Frequently Asked Questions
of Segmentation
What is a cluster?
The assignment to small areas of geography, such as the census block
group, to one of forty or more "lifestyle codes" established
through the application of multivariate techniques to detailed demographic
composition. The "clusters" are an impressive consumer segmentation
and predictive measure.
How long have clusters been used?
United States businesses have benefited from multivariate clustering
techniques for market segmentation since the seventies - however,
more widespread use of clustering occurred during the 1980's.
What is the difference between the clustering systems?
Lifestyle clustering systems depend on census data (both geographic
and decennial demographic information) as the fundamental metric(s)
for small area evaluation. Yet, differing cluster systems go beyond "just" census
data - both in development and in year-to-year updating. Today, clustering
databases utilize national list data, local community government
statistics - as well as, in some cases, aggregate financial purchasing
behavior. Another important difference can be found by examining
the variety of techniques used to perform the calculations.
Why do different systems have more / less clusters?
While all commercially available clustering systems may seem similar
because their underlying fundamental data sources are the same, all
systems employ a variety of unique data inputs. A variety of data
combined with numerous statistical processes and methodologies used
to calculate "homogeneous groupings" add to the mystique
of clustering. Simply stated - the multivariate distributions, contained
within each geographic unit (evaluated by averages and means) are
carefully measured and identified based on data input and modeling
technique. This process is often described as both a science AND
an art resulting in a varying number of classifications - dependent
primarily upon the "artist" and the tools and data available.
Keep in mind that the statisticians that create these systems for
marketing purposes have two main tasks in mind when they are formulating
cluster systems: to achieve groupings that are fairly homogeneous
while maintaining groupings large enough to measure and target efficiently.
One statistician may find that 50 groupings show enough discrimination
when used with certain data sources, another may choose and 80 or
even 100 cluster solution.
My company uses another system - can I (easily) convert to MOSAIC?
Yes, all cluster lifestyle codes are assigned to levels of geography
(either postal or census - or both) and can be easily assigned to
a file containing address information. The conversion (not matter
the volume) can be done in a matter of minutes or hours. To make
this process easier for you, Experian invites you to purchase a cost-effective "conversion
project" designed to let you see how MOSAIC classifies your
data compared to your current system. (Call CATENATE for details.)
What do I need on my file to use Segmentation?
Preferred is address, including street number, city, state, ZIP, ZIP
+ 4 and Census codes - State, County, Census Tract, Block Group and
Block.
What are the different levels of geography?
The Cluster codes are assigned at the postal level to ZIP Code and
ZIP +4 and on the Census level to Census Tract and Block Group.
Does Segmentation just work with ZIP codes?
No, Segmentation can be used with a variety of different levels of
geography. (See previous q/a). However, if you only have ZIP Codes
on your file, you can use Segmentation to profile your customer base.
However - most ZIP codes contain anywhere from a few hundred households
- up to and exceeding 20,000. Therefore - to achieve the greatest "lift" from
Segmentation - it is advisable to manage the lifestyle clusters a lower
level of geography.
What is geo-coding and why is it a necessary step?
Geo-coding is the process of enhancing your address information with
finer detail. Typically, this includes State, County, Census Tract,
Enumeration District, Block Group and Block. These are the geographic
boundaries used by the government in All State, County and Federal
surveys, analysis and evaluations.
What does `geo-demographic' mean?
During the decennial (every 10 years) census the government mails one
of two survey forms to every household in the Unites States -- the "long
form" or the "short form". The data collected forms
an enormous data repository, a veritable "Fort Knox" of
information on the American populace. This summarized demographic
information is evaluated and managed via census geography. Geo-demographic
-- the "marriage" of geographies and demographics _ has
become an indispensable tool for market analysis.
Can I use my own customer data and custom surveys with Segmentation?
Yes, the advantage of using a cluster segmentation system is the ability
to cross-tabulate most of your internal information and transaction
based information for analysis (profiling) and/or further multivariate
or advanced modeling.
What is the advantage of using a commercial segmentation system over
a custom model?
Custom modeling can be a key component to any marketing application.
Quite often the results are utilized exclusively for internal analysis
and for the internal sale culture. This greatly reduces the option
of understanding the market based on outside intelligence (market share
and competition). A commercial cluster segmentation database links
your data and your analysis to the vast resources outside your environment
(e.g. national survey data and lists).
Can I use both?
Yes, the Segmentation systems are extremely flexible, as a powerful
predictor in its own right - it naturally fits well into your own
data analysis and modeling.
I only have telephone numbers on my customer database - can I use
Segmentation?
One technique is to get a "reverse match" on your customer
file. By utilizing a mail house, we can take your list of valid telephone
numbers and cross-reference them with your customer database to append
actual addresses to your list. However, this process generally results
in a match only 60-75% of the time.
How can I use Segmentation to tailor my message to different customer
segments?
Each cluster not only describes the geo-demographic make up of an area
but also (via links set up to outside survey and data), outlines the
lifestyle purchasing preferences of the households. You can use Segmentation
to learn about your customer's vacation & sporting preferences,
media habits, product purchases and more. All of this additional information
can help you to craft messages that speak to your customers likes while
avoiding their dislikes.
Can I use Segmentation on my Internet customers?
Any record, which contains address information or telephone numbers,
can be easily appended with Segmentation. Coming soon! An ability
to unobtrusively identify web visitors by Segmentation type! (Contact
CATENATE for more information.)
How can I use Segmentation with Mailing list databases?
All of the major mail house databases are segmentation encoded. Speak
with your sales rep for specific information about the data you seek.
Can I get maps to show where my best customers live?
Yes, visualization via mapping is an essential component and is available
through CATENATE. Colorful maps can show where current customers
are and where potential customers can be found in high concentrations.
Maps are helpful for media selection, retail location applications
as well as competitive analysis.
Single assignment -vs - distribution, what do they mean?
`Single assignment': a unit of geography is given ONE (single) cluster
classification. For example, a zip code (a unit of postal geography)
might be assigned cluster D20 that represents the preponderant neighborhood
type in that geography. The zip code may be evaluated using smaller
units of (census) geography such as block groups. In this scenario,
a distribution of clusters within the zip code emerges:
| Geography |
Cluster Assignment |
|
|
| Zip Code |
D20 |
100% |
Single Assignment |
Block Group 1
2
3
4
5
6
7
8 |
A06
A08
C16
D20
D21
D22
F34
J31 |
18.8%
1.7%
8.3%
31.3%
6.8%
23.5%
5.3%
4.2% |
Distribution |
What's a profile?
The "standard" report and the basic "directional or
analysis report" is basically a report which summarizes your customer
database by clusters and compares that composition and penetration
to that of a "base" - the base being the US as a whole or
the market(s) where the majority of your customers are to be drawn
from.
What is the purpose of creating a profile?
To show the "lift" and potential (preference) of a cluster
or group of clusters for clear and effective targeting and scripting.
What is "lift"?
The "lift" is the level of discrimination a particular cluster(s)
shows in terms of potential usage for your product or service. Cluster
analysis is performed to identify those customer types who have a higher
propensity for your product or service than other types.
What does MPI stand for?
"
MPI" is the acronym for Market Potential Index.
What is the MPI calculation used for?
The Market Potential Index calculation is used to discover the high
to low concentrations of best performing clusters in local markets
or, in other words a specific markets propensity to use a certain
product or service. Therefore, once you have established a Cluster
profile (high performing clusters) you can then determine where these
high performers are located geographically. For example, if your
profile shows ten of the mosaic clusters heavy users of your product
- the next step is to understand where you can find more high performing
areas. This is when the MPI is essential - to find the "gold" hidden
in the diverse geographic market haystack. Most MPI analysis is achieved
through tabular reports - ranking high performing areas by ZIP code,
census tract or block group. Another way is visually, though data
mapping. This allows you to deploy your efforts (media, mail etc.)
right to the target audience (carving out unwanted areas that might
contain unwanted natural or manmade developed areas).
How is MPI calculated?
The MPI calculation - an advanced mathematical formula that associates
the average indices of a cluster profile to a (high to low) distribution
of small area cluster neighborhoods. In the calculation (which is
extensive), each geographic unit (ZIP code, Census Tract, Block Group)
is individually matched against the high and low overall performing
clusters. The algorithm continues - calculating and matching and
ultimately finished by ranking and categorizing the market for pinpoint
precision targeting.
How can I use this information for media planning/buying?
Segmentation gives you the power and ability to match your consumer
segment with their media preferences - reading, viewing and listening
_ by linking with Simmons Market Research Bureau, Mediamark Research
Inc. as well as a variety other data sources. With a Segmentation
license, you may code any custom survey you have commissioned to "link" the
two databases. (Please contact CATENATE to assist with having your
favorite syndicated sources coded.)
What does it mean to "link" different databases?
"
Linking" disparate sources of data is one of the primary advantages
of using a commercial segmentation system like MOSAIC or Community
Tapestry. Why? When you code your database and sum your customers by
clusters (creating a profile), you see a particular pattern evolve
- some clusters show a greater propensity to utilize your products
or services and "pop" or "lift" more than others.
Once you know which clusters are best for your business, you can analyze
information on other databases such customer surveys, internet respondents,
or prospect lists by cluster encoding those databases and looking for
similar cluster patterns. For example, you might do very well with
clusters A02, B09, and C19. Using syndicated sources such as Simmons
or MRI, you can learn more about what people who are also typed by
those clusters read, purchase, and enjoy. You are not physically "linking" the
databases (like a relational database), however, you are "linking" the
information through association. Remember that segmentation systems
are designed around the concept of "birds of a feather flock together".
That means that people on your file typed by cluster "X" have
a great propensity to act like other people typed "X" on
other databases.
Can Segmentation be used for site location?
Yes, segmentation databases are designed with site study in mind -
and as a geo-demographic tool, is well suited for both new site placement
and realignment of current locations (sales offices etc.)
I already have a lot of transactional data on my database, why do
I need Segmentation?
"
You can never have enough information". The addition of segmentation
to your data armory gives you the current state-of-the-art weapon in
direct marketing analysis and targeting. Additionally, the strength
of geo-demographic clustering allows you link YOUR database to other
databases as well as geography - transactional data doesn't let you
do that
Can I link the cluster deployment to profitability?
Knowing what areas bring you the highest revenue helps to refine your
search geographically. Accordingly, you increase profit via reduced
direct marketing costs plus target hitting acquisition of "best" customers.
Can Segmentation be used for business-to- business applications?
Segmentation is a system built for businesses that need to target consumers
more effectively. It scores neighborhoods, not businesses. It is
not intended for b-to-b usage. If you require a b-to-b solution,
please contact CATENATE directly.
How cost effective is Mosaic or Tapestry in comparison with other
systems?
MOSAIC or Community Tapestry are among the most robust, universally
applicable, easily executable systems on the market. They are competitively
priced, often 25-35% less than other quality systems on the market.
How often is MOSAIC or Community Tapestry updated?
Mosaic is updated by August of each year. Licenses include updates
on one year terms.
|