Relevance & Research Question: We use the reputation effect on selling performance as an indicator of the effectiveness of reputation systems in C2C online markets. The purpose of reputation systems is to solve trust problems and promote cooperation among buyers and sellers. The larger buyers’ needs for information about sellers’ trustworthiness, the larger will be the correlation between seller reputation and selling performance. Consequently, the more effective a reputation system is at screening untrustworthy sellers, the smaller will be the reputation effect.

Methods & Data: We integrate the results from 96 existing empirical studies with a meta-analytic method to investigate the existence of reputation effects. We compare the correlational effects estimated between three types of seller reputation variables (number of positive and negative ratings, as well as feedback score) and four types of selling performance variables (probability of sale, selling price, selling quantity and the ratio of selling price to reference price). Because of the differences in analysis methods across the 96 studies, we use conversion techniques to calculate comparable effect sizes.

Results: Our meta-analysis confirms the general existence of the reputation effect. Especially the number of positive ratings exhibits a consistent, significantly positive effect on all types of selling performance with correlational effect sizes ranging from 0.06 to 0.27. The feedback score (i.e. the number of positive minus the number of negative ratings) also has a positive effect on selling performance, but is generally smaller and ranges from 0.03 to 0.09. Results regarding the number of negative feedbacks are mixed although the effects are, as expected, generally negative.

Added Value: Our study corroborates the existence of the reputation effect across different operationalizations of seller reputation and selling performance. It also shows however that the range of the effect is considerable, which could indicate that in some C2C markets reputation systems are more effective at screening untrustworthy sellers and thus reduce the information demand on the part of the buyers. In a next step, we will meta-analyze these reputation effects to obtain a better understanding of the conditions under which effect sizes vary.

Relevance & Research Question

For the brands, influencers offer new opportunities in digital marketing. They might be a more intimate, unformal, and therefore more convincing touchpoint with consumers. But before decisions can be made some questions need to be addressed. Is it worth it for a brand to sponsor youtubers? Who is the real audience of youtubers ? What is their “true” influence?

Methods & Data

We made a selection of the TOP 50 most influential German youtubers among various industries (fashion and beauty, food, hi-tech, tv show and movies, gaming).

We have a panel rolling out in Germany of # 2000 nationally representative people who have agreed to share their navigation data (on computer and / or mobile devices) with us. They have installed a tracking software on one or several devices. As such we are able to detect who watched a youtube video from an influencer (roughly 25 % of our sample had seen at least one video from the TOP 50 influencers), and what they did before and after.

Influencers post some videos sponsored by brands, websites. With our methods, we can measure the impact of a sponsored content; to what extent it brings traffic to the sponsor’s website.

We also run a survey among the identified audience of these influencers in order to understand their motivations.

Results

The ROI of influencers strongly depends on the target and the industry. Youtubers have far more impact among young people. Cosmetics influencers are far more influential than food influencers. We have measures which prove that an important share of young people interested in cosmetics tend to use influencers as a reliable source of information and to follow their recommendations. We are also able to explain the logics of influence: What differentiates a good influencer from a “bad”one?

Added Value

For each brand we are able to build an index of performance for each influencer. With this method a brand is able to decide which influencer is the most impactful.

Relevance & Research Question:

In a world of insights that is rapidly changing due to new thinking (especially behavioral science), new data streams and new analytics and an increased time and budget pressure, we have decided to pursue a new path for screening packaging designs.

Methods & Data:

The AI Pack Screening Model is based on human expertise, data mining and artificial intelligence. Combining these three forces we have come to a predictive model of future success of pack designs without conducting surveys.

The "heart" of the Model consists of our own database with more than 25.000 studies on packaging research conducted over the past decades. It is trained by the input of variables that are validated to in-market Sales Performance, relationship with and across key metrics of pack designs and the impact of different market scenarios (e.g re-stage vs. new product, small vs large share of shelf, well established brands vs. new brands, etc.).

These Learnings have been applied to build a model for the screening process of pack designs which has been tested against survey data and which is continuously learning from new cases.

In addition to applying the model based on information by database learning to new designs, Human Experts are needed to provide a holistic perspective and to integrate dimensions/strategies specific to each brand’s situation. Those experts are Senior PRS IN VIVO professionals who will be activated at the beginning of the process.

Results:

The presentation will cover the main principles of the process as well as examples of outputs and will talk about first hand experience of combining human expert evaluation with AI based algorithms.

Added Value:

Thanks to the combination of Human Experts and AI results can be delivered within one week, catering the demand for faster turn around of research results in a rapidly moving insight world.