The automotive industry is undergoing a number of sweeping changes, from falling urban car ownership to recent breakthroughs in autonomous driving and eventually, to the new experience economy that will emerge around our vehicles. As these technologies are further integrated and proliferated, it is becoming ever more important to be able to create, refine and evolve the automotive technology experience at speed, to enable in-car experiences to keep pace with the rest of their drivers’ digital lives.
Photo by Martin Katler on Unsplash
With the recent rise of mirrored infotainment systems like Apple CarPlay and Android Auto, and the imminent introduction of native integrations like Android Automotive OS, customers are expecting an exponentially higher quality of digital experiences within their cars, and satisfaction with technology experiences, in turn, is driving repurchase intent.
A car’s interior is the makeshift dinner table, the conference room for calls, the bedroom for the kids on a long haul journey and the sanctuary when someone just needs to gather their thoughts. Drivers arguably spend more time in their cars than they do in any other space. It is a sacred space of sorts - a sacred space that is increasingly being eyed up by Google, Amazon and Apple. There are rumours that suggest Google and Apple may even be going as far as directly challenging the automotive industry with vehicles of their own!
Whether developing for mobile-based systems or native integrations, tomorrow’s infotainment systems are faced with three primary opportunities:
Google, Amazon and Apple are able to provide products and services that car manufacturers cannot compete with and neither should they aim to - it would be a futile exercise.
While early versions of ‘projected’ mobile solutions haven’t had access to proprietary in-car systems, that has been changing, with Google recently working with early partners such as Volvo and GM to create deeper-integrated native Android Automotive experiences (the infotainment system is built on top of Android and is baked into the car so there is no need for consumers to use their smartphone as the vehicle for Android), and mobile-projected solutions will soon have access to those integrated systems. Much of the initial concern around these systems had been around an inability for manufacturers to embed their own design language- but so far the Silicon Valley players have been willing to cede significant design agency to automakers - Volvo’s integration of Automotive OS looks substantially different to how GM’s will look, and BMW’s integration of Alexa maintains their design language.
If manufacturers opt for relatively vanilla native integrations or purely mobile-mirrored solutions, they risk forgoing an opportunity to innovate and build customer loyalty. For manufacturers who decide to go further with their in-car experience, a lot still remains in their domain despite the existence of plug and play Carplay and Android Auto experiences delivered via smartphones - there is non-trivial scope for creating exceptional and more bespoke experiences for different buyer types, markets and vehicle ranges, both in the design of the underlying system, iteration therein with OTA updates, and extensions such as bespoke apps.
However, to successfully do so, automotive manufacturers would do well to take a cue from the companies that are driving so much of this disruption.
Remember: Google, Amazon and Apple were putting lean startup principles into practice long before it had become a modern tech movement; they continue to extol the virtues of ‘prototyping fast’, ‘shipping early’ and ‘iterating often’. Despite Google increasingly dabbling in hardware, software undeniably remains their weapon of choice and rightly so - it affords them the flexibility of testing fast, changing direction quickly and scaling near-effortlessly.
Automotive needs to embrace software and the prototyping process more than it has done so far. Indeed, forgoing hardware to make software the key interface for the car is already happening at the high-end with manufacturers introducing large displays with virtual instrument clusters, but there is a case for going further and substituting digital for the physical across a manufacturers’ full product range as we’ve started to see across the industry.
Bain & Company have stated that automakers cannot achieve their innovation goals using conventional waterfall methods for R&D with automakers needing to “embrace Agile methods and iterative, test-and-learn processes, particularly for the development of digital features”. Bain have also said that they “expect innovation and productivity to accelerate as these ‘new methods’ catch on and become more consistent in their application”.
Bain and Company’s point of view shouldn’t be surprising, many manufacturers are tuning into a more rapid approach to early prototyping, and there are obvious reasons why, for instance:
If automakers are to ramp up their R&D efforts and bring their offering consistent with the standards soon to be set by the likes of Google, Amazon and Apple, a consistent and easily repeatable approach to designing and testing in-car interfaces is required. One that is small and lightweight enough to shortcut building out a physical prototype as the first port of call but one that is also rigorous enough to elicit true insights.
Admittedly, this being the interior of a car and the inseparable driving context makes this a challenge that requires careful consideration. We outline below a framework for designing and testing in-car interfaces whilst keeping this unique context in mind.
Comprehensive research is the precursor to any design process and as far as in-car interfaces are concerned, there exists a wide range of inputs to consider.
Depending on the challenge at hand, we can frame our consideration of existing in-car offerings in a number of ways:
The goal of this stage is to build a detailed picture of the current in-car user experience in order to set a benchmark for best in class, to consider trends to get a handle on where the industry is headed, and to synthesise that into clear guidance on where the imminent opportunities may lie. Remaining forward-looking is important as the lead time for producing a car means that we must be looking slightly ahead of the curve to ensure whatever we test and verify does not become outdated or obsolete by the time it enters the relatively inflexible production process.
The purpose of this process is to create digital in-car interfaces which will be used in a driving context. Therefore, rather than talking to customers, there is a need to go further and use a technique borrowed from the user-centred design school of thought: contextual inquiry.
Contextual inquiry involves observation of people in situ. From experience, there is a big difference between what people say they do when they are asked to think about an everyday scenario and what they actually do when they find themselves in said scenario. This can be due to a number of reasons ranging from a lot of our actions taking place in autopilot (which is likely to be true of many behaviours in cars) to it being difficult to systematically think through the minutiae of our behaviours, especially when you are being questioned by a complete stranger in unfamiliar surroundings.
Therefore, to circumvent these challenges, it pays to get close to a driver in their natural habitat. Clearly there is a need to be as discreet as possible so as not to overwhelm the test subject. That’s not to say communication shouldn’t take place because it certainly should in the form of focused, purposeful questions but the conversation shouldn’t become a distraction from observation.
Some general areas of focus for contextual inquiry include:
Admittedly, this is still an imperfect scenario as a large sample size of data is required to draw robust conclusions across a range of different scenarios such as a long trip, school run etc. To augment contextual inquiries, a video log of driver journeys over the course of a few weeks can be used to gain a more rounded picture.
The outcomes of contextual inquiries and video logs captured across markets and driver types can be organised into an affinity diagram (a diagram which allows us to give structure to a large amount of data by identifying themes and natural relationships) before finally making their way into a set of personas, which act as a focusing device for designers.
Before the design process is considered, serious thought must be given to what is arguably the most difficult aspect of designing and testing automotive digital experiences quickly and at low-cost - namely, how do you re-create a realistic driving experience and also integrate designed prototypes into something that feels like a cohesive whole that can be scaled?
The need for recreating a realistic driving experience goes without saying - you must be able to test how effective the driver is at completing tasks on a prototype whilst they are under the stress of driving a car and the prototype must be able to hold up its end of the bargain by reacting to driver input. Consumer hardware can be used at a basic level, and scaled up to more enthusiast/professional level equipment depending on the level of accuracy required.
Use of existing product prototyping methodologies and equipment can allow for a lot of flexibility, from visual prototypes running on tablet devices, or voice assistant prototypes and physical buttons, more or less any interface can be recreated and attached to a testing rig. High-fidelity UX/UI prototypes will be created using either a prototyping tool (such as Framer) or in native code (e.g. Swift) depending on requirements and will be connected to a real time cloud database. This allows us for prototype interactivity as we will be able to send data from one device to the next and keep track of the interactions.
The beauty of using software is that we can push updates to the prototyping system and move between different UX/UI implementations across vehicle types, buyer profiles and geographies - on demand - creating a truly versatile and scalable prototyping toolkit.
Once we have designed different UX/UI implementations and recruited representative test subjects (including both existing customers and non-customers), we are ready to conduct testing.
During the testing, users will be given a few minutes to familiarise themselves with the prototype before being asked to undertake a driving task such as staying within a particular speed threshold and/or changing lanes/following a particular route on a simulation whilst simultaneously being asked to conduct various tasks on the prototypes in front of them - these will be set scripts in order to allow for reproducibility and comparison.
Whilst the test is being conducted, quantitative and qualitative data will be collected to gauge the effectiveness of the design as well as the safety of the prototypes. Participant actions can be recorded on a camera mounted behind their shoulder whilst one on the windscreen display can be used for primitive eye and facial expression tracking. On the prototype itself, analytics packages such as Google Analytics can be integrated to track button presses and flows taken by drivers whilst Hotjar analytics can provide heatmaps of user behaviour for a post-mortem.
More qualitative data can be gathered through interviews with the participants where specific questions are asked about the design and ease of use, as well as broader opinions and experiences around driving. There’s a reason Silicon Valley relies so heavily on face-to-face research, and it applies just as much when developing digital experiences inside the car. Mental workload can also be assessed qualitatively via the Driving Activity Load Index method, where participants are asked to rank their mental workload when conducting tasks on six numerical scales.
These are some examples of the approaches that can be taken to get a holistic view of the performance of our prototypes and how they may translate to the real world.
In order to ensure a consistent analysis approach, outcome analysis must follow a common set of measures. Over time, a consistent range of questions and scoring approaches will lead to a benchmark for subsequent testing that might need to be undertaken for new features/refinements to prototypes. The performance of the design can be analysed via a range of predefined scoring areas including some of the following: Timeframe for each task measured against an acceptable ceiling and the average time taken across prototypes Number of errors or incorrect paths taken for any given tasks measured against an acceptable ceiling and prototype average Focus away from the ‘road’ in seconds Enjoyment - measured via a numerical scoring system Brand perception - measured via a numerical scoring system via questions such as propensity to buy, whether they would recommend to a friend, meets expectation of brand etc. Alongside a standardised scoring system, standout UX issues that arise from testing can be plotted on an impact severity vs incidence chart to gauge urgency, and also as a means for triaging and prioritising issues for the Design team.
The performance of the design can be analysed via a range of predefined scoring areas including some of the following:
Alongside a standardised scoring system, standout UX issues that arise from testing can be plotted on an impact severity vs incidence chart to gauge urgency, and also as a means for triaging and prioritising issues for the Design team.
In a world where software has proliferated through almost every industry, with UX/UI extensively tested to the point where the placement of a single button is scrutinised heavily due to its very real effects on usability and conversion, the opportunity to embrace those techniques within the automotive space is highlighted.
Automotive’s challengers such as Google and Apple are masterful exponents of software and the rigorous, scientific approach to prototyping, testing and refinement.
This piece aims to take the most transferable aspects of this mindset and apply it to the very real challenge of creating in-car interfaces and experiences that are befitting of today’s world, and turning software experience into a real advantage for a car brand rather than a ‘good-enough’ hygiene piece that it has been in the past.
The proposed prototyping kit and approach is our attempt to successfully trade off realism against the ability to test in-car interfaces rapidly in a scalable manner. The prototyping toolkit should be seen as a one-off investment that is relatively inexpensive compared to embarking on a more extensive prototyping process.
We believe manufacturers can reap significant dividends with such an approach and begin to overhaul the negative perceptions of current in-car UX. With faster lead times on features, shorter update cycles and rapidly growing customer expectations, adopting test-and-learn approaches to design, test, and develop rapidly is crucial to rise to the myriad challenges and opportunities offered by these technologies.
