The questions are bespoke to PredictiveHire and have been designed to measure attributes known to predict whether someone will perform in a role and stay longer in a role. Developed based on research covering 316 articles across over 40 years of research into what does and doesn’t predict employee performance and turnover. Out of more than 50 factors studied over that time, we have selected at least 27 that are suitable for applicant assessment. Examples of these predictors include personality traits, employment history, motivation, attitude, organisational fit and coping styles.

Actual performance data is used to predict the likely performance of a candidate, rather than the other way around.

Traditional Psychological assessments do not link to actual performance in the role, nor do they have any self-learning functionality. There is no performance data that feeds into psychological assessments and therefore they have limited predictive power.

In contrast, with true predictive analytics, between the questions and the applicant is a data model that differentiates actual performance. It is this statistical model which predicts an applicant’s success in the role drawing on many objective data points. Because data enables an efficient and immediate feedback loop about the actual performance of the hired candidate to the model, this improves the accuracy of the predictive model over time.

Very quickly the predictive model that a client organisation is using to select high performers becomes completely customised for their own, continued use. The customer is building its own bespoke Intellectual Property which becomes more valuable to that customer with use.

Input and training data is used to build predictive models.

The input data we use are the responses to questions asked of every applicant. Our questions are developed by our organisational psychologists who test, identify and remove any questions that could adversely impact the EEOC protected categories. We do not parse CVs nor do we capture any information about race, age or gender, or educational or professional background. This is why question based assessments like ours (as distinct from face and voice recognition via video interviews) remain the most rigorously tested for bias and validity in applicant assessment.

Our training data comes from our customers and is objective performance data, such as turnover, sales metrics and other kinds of job family specific performance data that is used to measure individual performance on a regular and consistent basis. We screen this training data and model outcomes for bias using statistical methods (i.e. 4⁄5 rule, Chi Square and the Fisher's exact test), both before deployment and every time we re-train our models.

Your organisation will benefit from a model that is agnostic of any personal information, instead focusing purely on behavioural attributes which correlate to success in the role. Your predictions will therefore be blind to the applicant.

The responses to our questions are de-identified and we do not share them with third parties. No data from your organisation’s employees or applicants is used, analysed or shared outside of your organisation.

You can get started today! PredictiveHire's out-of-the-box predictive models are perfect for sales recruitment, contact centre recruitment and customer facing roles - and can be used immediately for any of these types of roles. Should you want a customised predictive model for your organisation's specialist roles, we can work with you to create these (expect at least 4 weeks). A good example are the customised models we have built to recruit Nurses and Carers for the Health sector.

Our machine learning models are searching for the profile of people who will perform the best.

This is usually not one ‘type’ but a mix of personality types that align with high performance because the dataset we are drawing on to define the profile of the best is a representative sample of employees (typically at least 200 or more) with varying performance levels.

As long as that employee base shows a diversity of personality profiles so will the output of the machine learning. In this way, the machine learning models we build are not filtering out a single personality profile type, avoiding self -reinforcing hiring when you rely on either traditional psychometric testing and human based shortlisting