Cities in Europe mainly rely on accident databases to estimate the level of safety at traffic light controlled intersections. Cities have already contributed significantly to improve road safety over the past years. But what if we could predict if intersections are unsafe or if realized measures have the predicted effect, based on actual data of the traffic light? The data driven risk approach was implemented in a 3rd generation safety indicator. This generation aims at combining infrastructure and behavior information. With uniformed real time logging from the traffic light controller like state of signal head and occupation of loop detectors several questions are addressed in an algorithm, like: 1. Has there been a red light violation by a vehicle? 2. Were there conflicting vehicles (conflicting traffic) on other lanes that could have been impacted? 3. If so, what would the severity be? By combining “likelihood” (question 1 & 2) and “severity” (3) it is possible to assess the risk road users are exposed to and so it is easy to compare the effects on safety for any changes brought. Additional research was conducted to see if changes like extension of green light could have a positive effect on safety. For one examined intersection the improvements brought led to a decrease of the risk of 15% and had no significant impact on traffic flow. These first positive tests could be further developed and upscaled through additional collaboration between experts of road safety and traffic management to improve road safety.