have SoV larger than their SoM. He formalized this relationship
as the advertising intensiveness curve (AI). Jones championed
the AI relationship as a practical tool to help marketers
set their advertising budgets, noting that the relationship suggests
a maintenance level of spend; brands tend to spend
around this level and tend to be stationary. Jones’s (1990)
research confirmed an advertising economy of scale, which
had previously been acknowledged (Aaker and Carman 1982;
Arndt and Simon 1983; Comanor and Wilson 1969). Research
into advertising elasticities consistently showed that elasticities
tend to be smaller for larger, established brands and vice
versa (Assmus, Farley, and Lehmann 1984; Sethuraman,
Tellis, and Briesch 2011; Tellis 2009), which suggests a diseconomy
for larger brands. Subsequent investigations
extended Jones’s AI research and documented consistent, AIlike,
asymmetric relationships between SoV and SoM in a
number of conditions, for example, Buck (2001) using a U.K.
longitudinal survey of premium brands; Hansen and Christensen
(2005) using 1980s’ AdLab single-source U.K. household
panel data; Binet and Field (2013, 2007) using IPA Effectiveness
Awards data; and Field (2009) using aggregate data from
30 Consumer Packaged Goods (CPG) categories. These extensions
confirmed the widespread empirical regularity of AI-like
relationships observed at industry and campaign level within
manifold conditions of TV only, aggregated multimedia,
across countries and diverse data. An explanation of this pattern
is that it reflects many natural experiments, where larger
brands have found that they can spend proportionately less
without losing market share. A competing explanation is that
larger brands spend proportionately less because their elasticities
are lower and they encounter diminishing returns from
expenditures above these levels.
In 2005 Hansen and Christensen started documenting
changes in SoV and SoM over time. In 29 categories with
more than one year of data, about half showed the expected
association between changes in SoV and subsequent changes
in SoM. They identified conditions that refine the AI relationship
and noted vast differences in the pattern across categories,
related to the intensity and concentration of advertising. However,
their data were restricted to household diary recall data
for TV viewing and purchase from 1985 to 1990 when the
media environment was dominated by television. It is important
to test applicability in today’s dynamic media environment
and in different markets where different media dominate
with data widely used in current practice.
This phenomenon was explored further, when Binet and
Field (2007) documented an “equilibrium SoV” showing an
AI-like asymmetric relationship of SoV with brand size. They
demonstrated the metric that determines the level of a brand’s
market share growth is its extra (or excess) share of voice
(ESOV), defined as SoV minus SoM. Field (2009) extended
the analysis across 123 CPG brands. The relationship between
ESOV and growth was quantified: The “average” relationship
that a CPG brand could expect is value market share growth