Typically 80% of the time in market research is spent on time-consuming
tasks such as sampling, data acquisition and analysis, leaving only 20% for
decisive detailed questions. By means of innovative big data and AI processes, this process can be automated so that market researchers have more
time for really value-adding activities such as the interpretation of the analysis results and to derive recommendations and actions. Tomorrow’s market
research will be oriented less towards samples and interviews but rather pursue a real-time census approach with automated analysis.
By its very nature, market research is an extremely data-driven industry.
Market researchers have always collected, edited and analysed particular
data and then dealt with the interpretation of this data. In today’s fast-paced
world, however, we are facing an enormous volume of data, we have already
been juggling with zetta- or even yottabytes for quite some time. The global
data volume is doubling every two years, resulting in a task man cannot cope
with alone. Luckily, state-of-the-art technology not only provides memory space and the adequate computing power to be able to deal with the mass of
data, but also diverse evaluation and analysis possibilities. The latest developments in the area of machine learning allow making smart data from big
data and using data really economically.
Successful market research has to adapt accordingly and integrate these
innovations in its work if it does not want to be left behind. For example,
there is already software which automatically converts the answers of subjects from studies (CAWI, CATI and CAPI) into codes whilst not only considering the respective main statements but also extracting and semantically
linking all the other information. The significance is increased by a multiple
thereof. Far-reaching interpretation then follows hereby the code plans reach
a new level of detail difficult to achieve with manual processes.
But actually it is not about choosing either man or machine. AI systems are an intelligence amplifier. Poorly drawn up, poorly maintained and
poorly interpreted, they only produce costs, trouble and nonsense. Wellprogrammed, capable of learning and used intelligently, artificial intelligence
can save a lot of work and create time for depth of detail. When it comes to
decision logic, for example, artificial systems are always more complex and by
far more precise. And that is exactly why predictive analytics—i.e. the prediction of customer losses, of sales figures or price acceptance—is so useful. Also
concerning the question “What causes the customer behaviour”, i.e. a causal
analysis, AI systems are considerably better. Because humans can actually
only think in correlations and thus fall into the trap of spurious correlations
on a regular basis, human decision-makers also have to learn something new.
In a first step, market research with artificial intelligence can complement
the classic path, in a second step, however, in part even replace it. The digital index of the state government of Rhineland Palatinate compiled for the
first time in 2015 is one example. ZIRP—Zukunftsinitiative RheinlandPfalz (initiative for the future of Rhineland Palatinate)—had polled 260 of
170,000 companies in the state beforehand, which was not only cumbersome, but also time-consuming and costly. In contrast, software based on
artificial intelligence can provide information on 110,000 companies in an
instant.
Comments
Post a Comment