Video Telematics: Key Workflows

As a video telematics platform evolves, there is a tendency to obsess about adding new features that will, purportedly, add more value to the end-user.

Illustration about key workflows

As a video telematics platform evolves, there is a tendency to obsess about adding new features that will, purportedly, add more value to the end-user. These tendencies are exacerbated in the current moment, the golden age of AI, where sprinkling a little bit of AI fairy dust to anything is supposed to be indisputably for the better. Feature addition that is not in direct service to an end user’s everyday workflows, though, is just feature creep. Focusing on what to improve, enhance or augment then is a direct consequence of the key workflows that matter in day-to-day fleet operation.

As it relates to the use of video telematics by fleets, workflows can be broken down into 3 main categories:


The fundamental requirement of any video telematics solution is simple – to display video concurrent to an event of interest for review. Events of interest generally fall in one of the following categories:

  • An on-demand request by a fleet admin, based on inputs external to the system (e.g. reported accidents, driver highlighting the attention to something, complaints from other drivers on the road), also known as the DVR request. In such cases, the workflow needs to be optimized for fast and easy retrieval of video from the device. Approximate time and/or location have been the traditional methods for video retrieval. More recently, methods like requesting a compressed time-lapse version of longer duration of video for faster review, have helped augment the traditional DVR workflow. The fetch request is sent to the device and fulfilled immediately in case the device is online. If not, the request is queued for fulfillment at the next instance it is powered on and connected. To enable immediate fulfillment of DVR requests, many camera solutions have a low power mode that they go into when the vehicle is not in operation.
  • Video corresponding to events triggered by a telematics black-box. There are two different implementations of this. The first, where the black-box communicates to the camera locally over a wired connection, WiFi or Bluetooth, in real-time, and triggers video capture and upload instantly. In another instance, typical where the black-box and camera system have no viable communication channel between them and rely on their independent 3G/LTE network connectivity, the request has to undergo ’round-tripping’. That is, the black-box creates an event, which is uploaded to the backend, which then triggers a request back down to the camera for video for the requisite duration, and which is finally uploaded by the camera back to the cloud. It is apparent that the latter is inefficient with regards to both data usage and time required for video to be available, and most modern video telematics systems aim for the former.
  • Events generated through analysis of video and other sensors on the device, in real-time, using AI. These include ADAS-based events like tailgating and rolling STOPs, DMS-based events like driver distraction or fatigue, and detection of impacts or crashes. For such AI-based events that are generated automatically, the main bottleneck in the review process is the fleet manager’s time and attention, which are both limited. Events on the other hand, even from a medium-sized fleet, can run into the hundreds in just a week’s time. Providing tools to sort and display only the most critical events for review becomes critical. A ‘lazy loading’ UX paradigm, where more events are surfaced only on request, is one of the ways in which fleet managers can be given better control over the review process.

Measurement and reporting

Video telematics systems, especially ones with auto-curation of events, can generate a mountain of data in very short order. Distilling that mountain of data into key metrics that measure and report how drivers and fleets are performing over time is both challenging and much-needed. To paraphrase the title of a famous book, if it is not measured, it does not matter.

Typically, they include:

  • Gross metrics like driver scorecards as an aggregate of all events weighted in accordance with the importance of any particular event-category. They help in:
  • Measuring how a driver is performing over time, with the ability to export the same in driver reports
  • Validating how drivers are comparing against each other, on something similar to a leaderboard. This forms the basis for reward and coaching programs. Top drivers could have performance-linked incentives, and drivers in need of improvement get shortlisted for coaching.
  • More granular metrics to include in driver reports, as part of a subsequent coaching process, like:
  • The most frequently occurring violation categories, with associated meta-data. E.g., if speeding is the most common incident, how much above is the driver on average? This can set the basis for weekly or monthly improvement targets.
  • Event correlations with respect to trip duration, time of day, weather, etc.
  • If it includes AI-based real-time driver feedback, metrics around the effectiveness of the same. As an example, after a driver has heard a warning for a posted speed violation, how often has she slowed down?

Coaching and feedback

Coaching, and feedback that both drivers and fleet managers can initiate, form the critical final component that makes a video telematics system deliver tangible value to the fleet. Common workflows that ensure that are:

  • On the basis of metric-driven driver reports, a fleet manager initiates a review and coaching session. This can be in the form of a one-on-one session, but a more scalable solution includes a formal review checklist that a driver has to complete. The checklist includes sample events and videos that the fleet manager has shortlisted, along with all associated meta-data and comments/notes that a fleet manager has annotated events with.
  • The driver gets notified of a pending review, the next time they log in to the system – usually through a driver app provided as part of the overall solution. Completion of a pending review checklist can be made an essential part of a driver’s to-do list before starting the next trip. In instances where a driver wants to dispute or provide clarifying comments on specific events or overall scores, the review process should also provide for that.
  • Once the review checklist is completed and submitted, fleet managers get an intimation of the same on the portal, where they can formally close the coaching instance.

It is only when they are placed within the context of these key workflows, that feature enhancements and optimizations lead to the best RoI for fleets. It ensures that product innovations are making the job of the fleet manager easier, while still delivering tangible improvements to overall fleet safety. The opportunity exists to improve and enhance each of the workflows described above, and that acts as the North Star for the evolution of our own platform, RideView.