Retorio is a video analysis technology that fuses machine learning with scientific findings from psychology and organizational research. Even from short videos Retorio reliably detects behavior and important personality perceptions. Retorio helps individuals, HR professionals and executives to challenge their own perceptions of a person and to make principle-based decisions.
HR processes (such as hiring) involve important decisions that are strongly influenced by subconscious factors.
The great importance of gut feelings on the part of decision-makers often leads to biases and unequal treatment.
Overworked HR departments often have little time to screen CVs or to prepare for and follow up on their interviews.
They have to rely on their intuition in their accelerated workday.
Without new technologies, such as Retorio, it is inevitable that many applicants merely can upload a CV and will not get a fair chance to make a personal impression.
While a CV might inform you about pre-experience and a candidates' socioeconomic status, it tells you very little about personality, soft skills or attitude - crucial factors that will determine how well a person fits into an organization.
Pre-employment assessments, such as Retorio, introduce an element of standardization and objectivity into the hiring process by providing concrete results that can be compared across all applicants (or employees).
Employers can then use these data to make better informed, more defensible hiring decisions before proceeding to more resource-demanding procedures like personal interviews.
No; Instead it empowers them. Retorio challenges provides professionals with valuable and objective data that informs their subjective decisions.
With Retorio, professionals challenge their own decision making process and defend their choices.
They often end up making more elaborate decisions (e.g. whom to invite to a personal interview or for a certain training program).
Absolutely; you can invite candidates and view all their results without the need for an additional tool like an Application Tracking System (ATS). However, should you already have an ATS, you can easily integrate Retorio into your existing ATS.
Retorio is used in pre-screening processes, where our clients collect and analyze video applications in order to give every candidate a chance to leave a personal impression.
Retorio is used in internal assessments to derive a better understanding of an organization's culture and to reveal behavioral patterns that drive performance.
Moreover, Retorio is used in training programs to enable customized soft skill trainings and to enable private users and employees to prep-up for job interviews, client talks, leadership roles, etc.
Retorio is used by organizations around the globe from < 20 people growth startups to global enterprises with up to 400.000 employees (you will find some of their logos on our website).
Moreover, we partner with selected HR consultancies, headhunters, and training agencies to serve our global audience. If you are interested in partnering, please contact us through our partner page.
We’re committed to ensuring the security and protection of the personal information that we process.
We are based in Germany and exclusively store data within the European Union.
We provide a GDPR-compliant and consistent approach to our data protection. Find out more about our Ethics and Privacy.
Retorio is a spin-off of Technical University of Munich, Germany. Retorio’s analysis is based on the universally accepted, advanced psychological model of personality traits, the Big 5 model. This model describes an individual’s personality in terms of 5 dimensions: (1) openness, (2) extraversion, (3) conscientiousness, (4) agreeableness, and (5) neuroticism.
In a workplace context observer ratings have an incremental predictive validity over self-reporting. Accordingly, Retorio's AI is trained to analyze the Big 5 according to visible behavior rather than self-estimations. Thus, Retorio's output suggets how a person comes across in terms of personality.
Retorio lies on machine learning methods to analyze behavior in videos. Machine learning denotes statistical procedures that result in prediction models.
The models have neither consciousness nor intuition. They rely on data as experience. That is, an AI is an accumulation of experience (data points).
The experience base is thus crucial to the basis on which an AI later makes assumptions. It is not only the mass of data that is decisive, but also the type of data, the variety of data and the data quality.
Retorio's AI only learns under supervision. Retorio's models have been trained to predict a person's impression based on a representative dataset. The dataset comprises thousands of videos clips of people around the world that have been assessed by observers the according to the Big 5 taxonomy.
In total, we used more than 2,500 observers from five continents. The individuals in the video clips were also divided equally in regards to sex, ethnicity, and age. To promote objectivity, multiple ratings per video were obtained. The overall dataset consists of more than 12,000 people.
Our system has a prediction accuracy of around 92%. Imagine asking thousands of people how they perceive a person on video in terms of personality. We achieve the same results to 92%.
92% accuracy means that we can expect an unsystematic estimation error of ca. 8%. That means, the AI might sometimes be a bit off, for which reason we recommend to merely use it as an assistent system.
The Retorio observer based assessment can be expected to show higher external validity than self-assessment of personality traits. Since personality traits are crucial predictors of job behavior, it is essential taking these into account when making hiring decisions. For more background on the validity of Retorio, click here.
We ensure that Retorio does not systematically discriminate against people based on their ethnicity, age, gender, religion or culture. We do so by by checking the estimates in our datasets for possible systematic biases (e.g., discrimination).
"Algorithm-based differentiations become discriminations in particular if they constitute an unjustified disadvantage of persons characterized by protected characteristics (in particular age, gender, ethnic origin, religion, sexual orientation, or disability)"
(Source: German Federal Anti-Discrimination Agency)
Retorio ensures that only factors that are within the applicant's control are included in the results. For example, in our datasets, we compare mean scores on the Big5 dimension Extraversion between Caucasians and Blacks. When we discover significant mean differences attributable to group membership, we adjust the mean and distribution to compensate for discriminatory bias in the training and test sets.
We regularly test the corrected models against large, scientifically sound datasets, such as UCLA's Fairface dataset, which contains approximately 100,000 individuals from different cultures, age groups, etc. We publish our results transparently. Our research clearly indicates that Retorio assesses applicants regardless of their skin color, gender, or age.
Retorio's AI does not need specific training to provide helpful output. Send out assessment links and receive video applications from day one.
Over time you can add more data about your teams and your culture to increase the precision of Retorio's predictions.
Usually, a Retorio assessment shouldn't take longer than 5-15 minutes, with roughly one minute scheduled for each question and several trials per question.
However, the length of the assessment depends on the number of questions posed by the recruiter.
You will need a recording camera, functioning voice recording and quiet environment, preferably a white the background and sufficient light.
Just like there's no right or wrong personality in a partner or friends -they either fit your character or they don't-, there is no right or wrong personality for a job in general. Most importantly, Retorio will assess a fit between the candidate's personality and job tasks, as well as the company culture and existing team. Performing well therefore means attaining a high fit. But here it gets very individual and candidates should at all costs avoid faking responses, gestures or mimic. It is most important to try and behave naturally, lean back and relax. As a fit that's not based on your actual personality will neither be helpful for candidate nor company.
We make it our objective to assess candidates' personality as objectively as possible. We guarantee high objectivity and reliability by removing single observer-based bias from the data and using multiple observer ratings. We also set value on removing any biases resulting from gender, age, ethnicity or skin color.
The combined rating of the AI makes use of visual input from facial expression and gesture, as well as auditive information like speaking speed, language sentiment and engaging language.
Just like humans, any AI can be tricked. For you to cheat, there would have to be an ideal state the cheating aims at. In personality assessment, there can be multiple ways to reach a goal. In general, we do have a high test-retest reliability. That is, if the same person takes the test in the morning or evening, only a deviation of around 9% occurs. However, if you decided to intentionally act in a different role from your own personality, reliability can be hypothesized to be somewhat lower, just like in a face-to-face interview or on the job itself.
No test procedure is 100% reliable and every human being makes mistakes. A certain error is therefore always to be expected.
When using Retorio, this measurement error is even desired to a certain extent. In both recruiting and training contexts, it is quite desirable that participants try to leave as good an impression as possible.
Accordingly, factors that you can actively influence are sometimes taken into account in the AI's evaluation. For instance, if you apply in a suit instead of an undershirt, this may allow important conclusions to be drawn about your future behavior. The background of your video is also in your control and thus you should avoid backgrounds that might make a bad impression.
For instance Empty beer bottles in the background may leave a different impression than a bookshelf. A video application from a bathtub may create a different impression than a video application in front of a beautiful painting. Accordingly, you should take these factors into account and ensure a neutral background and decent clothes.