From Data retrieval to Benchmark: how we got Consumer Trust Index

Consumer Trust Index is the medium, the instrument that companies can use to understand the market and to significantly reduce their losses. It is a day-by-day indicator of Italian consumers’ trust, emotions and propensity to spend, as we have already seen in a previous article. CTI is a way out to overcome the Coronavirus emergency, together.

CTI is powered by PwC and KPI6. A collaboration born not only thanks to the high value that both give to data, but to the faith that we have in the entrepreneurial skills and the expertise of Italian companies and entrepreneurs. People, and companies, that we want to help recover faster.

Do you want to know more about Consumers Trust Index?

Book a Demo with KPI6 and PwC to get a hand-on demonstration of CTI

In the computation of this index, we cannot avoid having some Covid-19 Specific Indicators, to monitor people’s opinions and reactions about the Coronavirus outbreak.

This custom application of our technologies is built to intercept upcoming trends, monitor the most recurring words and hashtags in COVID-19-related posts,  and to foresee the end of the crisis, with the “unsolicited opinions” we got by listening to social network data.

CTI isn’t a survey-based index but is an omnichannel index that gives answers to specific needs and focal points:

It can be read as a global overview and it can be declined by industry. An index used vertically, taking into account any kind of industry and brand.

Global CTI: a nation-by-nation index that takes into account the general perception of consumer trust in the whole economy.

Industry CTI shows valuable insights like performances, top trends, and personas. It can be applied to specific market segments, such as FSI, Fashion or Food & Beverage, to track the daily trust level among users and fans when they refer to the Manufacturing famous brands.

The Consumer Trust Index is generated from a dataset built with public data coming from the web, social networks, and similar sources. 

We set-up the research environment handling a sample of Italian and geolocated conversations.

Geolocation is a very important feature because it allows us to ensure that the user is in Italy when he publishes the post; We also rely on other unsolicited real-time data, instead of telephone or web interviews (CATI, CAWI) that need months to be processed, more expensive and less updated.

The entire dataset is weighted with Artificial Intelligence algorithms. We have run two classification techniques: Sentiment Analysis and Emotional Spectrum Extraction.

Starting from the research environment and the conversations, we have extrapolated the sentiment trend. 

Sentiment Analysis infers the general feelings and emotions expressed by the customer. It doesn’t narrow in on specific details or emotions that are experienced. Instead, it uses the contrast of positive and negative experiences to determine a broad response.

A sentiment analysis conducted on Sentiment140 kaggle market benchmark, unlike other Lexicon-based sentiment analysis.

“Lexicon approaches assign words a pre-calculated weight that indicates the sentiment of the word and calculates the average to obtain the sentiment of the entire sentence. This approach can present problems in case of denial especially if it is far from the interested word.

For example, it is not true that football is fantastic”: you should be able to deny fantastic, even if the negation is put several words before. Furthermore, with this approach, it is impossible to understand the ironic sentences.”  – Andrea Salvoni, Machine Learning Engineer @KPI6

A neural network approach, used in CTI,  is able to overcome these problems because it works on a more abstract semantic level.

Emotional Analysis uses a complex system to understand consumer responses. While Sentiment statistics monitor simplified positive or negative markers, the latter deeply inquiry a full range of human emotions and sensitivities. This method measures the differences in the feelings that various viewers or buyers express -using emoji and text analysis.

The Emotional Spectrum is made up of two components:  Positive Emotions (admiration and joy ) and  Negative Emotions (disapproval, anger, sadness, fear)  calculated starting from the research environment described above. 

Thanks to Digital Consumer Intelligence technology, it is possible to classify these different emotions (from joy to admiration, from fear to sadness), to have different audience segments, profiling strategically and having a full vision of the potential target.

We have calculated, with an inversely proportional reading, the percentage of COVID -19 conversations upon the total of the sample.

Then, we weigh and aggregate them with other insights that compound the CTI. The aggregation itself is performed by Artificial Intelligence to predict the best weights to match our initial measurements.

The index has already been validated by running a sound benchmarking with real-world data, finding a correlation with respect to the other indices that measure, monthly, consumer confidence. 

CTI displays a daily variation, forecasting other indexes in the market.

The index was built on the assumption that there is a 0 moment, identified on the 2nd of January, to which we have attributed a starting value of 100.

The components of the index determine an absolute numerical value.

In the first image, we can see the last recorded value and the weekly average, with the relative changes compared to the previous intervals, the gender donut, and the age distribution, compared to the entire period.

In the second image, we can see the time trend of the index value, with 3 different zooms (selectable via the tab at the top) per week, month or over the entire period.

It is also possible to understand the maximum and minimum values recorded in the selected period. At the right of the image we can a percentage stacked chart showing the emotional spectrum on a daily basis.

Consumer Trust Index is born from a common vision, shared by PwC and KPI6. We believe that Italian brands will be able to overcome the crisis thanks to their expertise and entrepreneurial ability, combined with their empathetic understanding of their customers. We offer them the power of data to help them recover faster.

Our goal is to give our contribution to the Italian economy,  providing the highest possible value to companies, giving a powerful ally that will stay useful even after the current crisis will break up.

Our Index will provide timely answers and suggest the best strategies to get wonderful results, minimize losses and drive a brilliant recovery in the next months. Together with us.

Do you want to know more about Consumer Trust Index?

Book a Demo with KPI6 and PwC to get a hand-on demonstration of CTI