Abstract
Data Analytics in Retail industry or in short,
Retail Analytics is the process of gathering data from various sources and
using it for the making educated decisions. These decisions could be anything
that supports the retail business operations, such as market analysis, customer
spending, promotions, service and product procurement planning, etc. There are
several sources from where the data can be gathered and it is up to the
business to mine the relevant information and categorize it use it to optimize
and grow the business. Using the information gathered, these retail companies
can make smart decisions that could very well shape the future of the
organization. This could effectively eliminate the guess work and the errors as
it the proposals and the investment decisions would be backed up by relevant
proof.
This document throws light on some of the
different steps and layers involving data analytics that are applicable to
retail industries from the source to the final cleansed data and the various
application in retail domain. The document also highlights some of the key
benefits and challenges associated with this process and the potential future
of the retail analytics.
Keywords: Oracle in Retal, Data Analytics, Data Cleansing, Retail
Data, Data sorting, Information categorization, Data Compartmentalization, Data
Organization, Processing Raw and Unstructured Data, Data Mining.
1. Introduction
Retail Industry, in simple terms, is the mode
of business which procures products in bulk from a wholesale vendor or directly
from a producer and sells it to end consumer. This is also referred to B-to-C
or Business-to-Consumer type of business. In the past two decades, the retail
industry has transformed phenomenally, both in terms of how the customer s shop
and the way the retail companies do their business. Unfortunately, there are
several large retail chains have failed read the market and adopt leading to
loss of market share, at times complete shutdown and the ones still standing
are those who have made smart and data driven decisions and moved in the right
direction. Data Analytics is the process that helps provide this insight. As a
matter of fact, data analytics needs to work the hardest in retail industry,
because, unlike other industries, these companies deal directly with the end
users who could be in millions and each customer is unique and are potential
input provider and this can come in any form, such as verbal feedback, survey
forms, online reviews, social media influencing and comments and the choice are
endless. While it is true that this feedback is needed even by the product and
wholesale companies, it has a direct and immediate impact on retail industries.
Also, online shopping taking retail by storm, it has forced them to think of
ways to attract customer well beyond the conversional ways, such as visuals that
they see online, the delivery time, cost benefit and of course the quality. The
ability of the retail company to dissect this information and properly channel
it can assist in some of the major decision such as the market demand, the
inventory they need to hold, the promotions that they can offer. More macro
level pointers such as political conditions, weather and the awareness of the
consumer can help the retailer plan and predict the market swings.
2. Need for Retail Analytics
The new retail age is both glorious and a
challenging one. While it has opened doors for anyone with an appetite and
talent to start their own retail business, it has increased the competition and
the customer’s demand for personalized and cost-effective products. These
retailer, big and small, need to leverage the data analytics in some way to
tune their business such that they stand apart from other to their customers.
Additionally, it is important that the internal operating practices be reviewed
to find what works and what does not and make timely and necessary changes to
plug the gaps, which otherwise might lead to wastage or time, money and
resources.
Today’s market can be very unforgiving.
Customers no longer want to compromise on the products they buy and are ready
to scorch through the internet to get what they want. With the modern search
tools such as visual search and AI it is not that hard to find them. Also, with
globalizations and open market practices, the customer demands have widened and
enabled them to buy from any part of the world within the legal limits. These
changes have forced the retail industries to rethink their business strategy and
the operating models. With the facilities today, anyone with an idea and the
business acumen can start a business. Unlike the large retailers, they do not
need large warehouses or a fancy outlet to do their business and with online
shopping abilities they can pretty much reach anyone who is looking. Though,
these cottage industry retailers might not be a serious threat for large
retailers today, they can surely take away few targeted customers within the
locality and circle of influence and in due course of time corrode their
profit. To keep up with the growing trends in the market the retailers need to
rely on the findings from data analytics to guide and decide their direction in
several areas both internal and external.
Unlike peer-to-peer marketing, retailers need
to have warehouses to stock-up their inventories and outlets. Though online
shopping has increased exponentially, still majority of the of the shopping is
done in-store and people still would like to touch and feel the products before
buying them. However, nowadays, the retailers cannot expect their customers to
wait or return back if the desired product is not in stock, they decide to move
on. On the other hand, the outlets need to display such that it helps the user
to shop with ease and entice them to buy more by strategically placing related
products within the close proximity so that the end customer by see a wider
picture. For instance, in a clothing retail, placing winter jackets and beanies
next to each other would attract the customer to buy them both rather than two
ends of the store, in which case the user might just buy one or be frustrated
with not finding the right pair and leave without buying.
With the help of data analytics, these
retailers can gather, process and identify patterns to optimize their
inventories and outlets. Information such as the shelf time, search frequency,
the customer demographics and special events and weather conditions can be
collected and mapped with sales patters to decide when to stock up and when
not. Having incorrect or insufficient information in this regard can lead to
serious consequences, such as running out of stock leading to missing revenue
or procuring the products at a premium price from suppliers or over stocking,
forcing the retailer to offer unfavorable discounts or even total wastage.
Also, by analyzing the buying patterns, the retails can identify the products
that are usually bought together or looked for and smartly arrange them
alongside each other aiding the customers to buy them easily. Also, these
patterns can be used to optimize allocation and redistribute the products to
different geographical locations based on special events and local holidays to
meet the customer demand or to explore new market based on developments and
migration.
These are only a fraction of what data
analytics can do to retail industry, such as sales analysis, employee
reskilling, production planning and optimization etc. Depending on the need and
the focus area, data analytics can be a great partner in restructuring the
retail industry in every operating domain.
3. Type of Data Analytics in Retail
The Descriptive Analytics is the type which
breaks down the entire process into smaller chunks, categorizes it and lay it
down for further analysis. This type acts as the base for further analysis, by
providing detailed metrics in every dimension such as volume, time, revenue,
location, customer response, surrounding conditions, etc. This forms the core
of the retail analytics since all other analytics are based on the outcome of
this type. Hence extreme caution needs to be taken in design these analytics to
ensure all the relevant data and numbers are captured to the smallest level
possible.
Performance Analytics is the type in which the
system continuously tracks the flow to provide insight on the effectiveness.
This can either be real-time or batch analysis. This type of analytics usually
has pre-set target and conditions along with periodic milestones and
checkpoints. Any deviations to these metrics will be analyzed to identify the
factors influencing these changes and the impact. Additionally, the rate of
change and deviation and the impact of this is analyzed and identified. This
impact could be positive or negative. If positive, the case is closely
monitored and will be extrapolated to a wider range and if not, counter
measures are suggested to bring it back to track.
The Diagnostic Analytics is the type where the
data is analyzed after an event has occurred to identify the factors that made
it happen. It is more like a post-mortem of an event or series of event to find
what worked and what did not. Similar to performance analytics, there is a
pre-set target based on which these analytics is done. In this type, the
information is gathered form every source and for every factor surrounding the event,
it tries to identify the root cause or the combination of actions that led to
the outcome. While these analytics are predominantly used in negative
situations, it is fully relevant in positive conditions too, where the outcome
would have exceeded the client’s expectations. A detail analysis of these
strand of flow can show what point the curve took a sharp turn and if this can
be sustained or manufactured for the benefit of the retailer.
The next type is the Predictive Analytics. As
the name indicates, this analytics is used to forecast the future events. It is
based on the mantra “History repeats itself” with the idea that given the same
situation and influencing factors, the chances are that the outcome will be the
same. This type uses historical data and the factors preceding it and
succeeding to identify what might be the next big thing that the business that
the business can expect. It reads the current events in silos and in connection
with the other factors and scan for similar situations recorded and the maps
the outcome. These analytics can either be external such as market predictions,
the next global trend, etc. or internal such as efficiency, financial solvency,
employee and customer reactions and response and the subsequent domino effect
that it might cause. Typically, the predictive analysis provides multiple
possible outcomes in the given situation, which the management can weigh in to
identify the best net advantage for the company.
Finally, the Prescriptive Analytics, where the
suggestions are made to the company on what the next steps should be based on
where the retailer is, what the target and the market climate. This analytics
uses the knowledge gained by reading the market on various scenarios and its
outcomes, wither within the company or market as a whole and makes suggestions
on what would be the best course of action. For relevant prescription, the
system to have a complete and accurate knowledge of the company it suggests
for, such as their financial status, the long term and short-term vision,
business model and culture, operating model in terms of both strength and limitations.
and the external knowledge such as the market conditions, political situations,
customers spend capacity, their preference, etc. It is important to note that
the same situation does not guarantee the same outcome for two different
organizations.
4. Best Practices in Retail Analytics
There are several ways to define and implement
retail analytics to assist the retail organization in making educated data
driven decisions. There is no one right way to go about it as every
organization is different in every way imaginable. However, there are certain
fundamentals that needs to be kept in mind while designing the solution as
these are key for successful analysis and accurate outputs.
The Data Analytics system should be able to gather
data from multiple sources, formats, domains and anything else that is deemed
relevant. As mentioned above, effective descriptive analytics forms the base
for all other analytics and this can only have a large diverse set of data in
the central pool that the system can read, decode, categorize and store. The
system should not only rely on a templated data source but be able to read
through various sources and formats. With the new generation, there is a whole
new language used for conversation, such as emojis, memes, etc., this layer
should be able to scan and understand the emotions and feedback provided. Also,
with every age group there is a different slang and way of shortening or use of
phrase to express themselves. There needs to be a constant growing repository
of such jargons with relevant indicators to understand what is being actually
said. For instance, “Right” and “Yeah! Right!” mean tow completely different
things. While the former is the sign of acceptance, the later is a sarcastic
tone for disapproval. These understanding is a rightly capturing what the
customer says about the product.
The next point is that there needs to be a
clear set of goals. The organization as a whole, can have a large long-term and
short-term goal, but for the retail analytics to be effective, there needs to a
concise and acute sent of goals and measurements that the system needs to work
towards. Having a generic goal, such as maximize profit or optimize inventory,
can be too vague to the data analytics tools to provide any accurate result.
This might spit out too many solutions or open-ended prescriptions, which might
be too complex to understand and implement. While the data collection should
look for any and all data available, there need to be clear and concise target
and controlled feed of relevant information into the systems to get clear view
of the data transformation in each stage. Flooding the system with all
information available, might slow down the analytic process and in worst case
generate undesired result. Even the relevant data needs to be properly layers
and prioritized based on the company need identify the best possible path with
most benefit.
The Key Performance Indicators (KPI) needs to
be constantly monitored to track the progress made. These are the various focus
points identified by the organization or by the market which shows if the
retailer is performing good or bad. For instance, a retailer can have the KPIs
as the weekly sales, online traffic, inventory movement, revenue, operating
cost, etc. The upward or downward movement of these indicators would mean if
whatever the organization is doing works for or against the end goal. While
adopting the suggestions made by the retail analytics systems, these KPIs needs
to be periodically monitored to see if the desired results are achieved. If not
the prediction and prescriptive algorithms needs to be revisited and tuned to
factor in the new conditions.
The other key practice is to constantly learn
and evolve. Any organization, retail or not, which has failed to adopt to the
changing market has failed miserably and there are some really big brands that
can be shown as an example. There should be a constant experimentation with
trying various permutations and combinations and sampled at the right time to see
how the market reacts and perfect it. All algorithms, assumptions and
categorizations need to be reviewed and refined to best suit the current
market. Though it is recommended to have a concise target set for retail
analytics to work efficiently, there should also be effort made to wire the
various findings to observe the movements in various functional domain.
5. Conclusion
The need to collect data and use them for one
own benefit has been growing at an exponential rate. It has crossed the lines
of organizational scope and has expanded to analyze global and political trends
to plan one’s business. With the growth of AI and ML, data analytics has seen
capabilities that was never possible before. Yet, with the constant shift in
the way the information is produced, such as new jargons, vernaculars or
slangs, the algorithms, tools and the methods need to continuously adopt with
the new trends.
Though the way retailing operates has changed,
domain by itself cannot go out of relevance. Also, with online shopping
revolutionizing the retail industry, by leveling the playing field in terms of
customer reachability, has forced every retail company who wants to sustain in
the market long term to use every means necessary to benefit their business and
data analytics can be a great ally in decoding public’s thoughts into the next
big thing, benefitting themselves, the investors and their customers.
6. References