2. The Sufficiency Problem
Generally the more
reviews you have, the
more they converge on
a consensus assessment
of the experience. That
suggests to the prospective
customer that the
experience is very
predictable and low risk.
Some experiences are polarizing: some people love them while
others hate them. Sometimes random chance will bring
these two sides
together in near
equal numbers,
with results
confusing to the
prospective
customer.
How do
you know?
3. Was this review helpful?
There is another dimension to
many review sites, where readers
assess the helpfulness of reviews.
However, it is not obvious what
the implications of this
information are.
Explaining “helpful:”
• The writer’s total
“helpful” votes (85%)
• Length of review (5%)
• Mention of the word
“excellent” (1%)
4. Jumps in total
helpfulness
signal important
reviews.
A Business’ Helpful Peaks
After a certain point readers
consistently assess reviews
to be less helpful. This may
be the point were existing
reviews reach sufficiency.
The red line is a
locally-weighted
regression.
*Yelp review data from the area around Phoenix, AZ
5. Satisfaction Minimizes
At the approximate
point where “helpful”
peaks, customer
satisfaction seems to
minimize.
When the entity being
reviewed is sufficiently
portrayed, the proper
market begins to be
exclusively attracted
and satisfaction rises.
6. Matching Expectations with Outcomes
Since matching expectations with outcomes is the key to customer satisfaction,
consumers should be told when the available reviews might not yet allow them
to make an informed prediction of their outcome.
7. McPhee’s (1963) Theory of Exposure
There are four (4) consumer clusters based
McPhee’s Theory of Exposure:
on the proportion of niche venues in their
• Natural monopoly. The most popular
reviews:
Niche Proportion Clusters
products get the most users, and those
Field
1
2
3
4
who use them least.
Consumers 16565
9682
6910
5319 • Double jeopardy. Niche products have
Avg. stars
3.82
3.77
3.7
3.6
a double disadvantage: (1) they are not
-7168 12016 -2007 -2897
∑ Helpful
well-known; and, (2) when they
Niche Prop
2.7% 25.2% 48.6% 98.9%
become known, it is by people who use
Reviews
60090 124007 57599 11167
the popular products and prefer
Rev/Cons
3.2
12.5
9.7
2.1
them.
Check/Biz 5839.6 3576.8 1178.5 164.6
– But are they truly engaged by them?
Patronizing a mix of 26% niche
businesses does not yield the
highest satisfaction, but it does
inspire the most participation
and engagement. This seems to
be the ideal point of
adventure.
Hit venues
Niche venues
9. A Typical Business Network
• This network was
created by linking
businesses that are
categorized in the
same way by Yelp.
• It depicts a map of the
competitive landscape
as these businesses
likely perceive it.
• The competitive
perspective.
10. But there is another perspective:
Consumers link themselves to businesses in a hub-and-spoke network with their
patronage. The patronage of the same consumer connects the businesses
themselves into a network that in one way can be seen to compete for a share
of the consumer’s budget, and in another way sustain the consumer as an
ecosystem.
11. The Consumer
Ecosystem
•
•
•
The primary businesses
patronized by consumers
who also patronized Hotel
Tempe.
Note that the Mission
Palms hotel here was not a
member of the competitive
map, and none of the
competitors in that map are
in the consumer ecosystem.
The most important insight
though is that this provides
a rich multi-faceted portrait
of the consumption habits
of those who are patrons of
Hotel Tempe.
12. Applying the Ecosystem to Search Ads
Hypothesis: By shifting their repertoire of targeted searches beyond the obvious toward
those that are both relevant to their business and consumer needs, consumer-facing
businesses like Mission Palms can use search advertising more effectively.
13. A Nexus of Theory
• Market structure analysis (MSA) examines how products in
the same market compete more strongly with each other
than with those in different markets.
• The consumer ecosystem often depicts general brands and is
appropriately described as brand mapping; brand mapping
often uses consumer choice data similar to MSA.
• Many of the brands in consumers’ awareness compete for a
share of their disposable income across product categories.
The budget allocation research stream is often focused on
how consumers prioritize planned purchases.
• Business ecosystem research sees consumers and producers
as members of an economic community that coevolves in
their capabilities and roles (e.g., the increase in economic
activity that accrues to shopping centers that maximize
heterogeneous retailer agglomeration).
14. Measuring the Ecosystem Perspective
Mutual Information
I(A; B) is the mutual information between an
advertiser and its consumers, B are all the consumers
who clicked A’s ads, p(a, b) is the joint probability of
the advertiser and consumers using the same query,
p(a) and p(b) are the probabilities of either using a
query.
To what
extent is an
advertiser
targeting the
full range of
queries
made by
consumers
that click its
ads?
15. Hierarchical Logistic Regression
Parameter
Raw
Coefficient
Relative
Importance*
Query frequency
-.207
.901
-.432
-.039
-.371
-.514
-.048
-1.304
.36
.33
.15
.07
.07
.02
<.01
Ecosystem MI
Ad position
Query freq.× rel.
Past impressions
Relevance
HHI
Constant
*Relative importance is the percentage of the
overall R2 that is attributed to a variable.
Control variables from
prior research:
• Prior ad exposures
• Prior ad clicks
• Ad display position
• Competitive
interference
• General query
frequency
• Ad-query relevance
Nagelkerke’s R2 of .262
16. Authenticity & Promise-Keeping
• Advertiser’s
targeting of
ecosystem queries
must authentically
empower the
consumer to fulfill
more needs.
• Landing page must
match the ad.