Biren Shah was most recently VP of Product Management and Engineering at J.D. Power. He started his career in software engineering at National Instruments before transitioning into product management. Biren then joined Demand Media as a product manager and later transitioned to SolarWinds, where he led product management for the web properties function. Before his most recent position at J.D. Power, he served in various product management roles at Vast (acquired by VROOM) and Trilogy.
In our conversation, Biren talks about how he’s led initiatives to optimize the user experience across the entire customer journey, from initial discovery all the way to post-purchase. He shares how these innovations have helped transform customer experience within the automotive industry and discusses specific examples of how he led the redesign of a new customer portal at J.D. Power.
I’ve spent most of my career in automotive retail. Customer journey mapping is uniquely complex and rarely linear — it spans across multiple disjointed systems, and there are four primary digital channels from which a dealership gets qualified leads:
Unifying these touchpoints is an incredibly complex challenge because of disparate systems and the need to put the data together. At J.D. Power, we built an omnichannel automotive business analytics platform, which essentially integrated all these different providers and original equipment manufacturers (OEMs). As a result, we were able to ingest comprehensive data on traffic, website events, campaigns, lead submissions, inventory, and more. Additionally, we had low-level visibility into the entire dealer sales funnel via CRM events as well as sales.
We tracked the entire customer journey from the research to consideration to submitting a lead and going through the purchase funnel. In some cases, we could also track post-sales. There’s a very complex customer journey here, and the challenge was to integrate with all these different channels while also having consistent, clean data that could be incorporated within our system.
One example is from a time when an OEM did a marketing campaign and was successful in driving traffic to their Tier 1 website. While we saw strong traffic and lead submission volumes, the conversion of those leads into sales was not meeting expectations. We analyzed the data and, by watching the customer journey throughout the entire cycle, identified that the dealers were not doing a good enough job following up on those leads.
When we uncovered this problem, we allowed the OEMs to give the dealerships specific lead handling instructions along with the leads in real-time, so that salespeople could understand how to respond better. Hence, they’d provide a much better customer experience. We immediately saw a significant lift in the close rates because the dealer was able to accommodate the request appropriately. This showed that analyzing data across the entire customer journey is critical in running operations.
Understanding behavioral intent has been at the center of my work for the last decade in automotive digital retail. I’ve led the development of a buyer intention engine that uses machine learning to score incoming leads based on their likelihood to convert. I also managed a team of multiple data science engineers who built that engine for us. Now, through deep customer understanding and machine learning techniques, we were able to find patterns in the customer journey that would predict the purchase at the end.
For example, we discovered that certain behaviors, when performed in aggregation, would provide more propensity to purchase. One example is repeat visits within a short time, either across the same website or different ones within our network. We could track this behavior across multiple channels to see who’s focused with the intent to buy versus someone who’s in the initial stages of researching. Lower-level funnel tools like Trade-In and Finance calculators were good indicators as well. We discovered surprising factors like a customer’s proximity to dealerships was a reliable indication of the likelihood to purchase from the local dealership.
We uncovered all of these things by running advanced machine learning regression analysis. Those indicators were then fed into the intention engine to give us a propensity to purchase score, which we integrated into our operations in real-time.
We built a personalized incentive offer engine based on our buyer intention engine for a major OEM. It specifically targeted on-the-fence shoppers. These are prospects who are showing high intent but aren’t ready to pull the trigger.
In real time, we were able to identify these users and activate monetary incentives so that they would be incentivized to purchase the car. The system was designed to exclude people who were just tire kickers who had no intention to buy the car, as well as people who had extremely high intention to buy the car because they were already about to do it anyway. We saw a 3 percent increase in conversion rate by employing this kind of model in the sales process.
Build very robust instrumentation and map out the entire customer journey. You need to identify behavioral signals that are coming from the different stages of the journey. That’s the most important foundational piece that I would advise any product manager working on this problem to do.
We ran a lot of A/B tests to validate the fine-tuning of this buyer intention engine. We had a lot of hypotheses, and through regression testing, we were also able to understand certain behavioral patterns, such as the average household income, and its relationship to specific models. My advice is not to stop at A/B tests based on what you think is going to work — try others that could be on the fringes as unexpected results often turn up while doing A/B testing.
Lastly, it’s important to design a product architecture that will enable real-time decision-making. The holy grail is to be able to identify these behavioral signals as they’re happening to deliver a personalized experience to the customer. Unless you have a system that’s fast enough to identify that and personalize it, you’re never going to be able to make that effect.
The driver behind the portal redesign was the rapid evolution of our business. In the last five years, we doubled the number of participating dealers. We added four major OEM brands to our portfolio, and we significantly expanded the Smart Digital product portfolio. As our products and the customers increased, our platform needed to scale with it. Increasing complexity was a big driver as well.
Our portal needed new, more complicated capabilities based on the customers coming in and the features they requested. For example, customers asked for educational resources like product guides, best practices, and video walkthroughs to get more value from our tools, so we had to build all of those and a way to deliver them intuitively.
There was also a lot of tech debt that we had to clear up. The portal became increasingly fragile, and it was difficult to scale, so we had to turn to more modern architecture to address some of the problems we saw.
We decided to form a design committee of different stakeholders since the portal was going to be used by many types of users, including our own company and customers. The product team executed a deep discovery process by running targeted interviews and feedback sessions with various stakeholders to understand top pain points.
From there, with the participation of the design committee, we prioritized these pain points. And through an ROI framework, we essentially ranked all the different needs and aligned them with our long-term product goals. Once we had a good understanding of that, we worked with our design team to build wireframes and prototypes, which we tested in cycles with stakeholders before finalizing the roadmap. We discussed the prototype with the engineering team to understand the cost to build it, and everything came together in a big loop.
Our biggest lesson was that we couldn’t redesign this portal in a vacuum. We needed to incorporate inputs from all the different stakeholders and then go through an iterative cycle to build something that fits. That was the biggest key insight, especially since executive stakeholders had a very different understanding of the customer’s needs compared to our end users.
The biggest challenge in automotive platform strategy is balancing the need to integrate with the wide range of providers while also simultaneously building scalable features that meet the distinct needs of a particular automotive OEM. Each automotive OEM has unique program requirements, but the platform needs to be able to remain cohesive and maintainable across all the different clients. To address this, we focused on three main areas.
The first was standardized data integration, which we invested heavily in. Most dealers use a common set of CRMs to do their day-to-day work. By building and maintaining a reusable certified integration layer between J.D. Power and the different CRMs, we insulated our platform from many of the changes the CRMs are making. This allowed for normalized data ingestion, so we got clean and reliable data back from the different CRMs.
Secondly, we created a common framework for the customer journey KPIs and reporting. Rather than allowing each OEM to define its own customer journey metrics, we aligned the stakeholders around a common set and journey. We had typical things like total leads, unique leads, etc. We also measured the contact rate from the dealer to the customer and the connect rate from the customer back to the dealer. That helped us measure how effective the communication was with the dealership.
The third was adopting a platform-first approach for feature development. Instead of building custom, one-off solutions, we would design the new OEM requested feature so that it was configurable across the different platforms and clients. We could toggle features on and off depending on their preferences. We sometimes had unique use cases for one OEM versus another, so we built a Custom Report Builder. That was highly successful as it allowed for greater flexibility for OEMs.
Right now, one of the biggest areas of disruption is conversational AI, especially in the early customer stages of research and considerations. These tools are powered by LLMs and can simulate a real customer salesperson. They’re answering questions about models, features, and availability. They’re also scheduling test drives and appointments directly through CRM integrations, and then they are handing off the conversation to a real person when the time is right.
Traditionally, we’ve seen that dealerships don’t always do a really good job responding to leads. The industry average is that 40 percent of qualified leads don’t get answered appropriately. AI bridges the gap and optimizes the customer experience. With that said, most of the dealers still prefer to talk to a customer once they move into the pricing and negotiation phase.
It’s still in early stages and very tightly controlled. Trust and accuracy are really important when you get into the purchase phase. AI solutions have not totally cracked the lower funnel phases, so there’s still some work to be done.
Legacy ecosystems (dealer CRMs) are lagging behind in AI adoption compared to Conversational AI companies. Many of them are not even able to support real-time AI-driven engagement. Rather, they’re focusing more of their attention on identifying in-market shoppers using the data they have within the CRM or generating templatized messages for dealers to give more follow-up to leads.
AI is evolving at a rapid pace, and while it’s difficult to predict the exact changes over the next 5–10 years, I see AI playing a major role in transforming CRMs to become more deeply integrated into dealer systems.
Looking ahead over the next 0–3 years, I anticipate AI becoming increasingly embedded in dealership systems. AI will enhance inventory management by predicting which models are selling fast and suggesting competitive pricing based on market trends. AI-driven engagement will make it difficult to distinguish between human and AI interactions, with adoption across all channels like web, email, SMS, and phone.
AI will also be integrated into traditionally human-led areas, such as price negotiations, trade-in evaluations, and personalized virtual showrooms using augmented reality. Additionally, it will enable intelligent lead qualification, prioritizing high-intent buyers, and provide AI-based coaching and training for dealership salespeople. On the post-sales side, I predict that AI will support predictive maintenance alerts, proactive service reminders, and enhanced customer support to improve buyer retention.
LogRocket identifies friction points in the user experience so you can make informed decisions about product and design changes that must happen to hit your goals.
With LogRocket, you can understand the scope of the issues affecting your product and prioritize the changes that need to be made. LogRocket simplifies workflows by allowing Engineering, Product, UX, and Design teams to work from the same data as you, eliminating any confusion about what needs to be done.
Get your teams on the same page — try LogRocket today.
Want to get sent new PM Leadership Spotlights when they come out?
One of the biggest changes to prototyping has been the rise of tools that leverage the power of AI to simplify the development process.
Espen Scheuer, Head of Product, Patient and Developer Experiences at NexHealth, shares how NexHealth structures its product organization.
Itzik Mitzmacher talks about how to avoid becoming a feature factory by shifting your product team’s focus from outputs to outcomes.
Canva offers hundreds of thousands of free templates and 100+ million design assets to appeal to all different types of users.
Learn more →
