Case study

AI Dashcam: Product-Market Fit

Back to Case Studies

Outcome

29%

increase in pilot wins

$113k

increase to ARR (new units)

4,360

units added to sales pipeline

Overview

This case study centers around an AI-powered dashcam product with a web app for fleet owners, a mobile app for professional drivers, and internal thresholds that use machine vision to detect unsafe driving incidents.

Soon after joining the product team, I identified the root cause of a product-market misalignment through user research, identified accident exoneration as the most profitable target for roadmap realignment, and re-designed our most prominent accident exoneration feature to meet user needs.

Roles
  • Lead UX Designer and Researcher
  • Director of Product
    (Interim Product Manager)
  • VP of Product
  • Chief Product Officer
  • Technical Program Manager
  • Software Development Manager
  • 2 International Dev Teams (15 Developers)
Skills
  • Research Planning/
    Project Management
  • User Interviews
  • User Surveys
  • Competitive Analysis
  • Academic Research
  • Metrics Analysis
  • Visual Design
  • Collaborating with International Vendors and Developers
Tools
  • Figma
  • SurveyMonkey
  • Dovetail
  • Respondent.io
  • LucidChart
  • ProQuest
  • EBSCO

Problem Statement

The majority of customer pilots began failing almost immediately after installing trial cameras, with three issues emerging in feedback to Sales representatives.

  • No time for daily coaching: Customers claimed they did not have the time or personnel to coach their drivers. This would require continuously watching the web portal and engaging with drivers on unsafe driving behaviors.
  • Driver and union pushback: Customers were concerned about a coaching product causing employee turnover in the midst of a driver shortage, citing pushback from drivers and especially driver unions.
  • Accident exoneration: Customers were surprised that the product was not primarily centered on "accident exoneration", but rather on coaching unsafe driving behaviors to prevent future accidents. (It had several “emergency response” features, but this was considered an edge case/secondary feature.)

Proposed Solution: Re-Evaluate the Product-Market Fit

Before discussing the problem with my product manager, I took a few hours to better clarify the boundaries of the problem, check in with people at the source, and figure out if competitors were also seeing the same trend. Three internal interviews and a quick one-hour competitive analysis indicated significant changes to the market: "accident exoneration" was being proactively surfaced by potential customers and it was all over our competitors' landing pages.

This warranted a re-evaluation of the market landscape and how well our product met the needs of our target customer base.

Research Questions

Accident Exoneration
  • How big is the gap between our product's feature set and customer expectations?
  • Why is this gap only resulting in failed customer pilots now, when the product had been released for years?
  • Is there greenspace in this area to differentiate ourselves from competitors?
Market Segments
  • Is our target market segment shrinking?
  • Which segment is bigger: the one targeting accident exoneration, or the one targeting driver coaching? Are they mutually exclusive?
  • Are there other market segments that we are not aware of?
Driver Influence
  • Why are customers just now starting to reference driver unions and turnover when the driver shortage has been ongoing for three years?
  • Do all drivers and driver unions oppose dashcams, and if so, why? Can this rift be bridged?

Academic Research

An analysis of 18 industry reports, white-papers, and publications revealed that while dashcams were not directly causing the driver shortage, drivers worked in an increasingly hostile environment that trained them to evaluate whether any new equipment would help or harm them:

  • Drivers are wary of all telematics purchases because of past experiences with employer abuse, fearing that employers were using gathered data to penalize them.
  • The American Transportation Research Institute (ATRI) reported lawsuit abuse and “nuclear verdicts” as one of the top concerns for fleet owners in 2021. 
  • From 2021 to 2022, the driver shortage became so extreme that the transportation industry pivoted sharply back to driver retention strategies:
    • Signing bonuses and higher salaries
    • Safety reward programs
    • Greater protection against nuclear verdicts and coordinated accidents
  • Berg Insights’ 2021 report on the Video Telematics Market notes that fleets often use data and examples of accident exoneration to win over personnel, then get drivers on board with coaching.

Driver Surveys and Interviews

SurveyMonkey: Driver survey results.

  • A statistically significant portion of unionized drivers like or accept their dashcam.
    • 85% said that they personally viewed their camera as a positive (53%) or neutral (32%) tool in their daily work. 
    • 78% said that their driver union also viewed driver cameras as positive or neutral.
  • Driver interviews revealed a journey of mistrust to appreciation, with exoneration being the tipping point.
    • 5 out of 6 interviewees experienced violence or threats while on the job, such as aggression from other drivers, carjacking, attempted theft, physical assault from passengers, or threats of false complaints if the driver doesn’t break policy.
    • 5 out of 6 drivers said that they personally viewed their camera as having negative value when it was first installed.
    • After the first exoneration, all 5 drivers said they see the value of dashcams and prefer to have them.

Buyer Surveys and Interviews

Conducting user research with dashcam buyers reinforced driver findings, while adding extra context and insight into accident exoneration from the employer’s perspective.

  • In a survey of dashcam buyers, the most common reasons for purchasing cameras were "changing driver behavior" (42%), followed closely by "recognizing unsafe driving trends" (36%) and "accident exoneration" (33%).
  • 6 out of 6 buyers reported initial pushback from drivers and unions, but:
    • This was solved by working with drivers to answer questions and evaluate the dashcams together.
    • Pushback stopped after the first accident exonerations occurred.
  • 6 out of 6 buyers described the same evolution of their fleet’s dashcam usage:
    • After multiple accident exonerations, they started to see patterns in unsafe behaviors or precursors (i.e. route locations) leading up to accidents.
    • They begin coaching drivers on unsafe behaviors, setting safety performance metrics to see if reducing a behavior or changing elements of the driver’s environment reduces accidents over time.

The buyer research revealed a key takeaway: Professional fleets’ usage of in-cab cameras exist in a pipeline from accident exoneration to driver coaching, and we had unknowingly been targeting users at the end of that pipeline, when potential customers were already likely to have safety programs and KPIs coupled with a competing dashcam provider.

Expanding the user personas.

Competitive Analysis

A 3-day analysis of product documentation and hardware specifications reveals significant investments into driver exoneration, accident response, and driver-centric features.

Competitive analysis in FigJam, showing screenshots and callouts from 7 major competitors.
  • Direct Video Request (DVR) workflows that send videos from the camera’s local storage to the fleet’s online portal by entering in details of the accident (trip, date/time, vehicle).
  • “Panic buttons” that send text and email alerts and automatically upload videos from before and after the button press to the fleet’s online portal.
  • Camera systems with 360-degree auxiliary cameras and improved video storage to prevent older videos from being overwritten
  • Driver reward systems for avoiding unsafe behaviors and accidents
  • Driver face blurring, lens caps, and scheduled camera shutdowns to preserve long-haul truck drivers’ privacy in off hours.
Key Takeaways
  • The entire market is already pivoting to serve fleet owners’ goals of hiring and retaining drivers by addressing driver concerns.
  • Accident exoneration features involve both hardware and software components.
  • There is greenspace in the market through improved DVR workflows, camera storage, and driver rewards/privacy.

Roadmap Changes

As a result of the preceding research, I recommended that my product manager change the focus of our next year’s roadmap from improved coaching features to accident exoneration features, divided into three areas of investment:

  1. Driver-Centric Features: panic buttons, privacy features, safe driving and performance KPI tracking for easier driver rewards, and case studies and sales materials geared toward common driver and union concerns
  2. Buyer-Centric Features: Month over month fleet metrics to identify potential accident causes, camera health reports to ensure the fleet is always recording, and support materials for navigating driver buy-in
  3. Accident Exoneration Features: Improved Direct Video Request (DVR) workflow, collision alerts, improved camera storage, multiple camera integration

The user research and proposal were approved by my product manager, our VP of Product, and our Chief Product Officer. Our highest priority would be the improved DVR workflow, followed by camera health reports, panic buttons, and collision alerts.

Improved DVR: First Iteration

Figma: The modal used to request video from the camera's local storage.

To impact our potential and current customers as quickly as possible, I prioritized quick changes that could be accomplished within a sprint, but would make our DVR workflow easier to find and follow users’ mental model during an accident response.

  • Moved the DVR feature out of a menu and made it a Call-To-Action button
  • Added form fields matching accident details
  • Added time estimate to download video
  • Multiple ways to request video (form or slider)

Improved DVR: Second Iteration

Figma: The new DVR report with further improvements to the video request modal.

Next, I tackled the DVR feature’s place in the Information Architecture. Instead of being buried three levels deep in the site structure, I created a new primary navigation item that led to a report holding all DVRs. This laid the foundation for a central place to hold videos from the upcoming “panic button” feature and included the improvements:

  • Contextual links to evidence around an accident, such as the trip data, driver history, and vehicle history
  • Alerts when video download is complete, so a user can continue gathering accident evidence while a video download is in progress.
  • Video previews and timelapses to help users find the accident scene in a trip before committing to a download

Contextual links and download alerts were market differentiators, and very few competitors offered video previews or timelapses.

Lessons Learned

Map Assumptions and Unknowns

If inheriting an existing project, document all working assumptions that do not have direct support from existing user research. Work with my product manager to prioritize these and find open spots in the delivery schedule to address them before they become a problem. 

Structured Usability Testing

My biggest regret with this project is not walking away with an official System Usability Score (SUS), heatmap, time on task, and error rate for the Improved DVR feature. Under normal circumstances, I would not consider a design complete until it has an above average or higher SUS.

  • One available alternative was to pay for an unmoderated usability testing service that ran in the background and recruited from the general public.
  • A generic but structured infusion of public feedback would have been a good supplement to the unstructured but targeted customer feedback, since a good design should be intuitive to anyone regardless of career background.
  • I decided not to pursue this in order to conserve user recruitment budget for our other UX Researchers, since I was already responsible for a majority of our spend-down. However, we ended the year with room left in the budget and I would have outsourced the user testing if I could go back and re-make this choice.

Next Steps (Complete)

  • Prioritize all features in roadmap recommendations with Product Manager
  • Estimate work with Software Development Manager and Technical Program Manager, then add UX Design + Development estimates to product delivery timeline
  • Circulate research results with Sales and Marketing to facilitate creation of customer-facing materials
  • UX Design for next highest roadmap priorities: Camera Health Report, Collision Event, Panic Button