machine learning makes you more human - a marketing story
TRANSCRIPT
TODAY’S ROADMAP
1st half:● Quick review - What is machine learning?● Putting ML into marketing terms● Comparing ML marketing against other options● Current & near-future examples
2nd half:● Build a personality analysis service
WHY IT’S IMPORTANT
● “Smart” is coming, ready or not● Machine learning is here and its impact on our lives is
only going to grow● It’s a paradigm shift — Seems strange that we would
now have access to these “out-of-reach” tools● It’s much easier than you think ● And it will quickly become the norm● Now is the time to jump in, ahead of the curve
FOR EXAMPLE...
Identify the pattern in buying habits of women
who just became pregnant.
Locate that pattern in others and target them with special offers for pregnancy products.
Will the current app user buy “maternity clothes” now?
BREAKING IT DOWN
is female?
is age> 20?
is Y app installed?
is X app installed?
end
has used < 30 days?
was X function
used?
was Y function
used?
no
yes
no
yes
no
yes
no
yes
end no
77.1%
yes
end
probably
52.3%
93.6%
2.3%15.8% 23.4%
WHY NOW?
The expensive and complicated machine learning systems Target had back in 2012 are cheap and
simple today.
And you have easy access to them.
businessinsider.com/the-incredible-story-of-how-target-exposed-a-teen-girls-pregnancy-2012-2
THE PROBLEM WE’LL TACKLE
One of the biggest challenges we face as marketers is how to personalize messaging to individual prospects and customers so that it most strongly resonates with
each unique recipient.
boagworld.com/usability/adapting-empathy-maps-for-ux-design
“The goal of a company is to have customer service that is not just the best but legendary.”
~Sam WaltonFounder, Walmart
RELATIONSHIP IS END-TO-END
“You can always find out the true nature of a business by simply asking for a refund.”
RELATIONSHIPS 101
Listen to what they say,Become an expert in who they are, and
Find ways to make their life better.
IT’S A QUESTION OF SCALE
Building deep meaningful relationshipstakes a lot of work.
As human beings, we can only build great relationships with a handful of people.
But machine learning can amplify that effort so you can build great relationships with millions.
FOR EXAMPLE...
Pick one of your customers.
Read every piece of content they’ve ever put out — every blog post, tweet, facebook update, Instagram image, etc.
How well would you know them? Pretty damn well — you understand them and can share their feelings.
That’s empathy.
And how well do you think you could market to them now?
Machine learning is an amplifier.
It allows you build empathy at a scale and depth that is simply beyond human capabilities.
● Model the habits, likes, dislikes and values of your customers, then
● Predict the behavior of those customers and personalize their content accordingly
NOW SCALE THAT UP
IN EFFECT...
Computers can dig deeper and personalize at a larger scale than humans are capable of.
This allows machine learning-powered brands to develop deeper relationships — with more customers
— than previously possible.
CRAFTING MESSAGES
Today, we’ll talk about 3 different options…
● Old school — very limited● A better way — somewhat limited● The machine learning way — sky’s the limit
Plus, some examples from the near future.
HOPE & A PRAYER
Deliver one version of your message and send it to every recipient at
the same time.
Hope they like it.
VERY LIMITED
● No personalization, eg. “Hi, you”● Written for everyone, connects with no one● Single delivery time, opens will be stroke of luck
MULTIPLE FLAVORS
Deliver multiple variations of your message and
randomly distribute them across every recipient at
the same time.
Determine which variation performs the best — use it
next time.
GETTING WARMER
● Basic personalization, eg. “Hi, Sam”● Everything is reactive — after the action has been
taken (and after you spent the $)● Minimal access to underlying motivation factors● Have to create multiple versions of same asset
INDIVIDUAL FLAVORS
Deliver personalized variations of your
message for each and every recipient. Sent at
ideal time for each recipient.
A PARADIGM SHIFT
PredictOptimize vs.
React to signalsStart with guesses
Shape behaviorsStart with models
Up to now, marketers have focus on “optimizing” their campaigns. With machine learning, we can now shift to “predicting.”
Behavior Prediction
Interest Tracking
PREDICTIVE PERSONALIZATION
Pages & content they’ve visited
Emails they’ve opened/clicked
Resources they’ve used/downloaded
Products they’ve viewed/wishlisted/bought
Searches they’ve made
Blog
Store
Find patterns Determine what they want to see/do/buy next (and when)
Days/time they’re active App
Search
Devices they’ve used (& geo location)Email
Social
• Recommended posts• Recommended products• Delivery day/time
• Dynamic content• Related posts• Sales offers
• Related products• Cross/up sell• Dynamic pricing
• Dynamic content• Sales offers• Functionality
• Query suggestions• Results ranking• Sales offers
• Content curation• Delivery day/time• Retweet/reshare
Tribe• Recommended topics• Topic curation• Member introductions
Yourcustomer
PEOPLE LIKE PERSONALIZATION
● 69% of consumers believe a brand's consistency across channels affects their loyalty
● 59% of consumers who have experienced personalization believe it has a noticeable influence on purchasing
● 67% of consumers who have experienced personalization are highly in favor of personalized coupons
All of these are strong points for machine learning-based marketing.
infosys.com/newsroom/press-releases/Documents/genome-research-report.pdf
WHY IT ROCKS
● Unique personalization, eg. “Hi, Sam — we noticed that you were searching for...”
● Can fine-tune content per individual person● Demonstrates genuine empathy for customers● Driven by computers (faster, cheaper)
PERFECT MATCH
A quick note…
Computers are very good at doing the boring, monotonous tasks people don’t like to do.
Tasks that can build billion dollar companies.
PRODUCT RECOMMENDATIONS
techblog.netflix.com/2012/04/netflix-recommendations-beyond-5-stars.html templates.prediction.io/PredictionIO/template-scala-parallel-universal-recommendation
SENTIMENT ANALYSIS
keyhole.co
ibm.com/smarterplanet/us/en/ibmwatson/developercloud/tone-analyzer.html templates.prediction.io/pawel-n/template-scala-cml-sentiment
AUTOMATED CAPTIONS
“A group of young people playing a game of frisbee.”
Automatically create contextual
descriptions of your images for
accessibility.
io9.gizmodo.com/computers-wrote-the-caption-for-this-photograph-and-ch-1660450610
TRANSLATION
Write your message once, have it be
delivered in multiple languages through
automated translation.
PERSONALIZED CONTENT
Deliver blog posts and other content,
uniquely personalized for each customer.
techcrunch.com/2016/04/21/facebook-news-is-new
PREDICTING CHURN
Monitor for signs that a customer may bail.
Then deliver targeted messages and
promotions to retain them.
LIFETIME VALUE FORECASTING
LTV - arguably, the single most important metric
for marketing.
Forecasting the profit — or value — your business will gain from its entire
relationship with a particular customer.
10xnation.com/customer-lifetime-value
Customer
Group
Retention
time
Transaction
Value
Transaction
FrequencyCLV
Young
Mothers60 months $10 1 $600
Teenagers 24 months $10 1 $240
Single
Parents48 months $12 1 $576
NET IT OUT
The better you know your customers...The better your marketing will be…
The better your customer relationships will be...The more customers you'll have...
The more successful your business will be.
AND...
Machine learning amplifies your relationship-building skills to levels beyond human capabilities.
UNLEASH YOUR BUSINESSEMBRACE EXPONENTIAL
10xnation.com