The Perils of Algorithm-Based Marketing

MARKETING
The Perils of Algorithm-Based

Marketing
by Utpal M. Dholakia
JUNE 17, 2015
If I told you this article was written by an algorithm (and you believed me), chances are youd be
creeped-out, suspicious of the content, and unable to muster much if any emotional response to
it.
Thats a natural response. Yet companies dont seem to be able to see algorithms from the
consumers point of view. They think nothing of deploying algorithms as marketing tools.
Instead, they should be looking for ways to inject humanity and, yes, actual humans into
their eorts to reach out to customers.
Ill explain why.
Originating in computer science, algorithms are simply sets of ifthen rules. But a number of
factors easy-to-use predictive analytics and data-visualization tools, the proliferation of mobile
devices, and companies ability to track and measure customer behaviors have helped
companies nd an astonishing variety of ways to use algorithms, particularly in marketing.
Algorithms help marketers utilize customer-specic knowledge demographics, previous

behavior, fellow customers choices to craft customized oers and deliver them, often in real
time. They help companies track customers, cross-sell to them, and promote products. Banks use
algorithms to suggest new products to customers, online retailers deploy them to set and change
prices, and media companies rely on them to recommend and deliver streaming content and ads.
But despite the broad adoption and growth of algorithm marketing, companies should be
cautious about it, for four reasons.
Algorithms arent sensitive enough to context. Eective marketing relies on messages that are
attuned to the customer. Customer response, even for the most mundane of products, is
sensitive to a host of ever-changing factors. On any given occasion, everything from personal
factors such as how well a person has slept the night before, current mood, hunger, and
previous choices, to environmental variables such as the weather, the presence of other
people, background music, and even ceiling height can inuence how a customer responds.
Algorithms can use only a handful of variables, which means a lot of weight is inevitably placed
on those variables, and often the contextual information that really matters, such as the
persons current physical and emotional condition or the physical environment in which the
individual is tweeting, Facebooking, or buying online, isnt considered.
They arouse suspicion and can easily backre. In a climate where privacy concerns are
perennially at the forefront of customers minds and trust is at a premium, customized
marketing of any sort is risky. If customers feel the marketer knows too much about them,
algorithm-based personalization can seem creepy or backre badly. Something as innocuous as
a customized Facebook feed highlighting the past year can generate tremendous grief for users
in certain circumstances.
They encourage complacency. Having tools that capture exhaustive data about customers,
quantify minute aspects of their behavior, and measure their responses can create a false sense
that one knows customers really well and understands their motivations and triggers. At best,
such complacency may limit marketing activities to those that can be measured easily and have
worked in the past. At worst, it can lead to a completely inaccurate customer portrayal, ill-
conceived marketing overtures, uninterested customers, and wasted resources.
They stie customers emotional responses to marketing oers. By their nature, ifthen
rules imply a decision calculus (If you are 3545 and just paid 20% down with a mortgage,
then you should purchase a home equity line of credit). Marketing algorithms encourage
analytical consumer decision making. Yet many customers make choices impulsively, are
motivated by fun or a need for variety, and respond with spontaneous, emotional reactions. It
is in these unscripted and impromptu interactions that the customer forges the strongest and
most meaningful connections with brands. By shifting the customer into a more calculated and
methodical mind-set, algorithm-based marketing minimizes opportunities for forming
emotional bonds and limits the range of customer actions to the marketers detriment.
How should marketers solve these problems? By injecting a strong dose of human into their
customer interactions.
Marketers can inject the human into their oers and activities in many ways. Some approaches
involve bringing in an actual human to interact with customers; others require making the
algorithms work in more human-like ways.
INSIGHT CENTER First, understand the context and make it part of

the tailored oer. Eective marketing is a never-
Growing Digital Business
SPONSORED BY ACCENTURE
ending quest to understand the contextual
New tools and strategies. drivers of customer response. Constantly testing

the impact of dierent contextual factors
through eld experimentation and
observational research is part of the solution.
Translating this understanding to marketing programs may mean creating more-sophisticated
algorithms having more rules and greater exibility to apply those rules; in another important
sense, however, it involves the creation of processes and activities where the marketer can react
to a customer without the use of algorithms. Waitsta empowered to oer a complimentary
cocktail to a distraught customer, or trainers who can take the time and trouble to educate a new
software buyer after understanding exactly how she will use it, produce value for customers that
algorithms simply cannot. Fondue restaurant chain The Melting Pot takes just such an approach,
empowering servers and managers to oer complimentary food and cocktails and do other things
to deliver a superior customer experience.
Second, introduce unpredictability into marketing. Algorithms favor patterned marketing
activities, such as coupons or catalogs that are sent every quarter, emails sent each week, or
twice-a-day tweets. Such marketing loses its impact partly because it is so rmly structured that
customers become habituated to these overtures and tune them out. Introducing randomness
and surprise into marketing activity breaks the shackles of algorithms and goes a long way toward
counteracting customer ennui. An unexpected call or a surprise new product announcement, or
even a sale that fails to materialize when it should have, make the brand more authentic, human,
and exciting. Unpredictability allows customers to seek variety and respond emotionally to the
brand. TD Bank, a Canadian bank, turned its ATMs into Automated Thinking Machines last
year, surprising thousands of customers with $20 bills, owers, and merchandise, earning
accolades and garnering signicant attention on social media.
Third, encourage interaction with humans at key customer decision and experience points. To
date, adequate technological substitutes for empathetic, competent, and empowered humans
have not been discovered, which is why luxury brands, whether they are fashion labels, hotels, or
automobiles, overwhelmingly favor human sales and service sta over technological methods.
Within constraints of aordability, every marketer should thoughtfully consider where and how
to incorporate human interactions into the customer experience. Hybrid approaches utilizing a
combination of humans supported by expert algorithms for services such as nancial planning,
product recommendations, and building security are an eective way to inject the human into
algorithm-based marketing activities.
Utpal M. Dholakia is the George R. Brown Professor of Marketing at Rice Universitys

Jesse H. Jones Graduate School of Business.
This article is about MARKETING

FOLLOW THIS TOPIC
Related Topics: GROWTH STRATEGY | INTERNET

Comments
Leave a Comment
POST
5 COMMENTS
We Innovation a year ago

Great analysis and comments. Analytics and algorithms tend to have us believe we can measure the
unmeasurable, and marketers need to be aware of this tendency. Contextual marketing and psychological
approach to customer behaviors need to become key development sectors, putting emphasis on understanding
an ecosystem conditioning individuals and inuencing their choices. Even with that, analysts argue there will still
remain uncertainties and discrepancies, hence the need to bring more conversational marketing and
communications with customers as a way to share knowledge and value, and therefore bring a joint meaning to
data sets and algorithms. http://weareinnovation.org/2015/07/06/market-roadmap-new-strategy-new-
marketing/
REPLY 00
JOIN THE CONVERSATION
POSTING GUIDELINES
We hope the conversations that take place on HBR.org will be energetic, constructive, and thought-provoking. To comment, readers must sign in or
register. And to ensure the quality of the discussion, our moderating team will review all comments and may edit them for clarity, length, and
relevance. Comments that are overly promotional, mean-spirited, or off-topic may be deleted per the moderators' judgment. All postings become
the property of Harvard Business Publishing.

The Perils of Algorithm-Based Marketing

Uploaded by

Copyright:

Available Formats

The Perils of Algorithm-Based Marketing

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Perils of Algorithm-Based Marketing

Uploaded by

Copyright:

Available Formats

MARKETING

The Perils of Algorithm-Based

Algorithms help marketers utilize customer-specic knowledge demographics, previous

INSIGHT CENTER First, understand the context and make it part of

New tools and strategies. drivers of customer response. Constantly testing

Utpal M. Dholakia is the George R. Brown Professor of Marketing at Rice Universitys

This article is about MARKETING

Related Topics: GROWTH STRATEGY | INTERNET

We Innovation a year ago

JOIN THE CONVERSATION

You might also like