Root, Inc. Q3 FY2020 Earnings Call

Good afternoon, and thank you for joining us today. Root is hosting this call to discuss its third quarter earnings results for the period ending September 2020. Participating on today's call are Alex Timm, Co-Founder and CEO; and Dan Rosenthal, Chief Financial Officer. Earlier this afternoon, Root issued an inaugural shareholder letter announcing its financial results. While this call will reflect items discussed within that document, for more complete information about our financial performance, we also encourage you to read our final initial public offering prospectus dated October 27, 2020, and filed with the Securities and Exchange Commission, and our quarterly report on Form 10-Q for the third quarter of 2020 to be filed with the Securities and Exchange Commission this week. Before we begin, I want to remind you that matters discussed on today's call will include forward-looking statements related to our operating performance, financial goals and business outlook, which are based on management's current beliefs and assumptions. Please note that these forward-looking statements reflect our opinions as of the date of this call, and we undertake no obligation to revise this information as a result of new developments that may occur. Forward-looking statements are subject to various risks, uncertainties and other factors that could cause our actual results to differ materially from those expected and described today. In addition, we are subject to a number of risks that may significantly impact our business and financial results. For a more detailed description of our risk factors, once again, please review our final IPO prospectus, in our upcoming Form 10-Q, where you will see a discussion of factors that could cause the company's actual results to differ materially from these statements as well as our shareholder letter release today. A replay of this conference call will be available on our website under the Investor Relations section. I would also like to remind you that during this call, we will discuss some non-GAAP measures in talking about Root's performance. You can find the reconciliation of those historical measures to the nearest comparable GAAP measures in our shareholder letter released today in our filings with the SEC, each of which is posted on our website at ir.joinroot.com. I will now turn the call over to Alex Timm, Root's Co-Founder and CEO.

Thank you, and good afternoon. We are thrilled to be speaking to both new and prospective investors today on our first earnings call as a public company. We hope you have an opportunity to glance through our shareholder letter, which we posted to our Investor Relations site. The letter does a great job in providing insights into why we built Root, but perhaps even more importantly, lays out how we are executing on our aggressive vision to fundamentally and positively drive change in the U.S. auto insurance industry. This is a very exciting time to share our story with you. The opportunity we see before us is monumental, and we follow three core objectives to guide us as we execute on our plan. First and foremost is our objective to drive significant growth. Next is to enhance profitability via loss ratio and retention improvements. And our third core objective is to optimize customer acquisition via direct marketing and a strong user experience. But we have accomplished so much already in the five years since I co-founded the company with my partner, Dan Manges. We remain obsessed with building the Root of the future with a proprietary telematics algorithm and a burgeoning and integrated data set in a highly skilled responsive, smart, and productive team of people. I'll double-click on some of these main themes before turning the call to our CFO, Dan Rosenthal, to walk you through our successful execution across the core objectives I just laid out, as well as to introduce you to our financial principles and share our guidance, after which we will open the call for your questions. We are still in the very early days of Root. As we look to the future, we believe the opportunity in front of us is massive and that we have what it takes to create a historic market-defining company. We align all of our strategic decision-making with delivering long-term profitable growth, driving our market leadership with technology innovation and pushing our commitment to revolutionize the insurance industry. We founded Root on the belief that machine learning and modern technology could fundamentally revolutionize a state insurance industry that's ripe for disruption while delivering a vastly superior consumer value proposition. More simply put, we wanted to utilize technology and data science to solve some of the most challenging big data questions out there and fundamentally change the way auto insurance is underwritten. With $266 billion in annual premium, the U.S. auto insurance sector is an enormous market. The product is a government-mandated purchase for the vast majority of motorists. Nevertheless, the pricing of auto insurance has very little to do with how consumers actually behave behind the wheel. Consumers have little to no control over their auto insurance. Their policies are often priced using stale demographic information that is hard or impossible to change, such as age, gender, marital status or education. It's also an industry where historically, lower-risk good drivers are systematically overcharged to subsidize losses that emanate from the higher-risk bad drivers. There's a fundamental element of unfairness inherent to the traditional insurance industry. This is the industry that I grew up in, and it's the thing that has always bothered me the most. Consumers deserve better, and we believe technology and data science are the keys to unlocking products that will enable that change. While technology has radically altered and improved so many aspects of our lives, the insurance industry mainly still operates the same way as it did a century ago. The industry continues to rely principally on archaic variables that do not measure driving behavior and are unfair to consumers. Further, it has not awakened to the reality that consumers are walking around with supercomputers, otherwise known as smartphones, in their pockets that can offer a deluge of individualized driving data on a daily basis. We built Root as a mobile app to capture telematics data that uniquely emanates from the smartphone. As data and predictive analytics are the foundation of insurance, the industry is in a prime position to be disrupted by an innovator leveraging advanced data science and machine learning techniques. We believe that Root has a material first-mover advantage, given our central focus on telematics and mobile-first consumer experience, combined with our balance sheet strength. In the five years since founding Root, we've developed industry-leading and highly proprietary telematics and data science capabilities. We are particularly proud of our ability to discern distracted driving—a meaningful driving risk that cannot be measured effectively by OEM-integrated car data or dongle devices provided by carriers. The National Safety Council reports that one in four accidents in the United States is caused by distracted driving. So the ability to measure distracted driving is paramount to a strong telematics program. In addition, actions like hard braking cannot accurately be used to forecast loss cost without a corresponding claims data set and other contextual behavioral data captured via mobile telematics, which Root possesses in spades. We have been asked why most of the industry isn't focused on implementing something similar. The answer is that it's extremely hard to construct and even harder to apply to an existing book of business, particularly with legacy systems. Our telematics are different because they are built on a proprietary integrated data set of complex behavior data tied directly to actual claims experience. And we use the power of this data across our entire portfolio. In fact, behavioral driving data is the first thing we look at when considering a customer's risk. And in the case of the 10% to 15% highest risk drivers, we underwrite these customers out without considering any other variables. To highlight how this differs from the legacy carrier side, one of the large national players recently acknowledged publicly that although telematics is their 'most powerful rating variable', they solve it last in terms of their pricing algorithms. So why is this the case? Incumbents face a classic innovator's dilemma. If an industry incumbent sought to implement telematics-based pricing across its customer base, it would risk large-scale lapses as customers react to those price changes. This risk of cannibalization simply outweighs the benefits to the legacy players. Some have argued that industry players face such innovators' dilemmas all the time. But unlike the traditional rating variables, telematics data is collected by these carriers after they have already acquired the customer and spent an entire six-month term with them. We believe Root is the only carrier with scale and focus working to implement telematics-based pricing upfront. Moreover, no other carriers are built on proprietary modern technology like Root. That allows us to rapidly test new behavioral data elements and retrain our underwriting pricing models in real time. Our data advantage and engineering capabilities allow us to streamline quote flows, create faster and better claims experiences, and in the future, will allow us to even further differentiate our product, pricing and service. Ultimately, our goal is to implement a full behavior-based pricing model. We've already removed the use of education and occupation from our pricing, and we are committed to removing credit score as well. In conjunction with the National Association of Insurance Commissioners Summer meeting in August of this year, we announced our Drop the Score initiative, outlining our plan to remove credit score from rating factors. We have made a commitment to remove the use of credit score from our rating variables no later than 2025, and we're advocating for and challenging the broader insurance industry to follow our lead and do the same. We are still in the very early days of Root. With the successful completion of our IPO and concurrent private placement on October 27, 2020, and our revised reinsurance program in place as of July 1, 2020, we now have more than $1.2 billion in fresh capital and a capital-light business model to take full advantage of the massive growth opportunity in front of us. You can say that we're competing with legacy carriers who have really deep pockets and have been building their brands for nearly a century. But our real competition is inertia. The inertia of an old and archaic industry that is slow to adopt new technology and even slower to embrace meaningful change. So how do we know it's all working? I'll now turn the call over to Dan to take you through the significant progress we continue to make across our core objectives as well as provide financial highlights in the quarter and share additional color on our outlook. Over to you, Dan.

Thanks so much, Alex, and good afternoon, everyone. Full details of our third quarter results are available in our shareholder letter. So I'm not going to repeat many of the numbers, but I would like to discuss a few key results and strategic milestones, highlight some aspects of our financial framework to help you understand the fundamentals of the Root model and provide more detail on how we are thinking about the rest of the year. I'll do all of this through the core objective framework that Alex spoke about at the top. A successful insurance technology business like Root needs to drive significant growth, enhance profitability via loss ratio and retention improvements, and optimize customer acquisition via direct marketing and a strong user experience. Across each of these three core objectives, our year-to-date trends have improved significantly as our business continues to mature. Growth is our top priority as it enriches our flywheel with more data. To this end, we posted strong growth during the third quarter and year-to-date periods by all measures that we deem important. We ended the quarter with premiums in force of $600 million, up 41% from last year. Total policies grew 35%, and average auto premium increased 6% compared to the prior year period. Direct written premiums were $471 million for the nine months ending September 30, 2020, up 53% versus prior year. And the amount of our policies earned year-to-date, direct earned premium, grew 93% to $450 million compared to the nine months ending on September 30, 2019. The 41% growth rate in premiums in force based primarily on strong share expansion in existing markets launched in 2019 or earlier demonstrates the depth of share available in our currently active states. Our team is obsessed with understanding the local factors that drive our customers' decision and allow us to continue to grow in each market we serve. Now that we have a more mature product informed by millions of customer experiences, we plan to bring Root nationwide. After disciplined expansion into 30 states, we're ready to accelerate that reach. To this end, we are incredibly excited to announce that in November, we closed the acquisition of a shell insurance company with property and casualty licenses in all states, plus the District of Columbia. With this new access to the vast majority of the U.S. market, our teams are gearing up to launch in new states throughout 2021. We recognize from experience that state expansion requires individualized rate plans, tailored state management, and methodical growth. Beyond state expansion, we can further tap into this massive market by addressing the customer's need for insurance holistically. Although we firmly believe auto is the gateway product, we've recently expanded our offering to include both homeowners and renters as additional tools to building a strong and lasting customer relationship. Protecting our customers' other investments with these additional products is a natural way for us to improve retention and grow our premium base. Our data-driven edge has been built on the significant volume of rich data fueled by our customer growth. We believe we have a powerful first-mover advantage here, now five years in the making, which only strengthens as we continue to grow. Collecting more data enhances our predictive modeling capabilities in a virtuous cycle to power our flywheel. Our proprietary telematics solution integrates driving activity data with actual claims experience and applies our machine learning capabilities to derive precise insights from the growing dataset. We collect roughly four terabytes of rich behavioral data on a daily basis directly from powerful sensors within our customers' smartphones. These sensors allow us to track driving patterns that are most relevant in determining a person's driving ability, such as hard braking, abrupt turning, and distracted driving. In the third quarter of 2020 alone, we collected an additional 1.5 billion miles of integrated driving and related claims data, increasing our total miles collected to more than 14 billion. But it is not just the number of miles or claims that matters; it is the ability to translate this data into behavioral insights with a high degree of accuracy across hundreds of phone models and then understand how these behaviors cause losses that are not explained by other variables. And when they do cause a loss, how much exactly that claim will cost. It's also the ability to improve our business by identifying underwriting and claims fraud and managing our claims cost with real-time data. Not only are we constantly monitoring and analyzing this rich data internally for the benefit of our proprietary telematics program, but given our commitment to transparency, we now share our cumulative mileage data with the world. As the COVID-19 pandemic began to unfold in early 2020, we utilized our unique access to real-time driving trends and started providing it for all to see via our website. This transparency helps our industry answer important questions, such as how much driving actually decreased as well as the ways in which driving patterns changed. Beyond a simple mileage measurement, we've been able to gain additional insights through our telematics focus about drivers' profiles to more accurately assess their true driving exposure during this unique time. Our growing dataset powers our flywheel and plays right into our second core objective, leveraging our data science expertise to enhance profitability via loss ratio and retention improvements. Through continued improvement in our telematics scoring with the industry's largest and growing dataset of behavioral data and claims experience, we are creating a risk segmentation advantage and making auto insurance fairer for consumers. As our unique approach gains traction and we amass more customers, it improves the overall seasoning of the book and drives down our direct loss ratio over time. Our quarterly results reflect our strong progress in this area. Our direct accident period loss ratio was 85% in the third quarter, a 16-point improvement from 101% in the third quarter of 2019. Meanwhile, our direct loss-adjustment expense was 10% in the third quarter, a 3-point improvement from 13% in the same prior year period. The increased predictive power of our telematics and our state management program have driven material total direct loss ratio improvements. Through the first three quarters of 2019, as we entered seven new states, only four states in our footprint had direct accident period loss ratios below 90%. Conversely, through the first three quarters of 2020, as the predictive power of our telematics improved and we matured in our existing states with only one new state launch, a total of 26 states had direct accident period loss ratios below 90% and 21 states were below 80%. We believe that our first term post-telematics underwriting loss ratio compares favorably to first term loss ratios at legacy insurance carriers. However, our total direct loss ratio cannot be compared apples-to-apples against other auto insurance players, many of whom have been in business for several decades and have less than 20% of their total premiums coming from new customers as opposed to the approximately 50% share of new customers at Root during the trailing 12 months. Moreover, it is overly simplistic to directly compare a personal auto insurance loss ratio to another personal line like homeowners or renters given massive differences in complexity of rating models, average premiums, and retention. Forget about apples-to-apples; comparing auto to home to renters loss ratios is like comparing steak to yogurt to penne. All are nice foods, but they have little else to do with each other. Finally, our business model is uniquely based upon underwriting out the highest risk drivers due to their disproportionate likelihood to be involved in an accident. In fact, our year-to-date direct loss ratio for the pre-telematics underwriting period is more than 20 points higher than the post-telematics period. Given the youth of our book and how quickly we are growing, this significantly weighs on the total loss ratio in the short term. We expect to reduce this loss by more rapid identification of high-risk driver characteristics and underwriting out the unacceptable risks as well as improving the lifetime value of customers that we bring into the marketing funnel. Similar to loss ratio, a mature portfolio will naturally experience higher retention rates at a macro level as the longer customers have been with an insurance provider, the stickier they become. For now, our current portfolio is at a disadvantage in this regard. Also related to maturity is price volatility, which, of course, can also impact retention. A young insurance company like Root naturally experiences more price volatility as we launch new states, transition to company models for underwriting variables and develop new telematics scores. As we are managing the business for its long-term potential, we believe near-term volatility is always worth absorbing to drive the right long-term decision. While select markets will experience price volatility as we open new states and address existing states, we expect price volatility across the book to reduce over time. We are actively targeting retention improvements through product offerings and features, customer engagement via proprietary techniques and customer targeting, which all can drive meaningful improvements in this metric as the portfolio scales and matures. As an example, the addition of renters and home insurance offers twofold retention improvement potential. First, through cohort mix shift as customers who desire to bundle can now shop with Root; and second, with cross-sell. Auto customers with Root who add an additional policy have shown to retain approximately 15% better than non-bundled customers at the completion of their first term. Product flexibility includes the ability to easily adjust coverage with one click and even to rejoin Root in a simple new streamlined app feature called Boomerang. Since testing began earlier this year, Boomerang has successfully reinstated more than 15,000 policies, or the equivalent of 4.7% of our auto policies in force at quarter end. This is a great example of how basic blocking and tackling within our proprietary technology stack can capture meaningful improvements in our unit economics. Our data science-led customer targeting strategies allow us to better identify potential high-frequency shoppers as well as potentially longer-retaining customers and pay the appropriate customer acquisition price to drive a target customer mix into our funnel. Furthermore, we believe claims is the most important moment of truth for our customers and a long-term driver of retention as we build a more mature book of business. Our claims experience is truly differentiated and will allow Root to stand out to customers. From the beginning, we have always built claims with technology in mind, enabling us to handle claims faster and better than any of our competitors. We continue to automate a higher percentage of our claims volume in the third quarter, allowing customers to take pictures of an accident, answer a few questions, and within 24 hours, get a complete resolution and money in their bank accounts. This has improved customer satisfaction as well as reduced both claims and claims handling costs. So back to that important question, how do we know it's all working? Adjusted gross profit, our key non-GAAP profitability measure shows how our growth, underwriting, maturation of our customer book, and capital disciplines come together to generate variable profit and mark our progress towards building a sustainably profitable business. We measure our progress towards profitability with adjusted gross profit to direct earned premium in order to best capture the contribution margin of our customer revenue. Loss ratio and customer retention are significant drivers in our profitability, and we expect these to improve over time. As our company grows and accumulates more internal loss and premium data, our data science and actuarial teams can construct more accurate predictive models. This is the flywheel at work. We are now deploying the third iteration of our internal pricing model. Due to the increase in size of our internal dataset, this iteration reflects a step-change in our approach, whereby we are able to accurately adjust more rating elements. This, in turn, allows us to provide fair and more accurate rates to our customers. Insofar as we've improved our loss cost accuracy by about 20%. Early signs are that the fourth iteration of our pricing model will produce even more substantial benefits once it is deployed. We also expect further improvement in loss adjustment expense, which we believe already is in line with industry-leading levels and will naturally experience further operating leverage as the business scales and our claims-related technology continues to improve. Third quarter gross profit was $1 million, and our adjusted gross profit was $10 million. Again, the latter of which is our key profitability measure and was substantially better than a loss of $27 million in the third quarter of 2019 due to an improvement in direct loss ratio, loss-adjustment expense ratio, and variable expenses, net of reinsurance ceding commissions. Adjusted gross profit fully incorporates the work we've done on our reinsurance program, a critical efficiency lever for Root. We set out in 2020 to land the right reinsurance structure for our business today. We're proud to say that beginning July 1 this year, we transfer 70% of our premiums and related losses to reinsurers, while also gaining a 25% commission on written premium to offset some of our upfront and ongoing costs. This reinsurance program is exactly what we need to allow our equity capital to drive growth and build a deeper moat around our technological advantage, and that's what ultimately matters. Reinsurance has implications for GAAP revenue, and our new reinsurance program will cause a reduction in GAAP revenue versus prior quarters. This is why we use direct earned premium as our primary top line metric for the business. It removes the volatility of our reinsurance program and really captures the revenues received from our customers. Our third core objective is to optimize customer acquisition via direct marketing and a strong user experience. The structural tailwinds at play, whereby consumers are migrating from agency to direct channels and driving accelerated growth for direct-to-consumer brands leveraging mobile plays right into our hands. We acquired 75% of our customers through direct mobile channels, driven by our unique onboarding experience that can be completed as fast as 47 seconds, which is highly differentiated and not easily replicated with legacy systems. This is our primary distribution source, our lifeblood. Yet this is only one part of an equation driving a long-term cost of acquisition advantage versus the industry. Our data-science-led marketing strategy is another vital part of this equation and inherent to the data DNA of Root. We use data-driven targeting strategies across our marketing channels. Our digital distribution model also allows us to be more agile when we see opportunity present itself or, in some cases, slow with caution. At the onset of COVID-19 in March, we reduced our performance marketing spend, maintaining and monitoring it with a watchful eye during the second quarter. In the third quarter, we resumed pre-COVID levels of marketing spend as we saw signs that the pandemic was accelerating structural shifts in auto insurance that support Root's direct-to-consumer and telematics strategies. Our third and fourth quarter marketing strategies focus on a test and invest approach as we set the stage for our push into more states in 2021. We are targeting a build-out of new channels such as streaming video to establish a set of diverse acquisition channels that work together. Additionally, we launched a brand-focused campaign in the lead up to the presidential election in an effort to make Root more of a household name. The campaign features NASCAR driver and advocate Bubba Wallace. This work highlights the importance of bold progress and reflects Root's culture and commitment to fair pricing based on driving performance rather than demographics. We are extremely pleased with the positive exposure Root has garnered from this campaign thus far. At Root, we are always looking for new ways to solve a problem by leveraging technology. While we have a team building a differentiated customer acquisition funnel, the question was posed in the quarter, 'What about the customers that we underwrite out? Could we help them find insurance elsewhere and potentially launch a new revenue stream for the company?' In less than a month, we launched a program to redirect customers with their permission to other carriers. And in so doing, we're able to offset 3% of our customer acquisition spend in the month of October, yet another example of what is possible with the nimble technology infrastructure we have at Root. In the near term, we expect the amplified brand spend will take time to drive acquisition efficiency and could result in temporarily elevated customer acquisition cost levels. However, the longer-term benefit will far outweigh any near-term pressure, particularly as we expand our footprint nationally. I'll close with some thoughts on how we're thinking about the business in year-end 2020 and beyond, and then we will welcome your questions. We will continue to prioritize growth because we believe our business only gets better as more data flows in. Which turns our flywheel and helps to unlock the full potential of our business model. More data allows us to deploy even more advanced algorithms, which allows us to further differentiate our product from the rest of the market while becoming an even better underwriter. As our flywheel continues to develop, we expect our operational scale will realize economies of scale and grow margins. To be clear, we base all our strategic decision-making on building a business for long-term sustainable growth and profitability. The near-term targets that we're establishing today demonstrate that we are on track in delivering this framework. While 2020 has been a year that few could have expected and no one will soon forget, Root will continue to deliver strong financial results. Our current expectations for the full year of 2020 are as follows: direct earned premium of $595 million to $600 million, adjusted gross profit of $14 million to $16 million. Root is a long-term focused company and management team. We're excited about our 2020 accomplishments, but even more about what is to come. We'll share our 2021 outlook when we report our Q4 and FY 2020 results in the new year. Until then, we look forward to pounding the virtual pavement actively engaging with our new and prospective investors and keeping you updated on our progress as we continue on this exciting journey together.

Operator, please open up the lines for questions.

My first question is on policies in force. That obviously slowed down a little bit sequentially, and you guys have talked about just kind of pulling back on advertising, just given the uncertainty surrounding COVID. In your prepared remarks, you pointed to picking up on the ad side from here as there's more certainty in the environment. So can you give us a sense of how you think policies in force growth within auto could trend in the fourth quarter and then also some initial views on 2021?

Elyse, this is Dan. Thank you so much for the question, and nice to speak with you again. I think what you see on policies in force is a result of the way we've managed marketing spend this year. And if you think about it in the first quarter of the year, we spent about $36 million in sales and marketing. And we dropped down to $18 million in the second quarter. And now we're showing $37 million in sales and marketing spend for the third quarter. And that dip, as you said, was really related to COVID. The uncertainty in mid-March of what the world was going to look like, how consumers were going to treat their auto insurance. And as we then had a good grasp on where things were going, we ramped back up the marketing spend in the third quarter. And what I think you'll start to see, as we fine-tune our algorithms and continue building, is a progression on that in policies in force that won't necessarily show up in the fourth quarter, but will show up as we grow towards 2021. We're not providing guidance today on 2021. But as we show in the shareholder letter and as we just talked about, we do envision now that our marketing spend is back to the pre-COVID levels, and we're excited to build upon that.

Okay, that's helpful. My second question is regarding your focus on improving retention. As a newer company, your percentage of new business is significantly higher compared to established carriers. Could you share your targets or insights on how we should view the current retention rate and its potential trends over time? Specifically, how might the retention portfolio contribute to improving profitability as it becomes a larger part of your overall business?

Absolutely, this is Alex. You made a great point about retention. In the near term, our retention rates will likely be lower compared to our older, more established business. This is mainly because new business tends to have a higher churn rate as it exits our book more quickly than those customers who have been with us for a while. Typically, we see significant improvements in retention as time progresses. Additionally, our expansion into many new states has required us to experiment and refine our pricing strategy. The primary reason people change their auto insurance is due to price changes, and when we enter a new state, we initially lack substantial data. Therefore, our first pricing plan is somewhat basic. As we begin to acquire customers and gather claims data in that state, we adjust our pricing, which can lead to a temporary increase in churn. This phenomenon has been observed in several states. Thus, the blended retention figure may not accurately reflect the long-term trends we expect the business to achieve, which we believe will eventually align with the industry average. We do underwrite against the worst drivers and anticipate maintaining this approach as a competitive advantage. Our retention strategy is not solely focused on maintaining customers for its own sake; we also carefully analyze our customer segments. Over the past year, we have shifted our focus towards acquiring higher-retaining customers in states where we've had success modifying our rating plans. This effort has helped mitigate previous rate shocks in our customer base, significantly improving retention through targeted marketing initiatives. Moreover, our product features are also enhancing retention. Dan mentioned the Boomerang feature, which has exceeded 15,000 policies, allowing customers to return to us. We are also beginning to explore cross-selling, particularly in states where we offer renters and home insurance. We've observed a notable decrease in churn when customers purchase multiple policies from us, and we anticipate even greater benefits as we scale this effort. So while we might see some short-term fluctuations as we continue expanding into new states, our long-term outlook remains positive, with expectations for substantial retention growth supported by our data.

Just a couple of questions on something you just talked about, maybe, Alex. On the plan to launch in other states, can you just help us understand the pace of rollout that you guys plan on doing? How does that compare to your thinking just a few months ago, pre-acquisition, does this basically dramatically accelerate your plans, and therefore, could have a pretty material impact on your sales and marketing? Would be great to hear how you're thinking about sales and marketing next year, although not necessarily asking for guidance.

Those are excellent questions. Regarding the acquisition of the license for all states where we currently do not operate, we are moving forward with our expansion plans. We intend to accelerate our state rollout through this acquisition and will be launching in new states continuously. This represents a significant growth opportunity that was previously unavailable, as we had to approach each state individually. However, we still need to complete our product filings, and we are actively working on that. You can expect us to add more states throughout 2021, which will be a major growth factor for the business. We will invest in marketing for these new regions and will monitor them closely to ensure that our models perform well. Our initial rating plans will be implemented in these markets, and we will be prepared to make adjustments quickly. The pace of expansion differs by state; some states have lengthy regulatory approval processes, while others are faster. This will create a varied rollout throughout the year.

And then just quickly, what steps are you taking to identify some of these policyholders with high propensity to shop around? That's something you also talked about earlier.

That's a great question. We identify customers who are likely to shop upfront in various ways. One key method is through the marketing channels we use, specifically digital marketing, which provides us with extensive data even before we choose to show an advertisement to a particular consumer. For example, we can know if an individual has prior coverage, whether they own a home, their age, and how often they've switched insurance in the past year. We gather this data on many customers before we even market to them. We use a data science model that optimizes for the lifetime value of a customer across our digital channels. This model helps us determine if a customer is likely to remain loyal, allowing us to accept a higher acquisition cost for them. Conversely, if a customer seems likely to leave quickly, we adjust our strategies to limit acquisition costs and lower the first term loss ratio, which we are implementing in our current rating plans. Throughout the customer's journey, we continually update our expectations using this model to predict how long they will stay with us, which helps us effectively manage these customers. It's a combination of targeted marketing and careful consideration in our rating plans regarding their likelihood to shop. For customers likely to stay longer, we might tolerate a higher first term loss ratio compared to those expected to churn quickly, for whom we would require a lower loss ratio.

I thought your graph showing the loss ratio distribution by state in your annual letter was quite revealing. You mentioned it briefly in your commentary as well. Expanding into new states poses challenges regarding loss ratios for any company, and it could be particularly difficult for a company like yours that is experimenting with mobile technology and distracted driving pricing algorithms. This will likely put pressure on the loss ratio. My question is how you balance your plan for national expansion and the resulting pressure on the loss ratio against focusing on improving loss ratios in your current markets.

Mike, this is Dan. Thank you so much for the question. And you're right. We decided to share that data on the loss ratio because we thought it was very compelling. And what you're seeing there is a function of time, more than anything else. As the business has time for policyholders to mature in each state, you can see the loss ratio performance trend really be quite strong. So then the question around state expansion, Mike, is a good one because you're right, if we just enter a state and prioritize expansion, without prioritizing the overall portfolio, it will put pressure on the loss ratio. So for us at Root, it's not when we enter a state. It's how we enter the state. And we've done that in a couple of different ways. First is we've really fortified our team. We've invested a lot on the team side to bring in professionals who have really strong state management experience in some of the states where we are not today. We brought in a new VP of product and pricing in late August with nearly three decades in the industry. We brought in a new head of claims, who understands the pitfalls that you can find when you enter a state and we also have the learnings from the 30 states that we've already entered, that we can apply in large part to the states that we are not yet writing business in. So we are, frankly, very bullish on not just turning or flipping the switch and saying, 'Let's go into 48, 49 states immediately', but let's do it the right way. And I think that what you'll see as we enter 2021 is we'll be very thoughtful about that. We'll be prudent stewards of capital. We'll be testing because that's what we do at Root. We're very data-driven. We're very test-oriented. And so you'll see us do that pretty consistently. And the bigger difference versus, say, 2017, 2018 or 2019 is we have this very nice installed base of in-force premium in states that are already maturing, as you can see on Page 12 of the shareholder letter, and are going to continue to mature next year. And that's very powerful for us. Our largest state is Texas, Mike. And just look, in Texas, in 2020, we took base rates down. And year-to-date, we've seen loss ratio improvement of about 36 points on 59% earned premium growth. That's quite significant when you look year-over-year at a state which is our #1 state and really shows the benefits of maturity. So you've got that installed base to support the loss ratio as we dip our toe in some of the state expansion.

Okay, Dan, that was helpful. For my second question, which is somewhat related to your previous answer, could you clarify your comments about the first-term post-telematics underwriting loss ratios in comparison to legacy players? Can you provide some specific numbers, Dan?

Yes. We look at some of the publicly available data in some of the reports and see a first term loss ratio estimated to be in the mid-80s, somewhere in there from the large direct carriers. For us, that story is complicated. By just comparing apples-to-apples for two reasons. First, as we identify, part of our business model is underwriting out the poor drivers during that first term. And so that's where it's very hard to compare our loss ratio and retention to legacy industry players and what we identify today is that the loss ratio delta between the pre-telematics period and the post-telematics period of that first term is around 20 points. So quite significant. The second thing is that today, for us, about half of our premium comes from new writings versus renewal customers. And obviously, as you know, Mike, renewal customers have a lower loss ratio, not just at Root, but throughout the industry. And that premium difference is north of 80% for renewal premium at legacy carriers. So you also see what happens at some of the legacy carriers is absorbing that first-term loss ratio with a bigger base of renewal customers. So on a blended basis, when you're comparing direct loss ratio, it's very hard to look apples-to-apples. At Root, again, a function of time. We are trending that way as the book matures as we have a larger percentage of our premium come from renewal customers, we expect to trend in that direction from the overall blended rate. But we feel very good about where our first term loss ratio is, particularly in that post-telematics phase.

I have two questions, one regarding customer acquisition cost and the other about retention. Can you provide any insights on the new ad campaign, particularly the TV campaign? Also, how did customer acquisition cost perform in the third quarter, considering that sales and marketing aligned with our plan? Additionally, Dan, you mentioned the price increases and the third generation of your pricing model. How do these pricing decisions relate to retention? Is there anything you can do during the customer acquisition phase or the initial test-drive period to enhance the retention rate, perhaps by being more selective in acquisitions or by adjusting pricing during the test-drive period to improve retention? Any insights on that would be appreciated.

Thanks, Ross. This is our first earnings call, and I know you'll want to ask about CAC, which I'm looking forward to discussing regularly. CAC met our expectations for the quarter. As mentioned in our shareholder letter, we anticipate higher levels not only in the third quarter but also in the fourth quarter, as we invest to support our state expansion plans and enhance our brand. We believe the long-term benefits of this spending will significantly outweigh any short-term pressures, especially as we grow our national presence. We expect overall sales and marketing to increase slightly from the third quarter as we build on the campaign launch associated with Bubba Wallace and the upcoming election. We aim to leverage the momentum from national attention surrounding our IPO and the broader campaign to expand our reach. The Bubba Wallace campaign was very successful, generating approximately 13 million organic impressions on social media and over 400 million earned media impressions. This success powerfully represented our brand. Authenticity is a key aspect we aim for with Root, and Bubba Wallace's story exemplifies that authenticity. Initially, we did not plan to enter the endorsement space, but we saw a significant opportunity and are pleased with the campaign's outcomes so far. We look forward to continuing our partnership with Bubba. Now I'll hand it over to Alex for your second question.

Yes. The second question focused on our pricing decisions, particularly related to retention. We set a target for loss ratio and retention by customer segment in each state. After launching new plans, we monitor them for a period to gather enough claims data, mainly frequency data, to evaluate their effectiveness. We then adjust rates to meet our targets, which may require several iterations. Once we achieve our goals, we see retention stabilize and improve. To optimize retention, we try to anticipate the retention of customer segments in advance. If we believe a customer will stay with us long-term, we may offer them a lower rate since their lifetime value will be significant. Regarding pricing changes during the telematics phase to reduce churn, we discovered that quicker assessments of whether someone is a good or bad driver, which we refer to as the test-drive period, lead to better retention and customer satisfaction. Our predictive models have improved significantly and can make these determinations much faster than before, allowing us to offer telematics-based rates to customers more quickly. While this is not applicable for everyone, we can swiftly identify many drivers. We're also leveraging the motion chip in smartphones for high-accuracy readings, which has positively impacted our ability to telematically rate customers, resulting in improved conversion and retention.

Alex, I would like to follow up on your previous answer and ask if you could elaborate on that a bit more. Specifically, as the data flywheel progresses and you become able to identify the risk spectrum tails more quickly, how soon can that translate into results? You mentioned a 2 to 4-week test-drive period in your materials. Could that timeframe improve significantly in a couple of years, possibly down to a 1-week test-drive period, or do you anticipate a more gradual improvement?

Yes. That's a great question. So where it ends up, we don't know, but we can certainly get within a 1-week test-drive period. And in fact, we can get to the point with certain depending on where you are in the tails. On either very good or very bad, you can actually get some pretty accurate readings within just a few days often. And so you can do that very quickly. In terms of how fast we ship those; once we have that model in play, it's very, very easy to ship. So it goes into R&D. The R&D time and data science can take anywhere from a couple of weeks to a couple of months, depending on the nature of the data science problem. And then once we've identified that, then shipping it is actually pretty trivial. It just lives on our servers. And so we don't have to do an app update even to actually do that. So we can actually just almost real-time ship that. And so those are things that we deploy very quickly. In terms of showing up in the financial results, and of course, those premiums would have to earn through our book. And so it will take more time, probably a period of 3 to 6 months to maybe a year, to really show entirely through just because of the nature of the business that we acquire customer upfront and then obviously earn that revenue and pay those losses over time.

Great. My other question is about your three core objectives. Thank you for outlining these and providing additional data. As we consider the first two objectives, which focus on growth and profitability, could you help us understand how Root balances these aspects? Specifically, how does the company approach the need for growth to capture market share while also demonstrating profitability? I'm particularly interested in how this relates to other stakeholders, including investors on this call, reinsurers, and the potential need to raise capital in the future.

Yes, Matt, thanks for the question. This is Dan. It's a constant balance of growth and profitability for us, and we have the ability to focus on long-term execution. And that's really what we're focused on. We feel like we have built out three important moats around technology, around the size and quality of our customer base, and the capital that we have in place to now go execute on our plan. And that really is our singular focus. So as we gain the benefits to the next iteration of the pricing model or from refining the pre-telematics period that Alex just talked about. We can really choose whether to take some of those benefits to growth or to profitability. And for us, it remains a balance. So Mike's question earlier about state expansion is a good one. As we decide to ramp up some of our state, which is the right thing to do for the long-term to expand that footprint. We'll also be conscious about pressure on the loss ratio. So we don't want that state expansion to drive the loss ratio really hot. So part of what we will do is take the overall benefits of our telematics and our pricing model and take a bit more of those benefits into profitability. So that's sort of what I'm trying to suggest that it's a constant balance for us, really focused on executing the overall plan.

A couple of quick numbers questions, if I could. First, can you maybe tell us what the new writings were in the quarter? And then also if you could maybe give us, I think, similar to the S-1 disclosures, the renewal book loss ratio?

Yes, I will address that. We do not disclose new writings due to competitive sensitivity. Instead, we direct your attention to in-force premium, direct earned premium, and policies in force for an overall view of the top line. We have chosen to focus on these metrics. Regarding the loss ratio, we’re concentrating on comparisons between the pre-telematics and post-telematics periods and demonstrating the significant benefits of loss ratio distribution by state for the first nine months of this year compared to fiscal year 2019. This approach allows us to present the loss ratio in a straightforward manner, emphasizing the accident period direct loss ratios without differentiating between new and renewal policies.

Okay. And then if we do look at the loss ratio for year-to-date versus a year ago. Is there a way for us to think about the COVID impact there if we try to adjust that out to have a more apples-to-apples comparison?

Yes. I think COVID for us was a story of a couple of different chapters, and we're in the middle of chapter 3. Chapter 1 was March through May, and we saw probably about 15 to 17 points of loss ratio benefit from the periods of March through May due to COVID and due to less driving and obviously reduced claims frequency. And Chapter 2 was the period in the third quarter. We didn't really see the extended COVID impact or the size of some other carriers, and we think that's due to our portfolio mix. California, Massachusetts, New York, New Jersey, states with high commuting in normal times and states or cities and have higher levels of COVID restrictions in place right now than other parts of the United States, those are not in our footprint. So for us, we didn't see quite the COVID impact in the third quarter, really any significant impact from COVID in the third quarter, like some other carriers may have seen. Chapter 3 is what's happening now? We saw some signs of reduced driving over Thanksgiving. We were frankly glad to see people taking COVID seriously, and we think that's obviously important with what the country is dealing with right now. It's just too soon to know for November, and I don't want to make assumptions on the rest of the year. But think of that as chapter 3, that I'm sure we'll come back and talk to you about in the near future. We're obviously watching it extremely carefully.

Yes. We saw in the shareholder letter that the launch of the next version of the pricing algorithm, it seems like it's poised for the second half of '21. Can you talk to us about how the pricing algorithm will come through in the financials? Is this for new business? Is it for retained business as well? Should we expect a stair-step improvement in underwriting? Or as you talk about the acquisition of customers on the algorithm and then the time delay and the earn-through of that pricing for them?

Absolutely. That's a great question. When we launch and roll out the new algorithms, you can expect both the third generation of our traditional models and the fourth generation of our UBI models to enter the market, likely in the first half of 2021. This will impact our operations positively, especially in states where we have confidence in our loss ratios and projections. As we improve our segmentation, we can lower base rates while maintaining or exceeding our current loss ratio performance. For example, we've reduced rates in Texas and are seeing lower loss ratios and strong growth. In states where our unit economics are solid, you'll see rapid growth and reduced customer acquisition costs. However, in states where our loss ratios are not where we want them to be, we might not reduce base rates as much, or we might even increase them. This process takes longer to reflect in the numbers because it is tied to loss ratios. If we hypothetically launch our model in the first half of Q1, it will affect new business rather than renewals. Over time, new policies will be issued under the new rating plans, and it will take about six months for those policies to fully earn through from the new business segment of our portfolio. It will take even longer for all policies to transition to the new rating plans. Therefore, the impact on loss ratios may take longer to manifest, with initial results appearing within six months and most results becoming evident within a year. Yes, that's a good question. With each, we've now lived through lots of different iOS permissions throughout the history of Root. And typically, what we see is that each one requires us to do something slightly different. So it's each basically versioning of the operating systems. We're actually going back and looking at the way the code is written and making sure that we can still get the data we need. As part of that, we've actually even developed motion-only models and where we really just use the gyroscope and the motion chip. So that we don't even need the GPS chip because for a while their location data was what was really, really targeted. In terms of advertisement, I think because a lot of what we'll be doing too will be on mobile web. I think you will see a minimal impact to it, but it's certainly something we're monitoring.

I have a question. It seems unlikely that we will receive the renewal loss ratio. However, I'm curious about the flat PIF growth in the second quarter, which slightly decreased in the third quarter when looking at the sequential basis. It appears that there won't be significant growth in the fourth quarter compared to the third. I’m interested in understanding what impact the 85% direct loss ratio had regarding less new business drag or new customers, especially since PIF growth has slowed down, and how we should consider that for the fourth quarter and into the first half of next year.

Thanks, David, for the question. Yes, look, our renewal loss ratio compared to our new business loss ratio is in line with what we presented in the S-1; there's no material deviation or anything like that. And so we expect that, that will continue. And obviously, we're showing you that as a percentage of our overall premium. Renewal percentage is remaining, to your point, roughly the same or slightly higher. I don't think across the overall base it's not having a material impact on our loss ratio, nor do I think it will through the end of the year. I do want to be clear on one thing. In terms of our forecasted guidance range, we are forecasting growth in direct earned premium through the end of the year. And obviously, bringing that forward from about $450 million at the end of the third quarter to a range of $595 million to $600 million as we come through the fourth quarter. But what we're describing to you is the ramping backup of marketing spend, which really, as I said, started in the third quarter. Starting to pay dividends a little bit in the fourth quarter, but really more in 2021. And that's how we expect to go forward.

Alex, I wanted to ask about the targeted marketing you mentioned and how it might help improve loss ratios or at least benefit the initial customers you are acquiring. What other strategies are you implementing to enhance that first-term loss ratio?

Certainly. The primary focus for us in enhancing the first term loss ratio is our approach to managing it on a state-by-state basis. In many states, we are pleased with our current first term loss ratios. However, in a few states, we see room for improvement. In those instances, we are prepared to adjust base rates as necessary. Additionally, ongoing advancements in segmentation will consistently drive improvements in our loss ratio. We are currently expanding our data collection and rapidly retraining our models. For example, our UBI 4.0 model shows significant predictive capabilities compared to UBI 3.0. This positions us to make strategic choices about our growth trajectory. While we plan for considerable growth in 2021, we have the option to either accelerate this growth or focus on reducing loss ratios to enhance our bottom line. Our commitment to investing in Root and achieving exceptional pricing through data and technology remains strong. We are also exploring new data sources beyond just mobile data. As we progress in this area, you will notice ongoing improvements in segmentation, which will allow us to lower loss ratios while still achieving growth.

I want to touch on the competitive landscape today. Some competitors are implementing mileage-based insurance pricing, where customers pay per mile. We've also observed recent collaborations, like Ford partnering with Verisk and GM partnering with American Family. Can you discuss how the competition is evolving? It seems there are new opportunities within telematics that differ from your go-to-market strategy. Is this change happening more rapidly than before? Any insights on how you perceive the current competitive environment regarding telematics and the use of data in insurance pricing would be appreciated.

Absolutely. Those are great questions. In terms of the competitive landscape, data is going to play a critical role. The pace of change with OEMs is not new; these have existed for over a decade with initiatives like the OnStar program and efforts to sell insurance through connected vehicles. This isn’t a change but rather another chapter in an ongoing trend. There are significant barriers to gaining traction, including the fact that most consumers don’t own just one type of vehicle, which makes it challenging to have a single insurance carrier or product per vehicle. Bundling is important for many consumers as well. Additionally, the technical challenges with vehicle data are considerable, as there’s no uniformity or quality across the board. That said, we believe that over the long term, this data will continue to expand. This includes not just vehicle data but also behavioral data gathered from smartphones that isn't directly related to vehicles. At Root, we have contracts for approximately 47% of connected vehicle data by market share, and we're actively analyzing this data. Some of it is valuable while some of it isn't, but the focus is on integrating this data effectively to benefit consumers. The key question is how to seamlessly incorporate vehicle data into a user-friendly product. This data isn’t new; what’s innovative is finding effective ways to leverage it to address consumer needs. For example, we have a feature called skip drive that allows consumers to get a telematics rate quote instantly, making it a unique offering in the current market. As for other aspects of the competitive landscape, per-mile billing has been around for a while and appeals to a niche audience. However, our tests indicate that most consumers prefer stable insurance bills rather than fluctuating ones. We have been successful in accurately predicting how many miles a customer will drive based on their test drives, so surprises are rare in that area. Moving forward, we expect rapid advancements in vehicle technologies and an influx of data from various sources such as smartphones, vehicles, and wearables. The insurance companies that will thrive will be those that are proficient in technology and data science, enabling them to use this data strategically to better serve their customers. This is the path that Root is on. Thank you for your excellent question.

I'd like to add something quickly. Alex is absolutely correct about how we are approaching our business. The advantages I mentioned earlier regarding our technology, the size of our customer base, and our diverse capital really matter. Our premium is not dependent on a single state or type of customer, which is significant. The maturity of the state is important, as we noted in the shareholder letter, and our telematics effectively addresses distracted driving. It's crucial to highlight that one in four accidents in the United States is linked to distracted driving, and integrated car data cannot accurately identify distracted driving from the data available. These are critical issues that the industry will confront as we continue to grow the Root brand and community while striving to offer a fair and more personalized customer experience.

We just would like to reiterate our thanks for everyone joining today. We are thrilled to talk to you guys today and continue to update you on the business and the company results. So we will talk with you all in about 90 days again. Thank you.

Operator, this concludes today's conference call. Thank you for participating. You may now disconnect.