the PowerPoint deck

advertisement
METHODOLOGY FOR
BORRELL DATA
11/30/2011
Borrell Methodology
www.borrellassociates.com
The Borrell Ad Spend Model

Under continuous development since 1984

NAA Lifetime Achievement Award to chief architect
Kip Cassino, Executive Vice-President

In long-term use by major media companies

Advertiser-centric rather than media-centric

Bottom-up rather than top-down

Based on multiple sources
If the world only had
one market…
If the world had only one market…
…measuring ad
spending would be
easy
If the world only had
one market…
Unfortunately…
…markets are all
over the place.
www.borrellassociates.com
Ad spending generated in a market…
…may go anywhere. Ad
spending from any other
market may end up in the
market you’re trying to
measure.
How we do it…
www.borrellassociates.com
Borrell separates ad spending that’s
coming in from ad spending going out…
So, we can measure:
• Ad spending generated in a market.
• Ad spending directed to a market
from someplace else.
• Ad spending generated and spent
locally.
www.borrellassociates.com
We start with two databases…
#1 - Estimates ad
spending by more
than 20 million
US companies
#2 - Estimates
receipts for all US
Media Companies
www.borrellassociates.com
At the National Level…
=
1=2!
That is, total US ad spending
must equal
total receipts for US
media companies.
www.borrellassociates.com
Database 1 (Spending) Sources…





Dun & Bradstreet (updated quarterly)
IRS
The Postal Service
Annual reports & 10K’s
More than 40 additional secondary
sources (magazines, association reports,
surveys, articles), including –
Editor & Publisher, Advertising Age, etc.
www.borrellassociates.com
We don’t stop here!
Every company’s spending estimate is
modified to fit specific markets in
two ways.
Two Modifications
www.borrellassociates.com
First, every estimate is altered to
fit the media demand pattern
of its specific market,
using data from:
Nielsen, Scarborough,
Demographics Now, Claritas and
other sources.
Second, the estimates are not
expressed as percent of revenue…
but as Per Employee ad expenditures.
Employees are key!
www.borrellassociates.com
In general ad
spending increases
as businesses get
larger…
$
%
But…
Each SIC has its own distinct
pattern. We measure and
account for that.
…the per employee
percentage of ad
spending actually
drops.
www.borrellassociates.com
Database 2 (Receipts) Sources…
 Dun & Bradstreet
 Annual reports & 10K’s
 More than 40 additional secondary
sources (magazines, association reports,
surveys, articles), including –
NAA RAB DMA TVB IAB YPPA
Media Week, OMMA, Advertising Age ,
Our own survey of media company revenues with
4,800+ companies represented.
Comparisons Made
www.borrellassociates.com
Each media estimate is compared to other
sources (including McCann Erickson, Veronis
Suhler, Jupiter, Forrester and others) to insure
accuracy.
When discrepancies are found, that
are not due to methodology or
definitions, the estimates are recomputed.
www.borrellassociates.com
Once the two databases agree…
… the resulting estimates
(by SIC and media choice) are
distributed to all US counties.
Distribution to counties is a
three-step procedure:
www.borrellassociates.com
Once the two databases agree…
Step 1:
Allocation
Step 2:
Replacement
Media estimates are
allocated to each county
using the weighted values
of several variables,
including:
• households
• retail sales
• median income
• gross regional product
• population
• median age
Whenever possible,
allocated estimates are
replaced by actual known
information. Typically, about
25%
of the estimates are
replaced.
Step 3:
Recalculation
After replacement,
the sum of the estimates
will no longer foot to the
original
national totals.
So, all un-replaced
estimates are
indexed and recalculated.
Is it Soup yet?
www.borrellassociates.com
Not yet!
All of our work so far has produced estimates of
ad spending directed to each market. We still
don’t know how much is local.
To complete a market estimate,
we must return to Database 1 –
Ad spending estimates for every
US business.
Think of three Glasses…
www.borrellassociates.com
The first contains the ad spending generated in
the measured market.
Estimates from
Database 1
Estimates from
Database 2
The second contains ad spending
directed to the market.
…poured into one
www.borrellassociates.com
The contents of both must be
“poured”…
…into the third glass – which represents ad
spending in the measured market.
Imports
www.borrellassociates.com
For some media, the contents of the
first two glasses will not fill the
third. More ad spending must be
“imported” to make up the difference.
Exports
www.borrellassociates.com
For some media, the contents of the
first two glasses will more than fill the
third. The ad spending “overflow”
must be “exported” to other markets.
The Formula is Simple
www.borrellassociates.com
Ad Spending generated in a market
(Database #1)
+ “Imports”
- “Exports”
MUST =
Ad Spending directed to a market
How good is the Result?
www.borrellassociates.com
• Used by many of the nation’s largest
Media Companies for years.
• Used by Advo, TV networks, ad
agencies, telecommunications companies…
• Has shown to be accurate to within
3-5 percent In markets where
we can get “on the ground” evidence.
• According to Booz Allen: “Only
methodology that could work.”
What they’re saying….
www.borrellassociates.com

“Accurate enough to be actionable”
− Borrell client

“Solid guidance on scale and direction”
− Another Borrell client
Our Model in Action - Examples

Political Ad Spending
www.borrellassociates.com




Borrell Projected $20 million
Competitors projected $75 million
Actual was $19.5 million
Directories



Projected decline of yellow pages of 37%
Competitors had it at 7%
Later projections pegged it closer to 35%
www.borrellassociates.com
Thanks!
For More Information on Upcoming
Webinars, please visit
http://compass.borrellassociates.com
Download