Advancing Engineering with AI through the Next Generation of Strategic Projec...
Using Machine Learning to move towards Customer-Specific Pricing
1. technologyrevenue management
32 | March 7, 2016 | hotelbusiness.com
By Nicole Carlino
Senior Associate Editor
NEW YORK—Revenue management
is one of the most important fields in
the hospitality technology world these
days, so it’s no surprise that it’s an area
that’s garnering a lot of focus and inno-
vation. The latest entrant into the field?
LodgIQ, a NYC-based startup that has
secured $5 million in seed funding led
by Highgate Ventures, a venture capital
platform focused on early stage tech-
nology investments in the hospitality
industry, and Trilantic Capital Part-
ners, a global private equity firm.
“LodgIQ is really about us funda-
mentally disrupting the traditional
revenue management technology
paradigm in terms of how demand is
forecasted, in terms of how pricing
is calculated and then how we can
actually use all of these different data
sources and interpret the right signals
at the right time for decisions and rec-
ommendations,” said Ravneet Bhan-
dari, CEO of LodgIQ.
Bhandari elaborated on some of the
challenges hotels can face. “There are
so many different signals that are com-
ing at hotels or at revenue managers.
Customers are shopping more than
they ever have, and rates have more
transparency in the marketplace than
they ever have,” he said. “One of the
fundamental tenets of revenue man-
agement always was that you could
price differentiate and put fences
around different rate types. That’s
gone. Customers now react to not just
price, but they also react to things like
review scores and content—specifically
images.” Bhandari noted that a reve-
nue management platform should take
that into account. “That’s one of the
fundamental reasons why we wanted
to do this. If you’re able to incorporate
all of the different decision and data
elements into a single composite al-
gorithm, by definition, you’re making
better decisions,” he said.
The CEO has more than 20 years of
domain expertise in leadership and ad-
visory roles with companies like Hyatt,
Caesars Entertainment and Starwood
Capital, and noted that his team has
similar industry experience. According
to Bhandari, many of the established,
legacy players use forecasting and op-
timization algorithms that are two de-
cades old. “The operating environment
has changed, and there’s a lot of noise
out there,” he said, noting that there
is a lot of data available now with re-
gard to buyer intent. “The vision from
the beginning was that we would be
able to create a multi-source, big data
infrastructure and then disrupt the
forecasting and optimization meth-
odologies through advanced machine
learning. What that really enables us
to do is to incorporate this notion of
buyer intent and all of the signals that
are out there into the revenue optimi-
zation decision, which eventually really
means customer-based or customer-
specific optimal pricing.”
LodgIQ deployed its platform at five
New York City hotels last month. The
company has two products: a web-
based system and a mobile RMS. “Our
full-featured mobile RMS is perceived
as our entry level product for smaller
hotels and select-service properties,”
noted Bhandari. “Very specifically, we
want to initially stay focused on the
top 50 global lodging markets. We are
targeting everything from full-service
hotels to select-service properties,
independents, asset management com-
panies and the larger brands.”
LodgIQ incorporates a range of data,
from flight schedules and weather pat-
terns to events into its platform. “We
are consuming, for instance, vacation
rental demand data. We’re consuming
forward-looking market demand data,”
said Bhandari. “Obviously, we have
competitive rate shopping. We are con-
suming things like newsfeeds in real
time from Associated Press and a few
other data sources.”
The heart of LodgIQ RM is its
machine-learning platform, which in-
cludes Databricks, a company founded
by the creators of Apache Spark,
and which is used by companies like
Amazon, Netflix and Google. “We
can throw any unstructured data at
Databricks, and it’s very quickly able
to establish the relationships between
disparate data sets,” explained Bhan-
dari. Dato provides the second part of
the machine-learning platform. “That
allows us to do things like semantic
keyword extraction,” said Bhandari.
“For instance, to take a very specific
example, how do you, in terms of a
real time newsfeed, extract the key-
words that are relevant to being able
to forecast either market demand or
perceived price elasticity? How do you
establish positive and negative correla-
tion to news or events or other stuff
happening in your marketplace? One
of the fundamental ethos or constructs
of our data schemer is this notion of
any data, any source, and that’s some-
thing we’ve tried very hard to be able
to incorporate.”
A user-friendly user interface (UI)
was also a focus for the company.
Bhandari noted, “We have so much
data, and so many different data
sources, that we have to make sure we
present it in a manner that is easily
consumable.” Bhandari also said that a
vast majority of users of revenue man-
agement technology are Millennials
and Gen Xers—people well versed in
user interfaces in the B2C world—so
LodgIQ wanted to bring B2C intelli-
gent usage into the B2B environment.
“We focused very heavily not just on
design, but on this notion of intelligent
design. Our UI actually morphs intel-
ligently to different user types,” he
said, adding that a revenue manager
and a GM would have different needs
from a revenue management platform.
“If you’re a GM, very quickly, in about
10 days, the UI will recognize what
your clickstream is, and it’s only go-
ing to present information relevant
to you,” he said. “Our objective is for
people to spend time in the system,
not use the system as just another tool
to export information into an Excel
environment where they marry it with
other data and try to make decisions
based off of that.”
Bhandari looked toward the future
of the revenue management platform
and the field in general, noting that
dynamic behavioral segmentation
is something hotels should focus on.
“Traditional revenue management…
totally ignores the behavioral com-
ponents of customers’ demand. He
pointed out that leisure travelers flying
from the U.K. to New York, who book
flights 30 days out and hotels seven
to 15 days before a trip, are a unique
customer segment. “What is the price
elasticity of that behavioral segment
vs. someone else? That’s the notion of
dynamic behavioral segmentation, and
it’s something we have incorporated
into our technology,” he said.
“The perfect manifestation of revenue
management is one-on-one pricing at
the point of purchase as opposed to
using this notion of a hurdle rate or
marginal rate or the BAR rate, which
is one price typically charged to most
customers depending on what market
segment they belong to,” he continued.
“How do you eventually make the opti-
mal decision at the point of purchase in
real time when the customer’s in the act
of booking? That’s where we have to go
and we’re setting the stage for that.” HB
Using machine learning to move
toward customer-specific pricing
LodgIQ incorporates a range
of data into its platform.