Beruflich Dokumente
Kultur Dokumente
We found data is equivalent to oil while economic growth is taken into account.
With the use of oil , transportation has been way better than what it used to be in
long back in centuries, when animals were the option .So now we are supplying
adequate amount of goods as per the demand in different part of geography . So it
definitely takes our economy to higher stage . Similarly we have now machines
which are capable of producing as per the demand using oil as driving force . So
here also its an add-on to economy.
In same way data also taking ahead economy .By analyzing huge amount of old
data ( like huge amount of oil) , we can find the hidden pattern in behavior ( it
might be human behavior or behavior of weather or behavior of wind ) . For
example if we could find a pattern like in every odd year there is less rain , then we
can grow some crop that requires less water during odd year. Doing so we are
avoiding loss not by growing the crop which require more water ,also gaining
profit by growing less water consuming crops . So here also its an add-on to
economy.
IMPORTANCE OF DATA:
Data can be defined as the raw material and unprocessed information which needs
to be processed to become useful for its user. Processing data is an essential
requirement to convert data in to information so that it can be beneficial to its user.
The importance of data shall however never be undermined. Data can be
considered as a crude oil which is not much useful on its own but can become
useful by getting converted in to plastic, petroleum etc. according to the wish of its
user. If looked upon from the other dimension it can be said that to have petroleum
it is essential to have crude oil. Similarly in business to have a strategic plan it is
essential to have data.
ANALYTICS STRATEGY:
A company needs to develop strategies for marketing, sales, human resources and
operations. Getting the right information means knowing what information is
important to the company's decision-making process. The strategy starts with basic
demographic data; then, it considers pricing based on the education and income of
buyers, and how that group speaks. Education and income are important, because
the more educated and the higher the income a target group has, the more likely it
is that the business can logically sell a higher-end product to the group who can
understand, appreciate and afford the product.
For example, a company wants to market a new skin care line to women. There are
several key factors the company must know to effectively design and execute a
marketing strategy. Although the product might work for all women, will it be
branded more for the younger teen crowd? Or, will it be branded for the aging
woman who's trying to hold on to her youthful skin? The label design might be
different for each group. A higher price point might not be affordable to a younger
crowd, but research backing the claims might be required to sell a product with a
higher price point to educated, professional older women.
In a scenario like this, the company needs data on the cost of goods sold, with the
ultimate product pricing and who is most likely to buy it. It will design the entire
packaging around its biggest segment, even though others may buy it. The
analytics took not only numbers and group sizes, but what a specific demographic
respond to visually. Packaging for a young Millennial group is much different than
it would be for a Baby Boomer demographic.
It isn't enough to have the data. Great business leaders use data effectively to make
decisions. Assume that a cell phone manufacturer uses components made in China.
If tariffs are put on that line of components, the costs of that phone costs will rise.
Understanding the effects of that data and what it does to the business owner's
profit margin is imperative in developing a strategy that keeps the company
profitable. The business leader might seek to find a new resource for components,
which are less likely affected by tariffs or international political situations. The
business leader might decide to raise the cost of its final product, effectively
passing on the new cost to the consumer.
Initially, the company could also decide to do nothing and absorb the higher costs,
because it has enough of a profit margin to still be successful, while many of its
competitors will need to raise prices. This strategy is one that could reduce the
profits-per-sale, but it might increase the demand for products, because it is more
competitively priced for consumers; thus, the demand increases.
RECOMMENDED USE OF DATA FOR THE ORGANISATIONS:
It is essential for the organization while analyzing the data to not consider it as a
fact. Data and facts are two very different things and should not be confused while
planning or decision making otherwise it will lead to wrong plan or decision. The
major difference being that basic data is not always fact but a fact can be a data for
further analysis. Where data is unprocessed information, a fact is true and valid
information.
It is therefore necessary for the organization to screen data and facts from the
information being provided to it for strategically functions. As Taiichi Ohno of
Toyota Motors clearly states that data and facts are different by saying:
It is necessary that the available data is used in accordance with the context of the
market place or of the situation related to the data. The study of a segregated data
without it context will not bring the required meaning and effectiveness to the data
study as it should. A data studied without its context will lead to a decision which
will be inconsistent with the other decisions of the organization.
DATA AS ASSETS:
Data are raw alphanumeric values obtained and owned by data producers. When a
data producer uses data for operations, they can be treated as physical assets. There
are mature methods for assessing the value of physical assets. When data are used
for other purposes (decision-making, regulatory, and research) they often are
treated as intangible assets. These assets are more difficult to value and the
methods are less mature and less precise. Unlike a ‘typical’ asset that provides
value to the organization that owns it, data also have immense value for secondary
users creating value for multiple organizations for multiple purposes at the same
time. Here, data behave as a derived asset (or non-rival good) whose value is tied
to an end use. Methods for valuing derived assets are in their infancy.
A data producer is a user interface, automation, service or device that collects data
that is relevant to a business. In many cases, multiple systems and processes may
produce data for the same data entry. For example, a customer record may be
updated by marketing, sales, point of sales systems, customer service and self
service tools used directly by the customer.
The Historical Cost method assumes that data producers behave rationally and will
only spend the money needed to acquire an asset if they will receive at least an
equivalent economic benefit in the future. The method assumes a return on
investment (ROI) greater than or equal to 1. Data range from potentially having no
value to enormous value. For example, some organizations collect data first and
decide how to use them later. As a result, sometimes data are never put to use and
are only a cost. At the other end of the spectrum, data may be used to inform high
impact decisions, resulting in the data being far more valuable than their
acquisition cost. Not all data are created equal. To address these limitations,
Moody and Walsh (1999) created the Modified Historical Cost method. This
method adjusts for the value of the data based on the unique attributes of data, such
as the potential for an infinite number of users, the quality of the data, and that
duplicate data have zero value. This method has a bottom-up approach where data
collected are not tied to an end use, and assumes data are valued in and of
themselves, without being tied to impact (Figure 1).
Figure 1: Data’s value is estimated based on data costs and usage attributes.
The Modified Historical Cost method adjusts the cost of data collection based on
the attributes of the data.
Data costs include labor, equipment, and infrastructure. There are upfront capital
costs to install equipment as well as ongoing operation and maintenance costs.
There may be additional data management costs to store, process (such as quality
control), and use the data.
Beginning with the assumption that value is equal to the cost of collection, the
following modifications are made to account for data’s unique attributes.
Unused data have zero value. Data usage can be estimated by number of
downloads.
The more data are used, the greater their value. Therefore, the number of users and
data downloads serve as a multiplier to the value.
The value of the data are depreciated based on data purpose. Operational data for
day-to-day decisions have a short life span, while other data may have value for
decades based on the decisions they inform. Depreciation rates may be based on
the data purpose if usage is unknown.
Example application
IoW Water Utility collects operational data from sensors to inform day-to-day
operations. IoW Water Utility also uses public precipitation and water quality data
to inform broad decision-making. Twenty five percent of the data collected by IoW
Water Utility are for regulatory purposes only.
$10,000 to buy and install a sensor. There are 8 sensors. Total = $80,000
$1,000 annually to maintain and operate each sensor. Total = $8,000 per year
The total cost in year one is: $86,000. The annual cost for each subsequent year is
$4,000. By the end of year 10, $126,000 will have been spent on data with an
average of $11,450 spent annually.
MARKET METHODS: -
Market Methods can be used to value data by understanding the user’s willingness-
to-pay. Market methods assess the value of data as revealed through markets and
experiments or as stated on user surveys.
Market methods are focused on understanding the customers’ willingness-to-pay
(WTP) for a product or service. The methods for assessing a user’s WTP for data
vary depending on whether data are treated as physical or intangible assets. When
data are in the market and treated as physical assets (i.e., end users are charged per
download, service, or subscription), the WTP is revealed (revealed preferences)
When data behave as intangible assets outside of a market, their value can be
estimated through stated preferences, where users are surveyed to understand the
value of data to them and their corresponding WTP (Figure 1). The methods to
assess WTP are contingent on whether data behave as physical assets (revealed
preferences) or intangible assets (stated preferences). We understand revealed
preferences from market data and experiments, while stated preferences are
directly or indirectly uncovered through surveys.
Revealed preferences
Market data methods are only applicable if water data are a commodity where
actual sales are occurring. The value of the data is the market price that balances
supply and demand. Examples of water data entering the market include charging
data users per:
Once in the marketplace, the value is found by observing the impact of each price
adjustment on the number of users and/or downloads.
Typically, charging for raw data access leads to decreased usage or market failures
due to data’s unique attributes (such as infinite users for small marginal costs) and
their value is tied to end use. As a result, for data to enter the market they often be
transformed into higher quality products or services (information). For instance,
B3 collects, cleans, and combines water rights and property data to understand the
water security of future investments. WatrHub collects, scrapes, and converts pdfs
into digital data that match upcoming utility infrastructure needs with potential
technological solutions.
Experiment methods
Field methods attempt to overcome these biases by following real customers. For
instance, a company may vary prices for the same sandwich at different fast food
store locations. Location A may offer a sandwich for $4.50, whereas Location B
offers it for $7.00. The difference in the number of purchases based on price can be
used to determine the value the sandwich to consumers (controlling for other
socioeconomic variables). In the digital economy, companies may provide two
versions of a website and collect usage statistics to determine the higher value
website. A similar approach may be used to discern which data or interface is most
valuable to users.
Stated preferences
Expert Judgment
methods provide a quick and cost effective way to estimate value. Expert judgment
could be immensely valuable when there are known use cases for data. For
instance, the value of data in a flood event will be better estimated by emergency
responders than someone in a non-related profession. Currently, the value is often
communicated via anecdotal stories rather quantitative value. This method is best
suited when there is a clear use case; however, over-bias by experts is a concern.
Customer Surveys are used to identify (1) the price above which you would not
buy a product because the cost is too prohibitive and (2) the price below which you
would not buy a product because the quality becomes suspect. Customers are asked
to write-in their WTP. This typically creates over-bias because it is easy to
generously value a product on paper, and stated preferences do not mean they
would necessarily buy the product in real life. Additionally, stated preferences are
not stable. The same survey sent to the same customer could produce different
results depending on survey timing or customer mood when filling out the survey.
Because of these biases, customer surveys are generally not recommended.
Contingent Valuation asks respondents their WTP to pay for a specific good, or
their willingness to give up a good. This method is preferred for intangible assets
(such as water quality or ecosystem services) where market failures are
widespread. Despite skepticism, contingent valuation has been widely adopted by
government agencies to assess the value of non-market services. For instance, the
USGS used contingent valuation to understand the value of Landsat imagery and
users’ WTP using a series of Yes/No questions combined with randomly assigned
dollar ranges. This method controls for over-bias by providing values rather than
allowing customers to write-in values.
the full price is rarely assessed as much as the additive value of different
attributes,
it is not clear that varying attributes and prices is additive (rather than
multiplicative).
A data user is defined as a person who makes use of personal information for a
certain purpose.
The Business Model Maturity Index method values data relative to how much they
influenced an outcome by making a better-informed decision. This method
progresses through identifying a desired outcome, developing use cases to achieve
that outcome, and the relative contribution of data to each use case. The Business
Model Maturity Index’s top-down approach prioritizes which data to collect based
on the desired outcome (Figure 1).
The value of data is tied to its impact. A desired impact is more achievable when
data are transformed from descriptive (requires expertise to translate to action) to
prescriptive (minimizes the expertise gap) information. Let’s explore the difference
between descriptive, predictive, and prescriptive information.
Friday.
The Business Model Maturity Index method progresses from (1) identifying a
desired outcome, (2) developing a series of use cases to achieve that desired
outcome, and (3) using expert judgment to estimate the relative contribution, and
value, of data to each use case. The last step is to estimate the return on investment
for data.
Use cases refer to decisions and strategies that can achieve the desired outcome.
For each use case estimate the implementation cost and the potential impact. The
value of the decisions should be accounted for over the full life-time of the project.
Consider all the data needed for each use case and the relative importance of those
data to achieve the desired outcome. For each use case, assign a value from 0 (no
value) to 1 (critical value) denoting the relative contribution of each piece of data.
Sum the relative contribution of the data in each use case. The value of the data for
each use case is estimated as:
The return on investment (ROI) for each use case is the benefit divided by the costs
(estimated or real) to implement the solution. The ROI for the data within and
across use cases is the value of the data divided by the cost of data collection.
Final Thoughts:
The Business Model Maturity Index method assumes no decision will be made
without data. However, the value of the data should be linked to improvements
from the original decision. For instance, IoW Alfalfa already floods its fields in
March to recharge groundwater. However, the data show flooding can be extended
till the 3rd week of April without hurting productivity. The value of the data is the
change in groundwater depth and subsequent energy savings that come from
extending the length of time the fields are flooded.
Use cases refer to decisions and strategies that can achieve the desired outcome.
For each use case, estimate the implementation cost and the potential impact. The
value of the decision should be accounted for over the full life-time of the project.
(3) Adjust value based on how fit-for-purpose data are to inform decisions
The value of the use case is adjusted based on the frequency of data collection and
their quality relative to what is needed to make a good decision.
Frequency
Frequency compares how often the data must be collected to inform decision-
making. The frequency tolerance is the window around which decision-making
will be impaired if data are collected less frequently. The adjustment is:
Data accuracy is compared with the required accuracy for a use case. The
adjustment is:
When the quality is not known, expert opinion can be used. The recommended
categories are:
Quality Modifier
The Quality Modifier ranges from 0 to 2 and is the sum of the Frequency and
Accuracy score. The estimated value of the data is adjusted by multiplying it with
the quality modifier.
Labor
Hardware
Software
Utilities (such as building costs, energy costs, and so on)
If the data are currently used for a single decision point, then the full cost is
weighted against the benefit of that decision. However, if the data are used for
multiple decisions, then the cost can be equally divided between those decisions.
Additionally, infrastructure can be depreciated or amortized annually.
ROI <1: the benefits are less than the cost of the project
Note that while the economic ROI might be less than one, the reputational, social,
or environmental benefits may still justify the cost of the project. Initially the
potential ROI is calculated, but as the project occurs, the actual ROI can be
estimated by adjusting according to the Performance Value.
The value of the data can be estimated based on their relative importance to the use
case. The importance ranges from 0 (none) to 1 (critical).
The value of the data can also be weighted by the level of difficulty to transform
the data into usable information, such that data are more valuable with lower
processing time.
The value of the data is multiplied by its relative contribution and divided by its
processing value:
The ROI for data is estimated by dividing the adjusted value by the data’s cost.
DATA HUBS
Data Hubs are ideally the "go-to" place for data within an enterprise, so that many
point-to-point connections between callers and data suppliers do not need to be
made, and so that the Data Hub organization can negotiate deliverables and
schedules with various data enclave teams, rather than being an organizational
free-for-all as different teams try to get new services and features from many other
teams.
Data hubs may use their annual expenditures as the cost or they may track costs by
dataset or by individual data producers.
The starting value of the hub is then adjusted by data usage statistics. The value for
individual datasets can be weighted at zero (no usage) to one (maximum usage)
based on either the number of downloads or the percent of each dataset accessed
by data users. This means the value of the data are relative to their use within the
hub. Since the benefits are realized by data users, the benefit-to-cost ratio is used to
compare costs and benefits (similar to a return on investment but does not assume
the same organization accrues benefits and costs).
The value of the hub is then further adjusted by dataset attributes. For instance,
redundant or duplicated data have no value. Data may be weighted by their quality
from 0 (very poor) to 1 (excellent). For data of unknown quality, expert opinion
may be used. Valuing datasets within the hub allows the hub to prioritize which
datasets are of high value to their users and how to maximize the value of the hub.
(additional tools and resources) and seeks to calculate the annual (and long-term)
costs of data hubs, as well as the value of hubs to data users. Hub benefits are
categorized in terms of
cost (time and money) for users to access the hub’s data,
the savings to users from using hubs versus individually accessing data from
producers,
the impact from actionable insights resulting from the data (use cases).
The KRDS method compares the cost of the hub to the value it generates by
making data more discoverable, accessible, and usable. Hub value is assessed
through predominately market methods.
The first step is to identify the cost of the data hub. This can be as simple as using
the yearly expenditures or as complex as breaking out costs by individual
employees or datasets. KRDS also enables users to depreciate the cost of hardware
and adjust for inflation.
Use cases refer to decisions and strategies that can achieve the desired outcome.
For each use case, estimate the implementation cost and the potential impact. The
value of the decision should be accounted for over the full life-time of the project.
(3) Adjust value based on how fit-for-purpose data are to inform decisions
The KRDS method provides three dimensions to considering each benefit: impact,
timing, and who benefits.
Impact
Timing
Timing can either be short-term or long-term, and refers to when benefits accrue
from the data and has implications for hub funding strategies. Timing to realize
benefits are linked to data purposes.
Long-term benefits take longer than 5 years to realize and tend to be used for
purposes of broad decision-making and research.
Example: Data from a hub are used to develop a water quality model
eight years after the inception of the project.
Who
Who refers to whether benefits are realized internally or externally to the hub.
Data hubs may calculate internal, direct, short-term benefits within their
organization. However, for water data hubs, much of the value is likely to come
from external data users, requiring survey methods to understand how data are
being used to create impact and the willingness-to-pay for hub services. Surveys
may also be used to improve hub usability for producers and users.
The ROI can be calculated for producers, hubs, and users. For example, Bergie &
Hougton (2014) applied the KRDS method to three data centers and found:
Producers had an ROI between 2 and 12 to host data on a hub rather than
through their own services,
Hubs had an ROI between 2 and 23 in terms of increased efficiency in
providing data from multiple producers in one location.
Users had an ROI between 2.2 and 2.7 in terms of willingness-to-pay and
the estimated value of the hub from use cases.
ROI is solely quantitative, however many benefits are provided that are difficult to
put into economic terms. KRDS encourages hubs to articulate these qualitative
impacts through surveys that reveal stakeholder values and how hubs may better
meet stakeholder needs, thereby raising the value of the hub