Conversation with Impel global co-founder and CEO, Devin Daly, speaking with Cox Automotive Australia’s Mike Costello from the US.
Q: How do we demystify the opaque conversation around AI? You’re obviously someone who understands this stuff incredibly well, better than most. But for a dealer using AI really well, what can it mean for a customer and how can we demystify it?
A: I think a good place to start is to think about what the customer experience looks like when you’re not using AI.
In the US, and I believe in Australia as well, a lot of times a customer goes to buy a vehicle, they fall in love with their vehicle, they fall in love with their sales rep. It’s the second largest purchase of their life. And yet after they buy the vehicle and drive off the lot, they almost never hear from that dealership again. If they do, it’s often HTML marketing content.
So that sort of state of play with without AI, when you adopt AI, you instead have this 24/7 concierge that’s reaching out to the consumer, reminding them they might need service, nudging them, knowing they’re doing what’s right with the vehicle, which again, just really enhances the customer experience, all the while improving lifetime value for a dealership.
So, it gives you that luxury 24/7 VIP concierge that you can text with at any time, almost like, you know, it’s your rap bat, you know, luxury clothing store like Tom Ford. It’s like the most committed staff member you could ever get.
Q: We’re looking at this through sort of an Australian lens, but where is the US automotive market today in terms of the adoption of AI? And what are the main applications of AI being adopted by dealerships over there? Impel after all, is kind of right at the coalface of all this stuff.
A: We are definitely the market leader and kind of category creator of conversational AI for auto. You know, we’ve been using generative AI going back to 2018.We’ve done 3 acquisitions in the space. We’re trying to drive a lot of this innovation.
In regard to US adoption, it’s definitely accelerating rapidly. You know, we believe this is the year of mass AI adoption where, you know, we’ve really crossed the chasm, right?
There’s this thing called the Gartner Hype Cycle. It starts with the technology innovation, and then it rises up massively and there’s this trough of disillusionment, and then you cross the chasm and that’s where we believe we are today.
Roughly today, I think 1/3 of dealerships use AI in the US in some capacity whether that’s marketing, advertising, accounting or conversational AI. 81% of dealers surveyed plan to increase their AI my budgets in 2025.
In terms of main applications, I would say doing sales response right, there are a lot of challenges with having humans responding to a huge influx of inbound leads. They don’t respond overnight, they don’t do long term follow up. A lot of times they don’t even answer the customer’s question. So that’s sort of like asynchronous sales follow up. Second is instantaneous web chat in the US market at least. There are two forms of web chat solutions that are out there. One is human staffed, which results in 11-minute delays on average.
And you know, Google just did a study. If your page doesn’t load in 1.1 seconds, you get deprioritised. That’s how short people’s attention spans are today.
The second category is non-AI, a kind of glorified lead capture, you know, which is really just using analogue scripted logic, and they basically make you submit your personal information before they’ll answer any questions. AI has come in and really allowed us to disrupt both of those types of different chat providers, providing instantaneous responses that really add value to the consumer.
The third big area we’re seeing AI be applied is doing ongoing service outreach. Again, having that concierge-like experience reaching out to you, reminding you might need service, checking in, learning you have an oil change special that’s going on, letting you know if there’s a state inspection.
I don’t know if that exists in Australia, but it’s a similar concept. If you think about, in the service life cycle, every year there’s probably 50 touch points around when you should be reaching out to that consumer. The average DMS in the US has 5000 leads. You start to do the math of how much outreach would be required to actively reach out to those folks and keep them coming back.
It is completely unmanageable using human staff and is a great task for artificial intelligence.
Q: So customers in that situation who are engaging with an AI as a follow-up process, do they know that they’re talking to an AI and do they care? Or to them is it just a matter of ‘it doesn’t really matter if it’s a human or if it’s AI, as long as I get what I need’?
The vast majority of cases, they do not know that it’s an AI. In the US market, we typically recommend that they name it like a staff member. Some dealers go so far as to put, you know, a picture on the meet our staff page.
You would be shocked how often people honestly do funny things like ask the AI out for coffee or, in all seriousness, they will come into the dealership, and they will ask ‘hey, can I meet, you know, Michelle?’, which was the AI’s name.
Obviously in certain places there is required disclosure. So like California, there’s a bot disclosure law, which we obviously adhere to and see in the signature, we will denote that you’re communicating with an AI agent.
The consumers love it though. You would be shocked not only how often they believe it’s a human and indicate that, but there’s often cases where a human will jump in, and the consumers will actually ask ‘hey, can I go back to talking to Michelle?’
They actually think the human is a bot and vice versa. In most cases, consumers don’t know, and they absolutely enjoy the experience of engaging with the AI.
Q: Is there a degree of inertia or reluctance from dealers to embrace AI? Do they look at it and go, ‘this is going to take my job or it’s going to somehow be an existential threat to me? How do you confront that, and do you notice if that’s sort of dissipating and changing over time?
A: First off, never consumers. Again, they’ve always very much enjoyed it and know it’s an enhanced experience over what they experienced previously.
Some dealerships can be scared that, you know, AI is going to take jobs. The reality is, right, anytime there’s a major innovation, people are worried about job loss or job dislocation. In reality, things like the printing press, the automobile, the assembly line ended up actually creating incremental jobs or labour opportunities.
What I would say is in the last year, I would say as we’re seeing large public dealer groups, large OEMs that are adopting AI. I think that feeling is dissipating. And it’s our firm belief that if you don’t embrace it, it’s like dealerships that resisted the Internet in the ‘90s or the early 2000s. You will be passed by.
We firmly believe that in the next 5 years, there’s going to be an AI and automation evolution in our space, right? Dealership gross margins are 5%, 4%, 3%, yet they’re massively human capital intensive.
It’s our firm belief that in the next 5 years, we’re going to see dealerships go from massive human capital intensity towards AI and automation to cope with unprecedented volatility and low gross margins.
Q: You touched on a point that we feel over here in Australia too. Margin compression is real, and dealers are looking for every efficiency they can find. But before we pivot to that, are automotive OEMs who use franchise dealers, adopting AI? And more importantly, I suppose, are they driving the adoption of this stuff within their networks?
A: Absolutely. I would say leading OEMs are not just experimenting with AI, they’re embedding it across their operations and mandating dealer adoption.
I would say a couple things there. Many OEMs in the US have invested in consumer data platforms on a centralised repository for consumer data, and they spent a lot of money and time and energy, and they failed to really figure out how to drive Tier 3 dealership value from those CDPS.
AI is the perfect activation of a CDP in that, number one, it allows for very long tail personalisation, leveraging those consumer profiles. But number two, we capture a lot of incoming information and data from consumers when they’re engaging with the AI that can be used to actually further hone those consumer profiles within the CDP. So that’s one area we’re definitely seeing from OEMs.
The other area is dealerships realise that a lot of times the tier 3 communications are very brand non-compliant. You know, an OEM spends so much money on creating an incredible brand experience at the tier 1 level, but the tier 3 execution is very lacklustre.
Not using the OEM mystery shop guidelines, misspelling vehicle features, OEMs see an opportunity instead to establish those brand compliant AI communication protocols at the tier 1 level and then cascade them down for tier 3 communication.
Q: So, are we going to see a sort of a growing gap between dealers who are AI-enabled, already embracing the technolog,y and those that are a little bit more behind the times, that might have some reluctance?
To what degree are we going to see a divergence in terms of the ability for these businesses to deliver what they need to deliver? And is that going to happen quickly or is it going to be more of a progressive kind of separation?
A: To date, I think we’ve seen a slow divergence. Some dealer groups are seeing double the closing rates of dealerships not embracing AI.
But what I would say is, with how fast generative AI is moving — seems like every week, every month there’s a new model released, a new use case, a new video on the internet – you never know which one of those could fundamentally change a dealer’s business and be that step change that accelerates that divergence pretty massively.
Q: So, on that note, we really need to be thinking about, we’re at the very beginning of this really, aren’t we? We’re not even nearly at the end point of when this technology is mature, it’s still got so much to go. Where’s it going to go next in a general sense? How much more useful is it going to get?
A: I think two thoughts there, right? I think one is the rise of agentic AI, which is obviously something that’s been talked about a lot. Unlike traditional automation tools, agentic AI can fully and proactively manage and complete tasks, can make decisions in real time, continuously optimise based on outcomes.
This means AI systems that don’t just respond to leads, they manage them end-to-end, scoring, nurturing, following up, re-engaging cold leads.
Another key trend we’re seeing is predictive personalisation. AI is getting much better at using real-time data such as browsing behaviour, service history, market conditions to serve hyper relevant messages across every channel, email, text chat, WhatsApp, soon to be voiced.
This creates really meaningful and relevant engagement and really improves lead-to-appointment conversion rates materially.
Last thing I would say is we are definitely seeing a rise in dominance of, you know, one AI brain, right? Impel’s sort of product vision is having one AI central nervous system that controls the entire life cycle from sales to CSI survey to F&I to accessories to that bond of service loyalty, eventually the vehicle acquisition that allows us to really share insights across all of those various touchpoints.
Q: In a market like Australia, where dealers are facing so many challenges, do you reckon the use of AI to bolster productivity and automate repetitive tasks comes into its own? And does this explain perhaps why Australia is a strategically important market for Impel as it rolls out across the globe?
A: I think starting with the second part of the question there. Australia is important for us as you know, it’s an English-speaking market. Even though generative AI allows for very seamless translation, localisation into another language is still not totally turnkey.
We also find it’s somewhat similar to the US in terms of, you know, the free reign of dealerships, the economics, the structure, so it’s a market that we’ve been excited about and potentially it may provide a springboard into some of the more Asia-Pacific markets.
I would say what you’ve just described though, absolutely underpins a situation where Australia is even more poised for rapid AI adoption. You know, if you think about in every market, the volatility, like I alluded to a minute ago, over the last five years has been unprecedented.
You went from COVID where dealerships were shut down to making record profits and inventory scarcity, to a bit of a glut of inventory in 2024 to now this tariff environment, right?
We have unprecedented volatility in the Australian market. You add on top of that the uncertainty that Chinese EV entrance brings to this entire dynamic, and are they going direct to the consumer? Is it an agency model?
I mean that is the benefit of AI, it allows you to scale up and down very seamlessly with the ebbs and flows of market dynamics. It also allows you to shift strategy very, very seamlessly and a lot more easily than human capital.
If you think of this tariff environment in the US, for instance, we’re going to see a decline in new car volume, you want to focus on used vehicle acquisition. You can quickly deploy an AI agent that does that in a day instead of letting five people go from your sales team and hiring five people to do car buying.
AI allows for very, very limitless adaptability and scalability. I think that’s a really good way to think about it, you’re saving so much time and resource and having to redeploy staff because you know, these products and services, they’re able to just pivot like that.
Q: Just to wrap it up, what is your advice for dealers just starting or thinking about starting, but not quite sure, to implement AI, particularly regarding trusting the technology? I know trust is a huge part of this and I sense that’s kind of going to be everything when it comes to the roll out of this.
A: I think it’s really important to start with clear, measurable use cases. You treat AI as an enabler, not a replacement for staff. Start where AI can immediately reduce friction and remove organisational drag by offloading menial, repetitive tasks.
So things like lead follow up, appointment service, appointment scheduling, answering repetitive customer inquiries, these are areas where AI is already proven and success there will build confidence in incremental use cases.
Secondly, I would say it’s critically important that you choose a trusted partner that’s got a lot of automotive experience. In and outside of auto, there’s increasing evidence that verticalised AI will win out due to training on specific vertical or industry use cases. So, identifying a vertical specific provider, ensure they integrate with your existing tech stack, your CRM, your DMS.
Lastly, definitely have a focus on security. You want to work with an enterprise-grade provider. AI is not something you could just plug into ChatGPT and run wild, right? AI unbridled is subject to hallucinating or lying about your business. It will not drive business outcomes. So, identifying somebody that is enterprise grade and has those security measures in place is really critical.
Learn about LeadDriver
LeadDriver, with AI-driven SMS and email conversations, boosts conversions by responding to online leads within minutes, providing after-hours coverage, engaging shoppers in genuine dialogue, answering VIN-specific questions, capturing trade-in information, and following up with dormant leads for up to 51 days.
To learn more, click here.
