifically, I delve into unsupervised machine learning, deep learning, and software robotics as applied to customer relationship management.

Watch my  entire “Mini-Cast” series of short 5 to 10 minute videos on machine learning (ML) and artificial intelligence (AI) used in marketing & CRM to improve customer experience (CX).  The entire series is available from my website linked to my YouTube Channel:

Episode #1: Machine Learning Introduction & History (this one above)

Episode #2: Smarter CRM – Machine Learning Applications for Customer Experience

Episode #3: Marketing Timing & Content – 2 Cool Use Cases for Machine Learning

Episode #4: Marketing Results – Amazing Outcomes using Machine Learning

Episode #5: Marketing Learning Measurement & Monitoring

Episode #6: Major Players in the Machine Marketing Ecosystem

Episode #7: Adtech & Martech Machine Learning Convergence 

Episode #8: Machine Marketing – Hot Trends (coming soon)