Scott McDonald is optimistic about cross-platform
measurement. “I see solid signs of progress,” he stated, with “better use of
projectable samples and panels for calibration, better integration of
statistical adjustments to correct for known biases in behavioral data streams (that
previously were venerated simply because they were large) and better automated content
recognition.” He is also fairly bullish about attribution measurement, though he admits that there still needs to be better inputs and a more expansive ability to calculate across all media and marketing.
I sat down with him to find out what we might expect to see
and hear at the next annual ARF AUDIENCExSCIENCE event scheduled for April
15-16, 2019.
Progress in Cross Platform Measurement …
The overall
theme of the conference will be the vectors of progress in cross-platform
audience measurement. According to McDonald, the industry is “making progress
in coverage, in integrating audience data from different streams and in
automating content recognition.” But, he
averred, “We still need to build better consensus on the core metrics to use
for reporting and trading.”
The
obstacles seem to be more political and economic than technical, he noted while
adding that “there is compelling evidence from the UK around how it is a costly
mistake to under-invest in upper-funnel brand support and over-invest in
lower-funnel activation but,” he added, “it is rarely fatal if recognized and
remedied.”
There have been milestones in these areas since last year’s conference.
“The MRC’s proposed duration-weighted standard for cross-platform video
measurement represents an important milestone in the long-running effort to get
consensus on the underlying metric,” stated McDonald. “That standard will be
hotly debated at this next AUDIENCExSCIENCE, but I consider it a significant
step forward to be having that debate now about a very specific proposal.”
Progress in Attribution …
McDonald
sees some signs of progress in integrating digital attribution with market mix
modeling approaches, but, “there still a need for better inputs (ad exposure
data) and greater transparency about underlying models.”
He believes that, “when properly executed, attribution
efforts premised upon true, randomized control tests have the strongest causal
claims.” However, attribution models are still operating under very limited spheres
that don’t apply across all elements of the media and marketing spectrum, “so
they don’t have as much practical utility for many marketers.”
Plus, they are often siloed, only working within one
complete end-to-end platform, resulting in limited scope and of potential
conflict of interest, he noted. “So when we go beyond these methodologically
pure randomized control panel (RCT) exercises, the most meaningful attribution
studies are those that reflect a bit of modesty and humility about their own
limitations.”
Time Table for Measurement?
I asked
McDonald, how soon did he think it will take for us to get to an industry
standard cross platform measurement? “To some degree, it will always be a work
in progress,” he admitted, “So I don’t think that we will ever be at a place
where we can just go on auto-pilot.” But, looking forward, he saw that, “5G will
have all kinds of knock-on effects on how we consume media and advertising, and
the whole field is so ripe with innovation that the goalpost will constantly be
moving.”
He believes
that there has been a lot of progress so far through the efforts of individual
companies and of industry organizations. “To some extent we are on the
threshold of a new era if the MRC standard survives the inevitable scrutiny and
goes on to win the support of marketers,” he stated.
In terms of
actual metrics, McDonald ranks the following as the most important:
Ø
Exposure metrics (impressions, reach,
frequency),
Ø
Attention metrics (duration, cognitive
measures),
Ø
Affiliation metrics (often reflected in social
media), and
Ø
Market metrics (sales, pricing power, brand
equity).
“Looking ahead to the next five years where do you see the
media ecosystem in terms of measurement, data, analytics, privacy and
attribution?” I asked. His response was swift. “Five years is an eternity in
this business. We will be farther along
– powered more by AI and 5G – but still with plenty of work to do solving the
problems that will be frustrating us in 2023.”
No comments:
Post a Comment