Trupanion – a long-term compounder!


Trupanion provides medical insurance for dogs and cats. The company issued its first policy 20 years ago in British Columbia, Canada and entered the US market 10 years ago. TRUP quickly rose to become the second largest insurer in the space by radically changing the business model.

  • TRUP has been growing at +30%+ for the past few years, capturing 40% of the market growth, and is projected to keep similar growth rates for the next 5 years.
  • TRUP has a 70%+ ROC and 80%+ reinvestment rate (that is not a typo).
  • TRUP has high recurring revenues (98.6% monthly retention rate[1]) and trading at 30x TTM FCF, but 9x 2020 sustainable FCF, at which point they can still grow at double digits.


Pet owners are often surprised by the cost of veterinary care and can be financially unprepared if their beloved pets are injured or become ill. The cost of medical treatments for pets have become burdensome[2] over time due to increasingly advanced veterinary care. An example area of improvement has been technological diagnostics (CAT scans, MRIs, etc.), which increase the frequency of medical interventions (prior to this many medical issues were seldom diagnosed since pets cannot communicate their complaints).

Some statistics to sum up the industry[3]:

  • Total pet expenditure reached $69 billion last year in the US alone.
  • Half of US households own a dog and 38% own a cat
  • 35% of millennials own a pet, a much higher percentage than last generations. This target demographic is likely to spend disproportionally more on their pet.
  • 53% of dogs and 58% of cats sleep in bed with owners forging strong emotional attachments.

The first pet insurance was issued in 1982 by VPI (now a subsidiary of Nationwide), the largest pet insurance in North America. Most pet insurance companies (except TRUP) replicated the legacy insurance model for pets. The legacy insurance model is extremely complicated and dysfunctional, lacks transparency for customers, and always benefited the insurance companies. That is why veterinarians—the gatekeepers of the pet industry—were always resistant to any form of insurance, and winning them is the key to success. Since 1982, the insurance penetration rate in North America remains under 1%, a drop in the bucket compared to Sweden 40%, UK 25%, Western Europe 5-15%.

Vets run local owner-operated businesses[4] and usually have a great relationship with pet owners. Vets are more idealistic than their medical counterparts; enrolling and graduating veterinary school is equally as difficult as dental or medical school. However, in the US, vets on average only earn 75k a year, 5x less than doctors or dentists. This goes to show that there is an idealistic element to pursuing the veterinarian profession that encompasses genuine concern for animals.

TRUP generates 85% of their growth from vets, as opposed to 37% for the industry; if a vet is going to recommend a product, they are going to recommend TRUP – see the table below for calculation. But why are they recommending TRUP and not even getting paid for it? The short answer is because vets are well incentivized financially and morally, TRUP is a superior product than its precursors in every aspect. Insurance helps lower the financial burden, allowing vets to perform all necessary treatments. TRUP policyholders spend twice as much at the vet compared to a client without insurance. However, financial incentive does not explain the full story, otherwise vets would have recommended other companies as well. TRUP showed to the scene 9 years ago and is the second largest player, eating 40% of the industry growth.

Number of pets New pets last year Vet recommendation % Vet recommendation
Industry  1,600,000  171,429 37%  63,429
TRUP  343,000  72,921 85%  61,983


Why would vets and pet owners choose TRUP?

Because TRUP has a superior product.

  1. TRUP pays 90% of the costs of veterinary treatment if your pet becomes sick or injured (that includes cost of medication). Most competitors will cover at most 70% depending on the treatment, some do provide up to 90% but limit what they cover.
  2. TRUP has unlimited lifetime coverage, and no limit on the amount covered for any one claim. Competitors have not been able to match that.
  3. TRUP has a 5-day waiting period for injuries and 30-day for getting sick whereas competitors have 6 months to a year; that is the equivalent of a human decade in a pet’s life.
  4. They offer 24/7 customer care and their customer care is top notch. Their phone sales conversion rate is 40%.
  5. TRUP allows you to pick your own deductible up to $1,000 and more importantly the deductible is for the remaining of the pet’s life per injury/illnesses. Competitors have only a couple of options for deductibles ($500 max) and most renew annually. This is an important distinction because assuming your pet needs an intervention each year, under the competitors plan the pet owner will pay deductible x number of years. Pets generally have a few common injuries/illnesses. Assuming a pet suffers from diabetes[5], under TRUP’s plan you only pay your deductible once and you are covered every time the pet needs diabetes treatment.
  6. TRUP is the lowest cost provider in the market; they spend 70 cents on every dollar of premium in claims whereas competitors spend 50-60 cents.

Yes, it is possible to find lower price insurance, but it will cover less, so in the long run pet owners will pay more. Vets know this and pet owners are smart enough to figure that out. Most insurance renews every year whereas TRUP is on a monthly basis and can be canceled any time. Common techniques employed by other insurance companies, but not TRUP:

  1. Most insurance companies increase their price based on age, number of claims, etc. Age inflation alone can double your premium in 7 years – that is a 10% CAGR. Once you sign up for TRUP (depending on age), your premium will only increase based on your local vet’s prices. This is more controlled, transparent, in line with the customers and hovers around 5%-6%. In 7 years’ time, the difference between 5% and 10% is huge. The initial premiums are all teaser rates that could grow tremendously and cost owners more in the long run, in addition to paying out far less than TRUP.
  2. Insurance companies will reimburse you according to a benefit schedule, not the bill. So if a condition “should” cost $2,000 based on insurance calculation and your bill is higher, they will only cover according to their “price”.
  3. Insurance companies cover only menu-specific treatments which owners, way over their head at this point, are forced to choose. They might exclude medical conditions that were previously covered; these conditions are more likely to recur and would be more costly for the company.
  4. Insurance companies need to approve you for your coverage (financial standing of owner, pet condition, etc.). TRUP does not because it is a monthly-based payment (98.6% retention rate).

Pet Insurance worldwide

As we discussed above, pet insurance is widespread in countries like Sweden, UK, Australia; in fact, Sweden was the first country to underwrite insurance for a dog in 1924. In 1890, Sweden was the first to underwrite animal insurance on livestock and other farm animals; now 40% of pets in Sweden are insured. However, factors such as household income, veterinary cost, etc. are quite different worldwide. The table below summarizes some interesting statistics (all figures in USD):


  • TRUP has a long runaway for growth; pet insurance in the US is a fraction compared UK, Australia and Sweden.
  • Most pet insurance in mature markets is underwritten by one or two big players. As the US market matures, it will force players out of the market and force them to merge. More importantly, fewer than 2-3 will be able to underwrite profitably and competitively in the long run.
  • Annual premiums in the US are not egregious, they are comparable to Australia and Sweden. However, premium growth expectations would be more moderate than the previous years.
  • TRUP paying 70 cents of claims out of every dollar is not abnormal, only in the US.
    • UK firms pay significantly more in claims than US firms – further proof of a dysfunctional insurance system in the US.
    • The average claim in the UK is almost twice as high as in the US
    • Agria Djurforsakring, one of the largest pet insurer in Sweden, paid approximately 70 cents on each dollar of premium.
  • Management’s projected profit margins are reasonable:
    • Agria generates 10% profit margins[6][7].
    • Hollard, which underwrites most of the Australian market, claimed that they were generating 8% profit margins.
    • Allianz’ PetPlan, insures a third of all pets in UK, reported a loss throughout their operations[8], but were generating 11% profit margins on their mature markets.
  • US households are twice as likely to have a pet.
  • UK and Australia are experiencing similar increase in veterinary cost; most insurers are scrambling to adjust prices.
  • Pet insurers in Australia have annual limits of $8,000 (standard) – $15,000 (premium) and co-pay varies from 75% to 80%. Important to note costly medical intervention exclusions are common such as; cancer treatment, dysplasia, diagnostics, etc.
    • Pet insurers in the UK have similar annual limits, co-pays and exclusions to Australia. We were unable to find much information on Sweden.

Trupanion Express

Legacy insurance works in a similar way as other medical insurances. You pay the vet out of pocket for a treatment, file a claim and wait a few weeks from the insurance company to send you a check. Trupanion pays claims very rapidly, 60% the same day, 80% within 5 days. TRUP rolled out Trupanion Express, a free to use software for vets that allows immediate payment, under 5 minutes before the owner pays a cent. Trupanion Express processes around 22.3% of the claims (from none four years ago) and is growing 13 percentage points faster than TRUP’s top line. It is a matter of time before Express will be rolled out throughout most of TRUP’s network. Express can be easily replicated from a technological point of view, but not from an executional one; competitors will be unwilling to since it highlights their flaws and opaque system.

How are they able to pay 90% of the cost?

Because they do not have frictional costs like other insurers. A typical insurance company will have third party underwriters charging 15% of the premium and 3%-5% for lead generation which totals to 18%-20% extra costs. Furthermore, recent acquisitions in the space put a strain on keeping or expanding margins (cutting corners). TRUP is the lowest cost operator.

Pet Owners

Why do people buy pets?

“People are happier and healthier in the presence of animals. Scientifically-documented benefits… include decreased blood pressure, reduced anxiety and enhanced feelings of well-being” – Human Animal Bond Research Initiative Foundation.

People buy pets for a variety of reasons, but at the core of it is the inherent human need for companionship. Most dog owners can attest to the unconditional love that a dog provides. Emotional connection is the biggest driver behind pet insurance (excluding the effects of income). Pet owners do not want to be in a position where financial condition prevents them from providing a necessary medical intervention for their pet. At the core are two trends: the first being baby boomers who are launching their kids into the wild and replacing them with domesticated animals. In almost all pet spending categories, spending declines as a person reaches 55 years of age – but then pet spending peaks between the ages of 55 and 64. The second trend is millennials (people born between 1985-2010) who are probably the first generation to grow up thinking that pets are more like humans than wild animals, a process called humanization of pets.

Is it affordable?

Affordability is ultimately a function of the value provided. The value here is a healthy pet and a peace of mind. $55 a month may sound steep, but surprisingly money doesn’t seem to be a problem. 65% of TRUP’s customers are from mid to low income. Pets require frequent assistance; 14% of their clients have an invoice every month and more and more owners are discovering that medical insurance is a necessity. Speaking of necessity, here are a few stats on how much pet owners are willing to spend on pets:

  • 10% of dogs and cats have designer clothes
  • 5 million dog owners stay with their dog in pet friendly hotels
  • 6 million dog owners frequent dog-friendly restaurants
  • 8 million dogs go to work with their owner

Sales and capital allocation

How do they approach vets?

TRUP’s pet acquisition cost is largely comprised of their salesforce which they call territory partners (around 104 in total). Following the earlier Coca-Cola distribution system, each territory partner is responsible for vet hospitals in a certain area. They meet with each vet every 2 months to build a relationship. Their job is to show the vets TRUP’s product and leave it to the vets to recommend it. TRUP pays their territory partners a fixed fee per pet in the area that cover. This cost is much smaller than having a broker model (which usually takes 15% of the premium – hint: Geico). The partners are trained rigorously; TRUP has an in-housed university and each partner receives constant training. If a territory partner does not do his mandatory rounds upper management can easily spot it in the data.

A timeline of their sales process

It takes time for the vet to trust the product and provide full support. One hospital in Canada took about 3 years to reach 1% penetration rate. After 10 years in the hospital, 5% of new enrolled pets are signing up for TRUP – that’s a 5% going forward penetration rate at 0 cost. TRUP is active[9] in 8,000 hospitals out 28,000 in the US, and the majority of those they entered in the past 4 years. Growth will be horizontal (more hospitals) and vertical (higher penetration in each hospital).

Lifetime value/ Pet acquisition cost (LTV/PAC)

The ratio is a proxy for TRUP’s capital allocation. TRUP’s cost structure 70 cents are spent on claims, 10 cents on filing claims/variable expenses, and lastly they expect 5 cents to go to fixed costs – once they reach scale three years from now. Lifetime value is defined as the gross margin (revenue x 20% – excluding fixed expenses only) multiplied by the average policy lifetime which is currently 72 months or 6 years. Pet acquisition cost is the cost spent on sales and marketing (mostly territory partners) to acquire new pets (that includes replacing the churn rate). Last year, discretionary margin was only 8% of revenues (so $15 million) and LTV/PAC of 5:1, meaning they get $75 million of new business over 6 years; that is a +80% IRR. TRUP will go into cash flow negative territory if they spend a cent more on growth.

When they reach scale (in 3 years by management’s estimates) they will be able to make 15% operating margin, which is double the current margin. In 3 years, revenues will grow to $482 million so potential pet acquisition dollar increases to $482*15% or $72 million, almost 5x current spend rate. I don’t expect the LTV to PAC ratio to remain at 5, that is really high. I do expect it to decrease, but management does not expect it to drop below 3x which translates to +40% IRR. How many companies can achieve that IRR? Not many, that is why I encourage the company to run at 0 profit even if they get a 25% IRR. Why? Because if they paid that to me as dividend I would not know where to allocate it at +25% rates. At the current LTV/PAC an investor would need to return 40% annualized to match management’s current opportunity cost of distributing the cash to shareholders. When the PAC drops to 3x[10], TRUP’s opportunity cost is still 28%.  This starts to look very much like Amazon, spend every cent you make because the IRR is too high to report a profit.

TRUP adjusts its premium depending on vet care cost in the area. They won’t raise the premium due to age, number of claims, annual costs, etc. which is quite the normal in the industry. That incentivizes pet owners to insure their pets as puppies since insurance will be low. But that will mean that the lifetime value of the pet should be higher, and that is exactly what management has seen. Pets generally live more than 72 months so each new puppy owner has a higher lifetime value. That is partially offset by the lower early age premium; but it is still higher than current LTV.


I won’t go into too much detail on management, I think reading the 3 shareholder letters[11] is more than enough to get comfortable with the CEO. Trupanion was founded by Darryl Rawlings, the current CEO, and he owns 7.65% of the company. He has been able to grow TRUP into a fantastic business with a great product into second place from non-existent almost a decade ago. TRUP is eating 40% of the industry growth, growing its top line 27% and has an internal IRR of +80%; the CEO must be doing something right.

Maveron, the VC firm founded by Howard Schultz and Dan Levitan, were instrumental in investing and helping Darryl transition from a start-up to a public company. Dan Levitan is a prominent VC investor and Howard Schultz does not need an introduction. As of 2016, Maveron owns 18.9% of the shares and prominent investment firm Ruane, Cunniff & Goldfarb also owns 6% of the shares.


  • Pricing
    • This is the primary risk to any insurance company. In the past, it has been difficult to sustain unexpected losses, but they have mostly been within a controllable range. If TRUP has been able to keep risks within bounds when sub scale, the more they grow the better their pricing will get.
  • Capital allocation
    • Despite having high IRR, as the margins and the top line increases, the potential PAC dollars will increase exponentially. That would make it hard for the CEO to allocate it efficiently. The good news is that the IRR is so high that declines are expected, but the margin of safety is still very high. The CEO has said that he will keep a minimum 40% IRR.
  • Succession
    • Rawlings has trained upper management well and all of them have +3 years working with him. Another example of quality hiring is that claims employees have worked +5 years in a vet hospital. Territory partners go through a rigorous training; Rawlings believes that it takes two years for them to become thoroughly ready.


Last quarter the company generated $55 million in revenue. Because this business is highly recurring and growing fast, it is unfair to use TTM. We prefer multiplying the latest quarter by 4x and use that as the yearly revenue, for simplicity purpose, so FY 2017 would be 55 x 4 = 220 million. For FY2020 we grow Q1 2017 revenues at 25% for 3 years and multiply by 4x to get $482 million.

  • Claims and variable are always going to be 70% and 10% of revenues, the company has been pretty spot on the previous years.
  • They expect fixed expenses to normalize at 5% of revenues when they reach scale, roughly 650k-750 pets. If we grow current pets enrolled by 21% for 3 years and price increases of 4% we arrive at the lower end of the scale or 677,572 pets enrolled.
  • We assume a maintenance capex of 1.5% of revenues. TRUP has an annual churn of 16.8% (1-98.6% x 12). Half of that churn is replaced by pet owner referral and/or pet owners buying a new pet after their previous one passes away. The other half is replaced by spending only 1.5% of revenues; the model displays it better.
  • At scale, EBIT margin would be 13.5% and we assume 20x EV/EBIT, which is slightly higher than the current EV/normalized EBIT of 18x.
  • Base case, we double our money in 3 years or a yearly return of 25%. We believe that the downside is quite protected and we lose.
  • There is a lot of speculation around the operating margin being sustainable at 15%; after all it is management guidance.
    • To those of you who think this is too high, not quite. The average industry margin in US is 5% when only the largest player (Nationwide 40% of the market) is of scale[12]. Assuming the rest are reporting losses, the profit margin could be around 10% which is EBIT margin @ 15% x 33% tax rate.
    • Furthermore, TRUP is able to maintain a higher margin because of the frictional cost savings @20% that we discussed above.
    • To those of you who think this is too low, it is debatable. We believe that smaller margins and showing 0 accounting profit will keep new entrants away.

[1] Half cancel policies due to financial reasons and the half because the pet is no longer with his pet owner (in most cases, the pet has passed away).



[4] 26,000 out of 28,000 vet clinics/hospitals are single-vet operated

[5] One of the most common illnesses for old pets


[7] Note that Agria has other lines of insurance

[8] Including countries like Brazil, Australia, Canada, etc.

[9] Active means that there is at least a pet enrolled under Trupanion.

[10] Lower than the management’s projection of 3.2x


[12] Since TRUP, the second largest player, has not reached scale yet we assume smaller players are not of scale either.

Posted in Categories | Leave a comment

Baidu – A hidden gem!

All figures are in USD


  1. Baidu is the dominant search engine in China with 80% market share, 50% operating margin trading at 12x TTM FCF with a top line growth rate of 40%-50% and 300% return on capital[1]
  2. Under-monetized mobile platform[2] comprising 2/3 of their revenue
  3. Optionality on an Online-to-Offline (O2O) venture with huge potential

Chinese Market Overview

  1. Growing Income
    • Annual average income is $8,000 a year but growing at 10% CAGR; in the US it is $50,000 but growing at less than 1%.
  2. Elevated levels of Internet consumption
    • Last year, Chinese people spent an average of 26 hours per week or almost 4 hour per day[4] online while in the US people spent only 2.5 hours. Since income, space, and pollution make it difficult to partake in outside activities,  the smartphone is the only entertainment gateway for many.
  3. Technology Leapfrog and Internet under-penetration
    • As is happening with a lot of developing countries, most Chinese people did not grow up having a personal computer or laptop but leapfrogged into mobile devices. The average smartphone in China costs[5] $319 while Xiaomi offers smartphones on average for $149. There are 691 million[6] unique smartphone users and 913 million smartphones in total in China, and that represents only 50% penetration[7]. A third of the population online is under 20; most of them have limited income but in time will bring more dollars online.
  4. Government Support
    • Generally, the government is very supportive of Internet and Internet companies. Recently, the Premier Li Keqiang announced Internet Plus. “The plan,” Li stated, “aims to integrate mobile Internet, cloud computing, big data and the Internet of Things with modern manufacturing, to encourage the healthy development of e-commerce, industrial networks, Internet banking, and to help Internet companies increase international presence.” The government is also pushing hard for Internet penetration in mobile, broadband, etc.

Search Engine


Baidu was founded in 2000 by its current CEO Robin Li and it was not the first entrant in the search engine business in China; Google, Yahoo and a few other players were already operating. Google’s exit was seen as a big opportunity for Baidu, but by then they already had a dominant market share of 70%.

Baidu is in the business of providing ad placements on the Internet in two ways: 1) search related and 2) contextual. Search charges customers when a user clicks on an ad generated by the search engine, while contextual is a revenue sharing agreement between a search engine and advertisers for publishing on third-party websites. Revenue shared with the websites is considered traffic acquisition cost (TAC). Their business is similar to Google, that is why I won’t cover the basics but go straight to the juicy part.

The existence of fixed costs, together with the difference in platform value stated above, has a potentially strong effect on the economics of the search-ad platforms given the CPC pricing structure. Advertisers incur two costs of running campaigns that are independent of the number of clicks. First, they incur costs of setting up the platform, installing software, and learning how to use it. Consequently, the advertiser must exceed a minimal volume of advertising (or more specifically, a minimal level of incremental profits) from this campaign before contracting with another search ad platform. Platform set-up costs discourage smaller advertisers from joining smaller platforms.

Second, advertisers incur costs of running a campaign on keywords. They have to make decisions on the bids and monitor the performance of the campaign. These tasks generally cannot be automated fully and therefore require humans. Thus, the advertiser must exceed a minimum volume of clicks on a campaign before mounting it on an ad platform that has been set up. To take the example above, if it cost $200 per week to monitor a campaign for widgets, the advertiser would run the campaign on platform 1 but not platform 2. Therefore, campaign-monitoring costs also discourage advertisers from mounting campaigns on smaller platforms. On Baidu it costs around $200 or $1,000  to set up an account on Phoenix Nest.

Baidu Exchange Services BES

Initially online advertising was done simply with search engines acting as the middlemen. In the west, advertising is increasingly programmatic, ads and inventory are matched via an ad exchange similar to a stock exchange. BES charges each time they render a service, as well as provide additional services such as demand/sell side services. Currently they are under-monetizing and generate less than 1% of their revenues, Google generates 25% of their revenues from similar services. It still is in its infancy in China. BES will be another revenue stream for Baidu; currently their system has the largest market share at 30% while Alimama (Alibaba) and TANX (Tencent) own cumulatively 40%. BES can handle over 15 billion impressions a day (currently performing 300 million) while Google can handle 30 billion.


Online advertising is superior in most cases to other types of advertising due to higher accuracy – they have more information and know the consumer better, hence higher ROI. In order to know the consumer better it is necessary to:

  1. Know who the consumer is (e.g. late teens University student interested in football and Rock ‘n’ Roll)
    • Have superior information and also analytical skills to process the information
  2. Track the consumer through multiple websites (crawl on him) and devices
    • The latter one is hard but can lead to immense monetization possibilities

On PC, tracking was made easier through the use of cookies[10], but even cookies have a 59% success rate in tracking. On mobile, however, where consumers are increasingly spending their time, cookies have limited use. Not only that, but the screens are smaller and cannot contain as many ads. The likes of Baidu, Alibaba and Tencent (from hereon referred to as BAT) had to reinvent themselves.

Mobile is an important revenue generator for Baidu. Last quarter, 56% of their revenues (and growing) were from mobile by reaching total monthly active users (MAU) of 647 million for search and 302 million MAU for the maps app. The maps app has 70% market share, 50% of automobile navigation market share and provides interior maps of malls, airports and other large facilities. Currently, they are building their O2O business and maps will be extremely helpful. Mobile was helped by several factors such as the leapfrog in cheap smartphone devices and the increase in competition between carriers. The number of carriers increased from 1 to 3 and all of them started offering 3G.

Baidu spent considerably to have its search app pre-installed with almost all the smartphone makers, and currently controls 90% of the mobile search traffic in China. They require people to log-in with their phone number[11] as opposed to e-mail and they can track not only your internet browsing history better, but also the location you made these decisions; it also supports voice and image-based search.

On top of that, the search app acts as a one-stop shop: you can access everything you need from your phone on the app without ever leaving it, including other apps. As the mobile is a different ecosystem, apps enhance the use of the hardware and this area is a unique opportunity for Baidu for 3 reasons:

  1. Acquisition of 91 Wireless
  2. Software Development Kit SDK
  3. The invention of Light App[12]

Google store is not present in China so therefore the app market is taken by third parties and is very fragmented. 91 Wireless, one of the largest app stores in China, was acquired[13] by Baidu in late 2013. Immediately afterwards Baidu provided a platform named SDK (software development kit), allowing third party developers to build, distribute manage, store, and optimize their apps on Baidu cloud platform. It is similar to Amazon web services, but it is offered for free in order to increase the development of apps providing a superior value proposition for app developers. As such, developers claim their development costs decreased by 90%. Baidu’s app market share increased from teens to 42% in a couple of years.

Robin Li has claimed that the current business model of app stores is flawed because 70% of all downloads are for a few popular apps. In addition, the smartphone can only support and efficiently operate so many apps. Light App introduced a new concept in which while searching for a specific query you are shown apps as well – to use without downloading them. In this case, Baidu helps direct traffic more evenly to apps, making the market more efficient and incentivizing developers to move onto their platform. Baidu reached 174 million apps[14] (downloaded or used) on a daily basis for 2014. Developers receive a superior platform to develop their apps while Baidu can access their ocean of information.

Baidu also offers cloud storage to registered members of 2 Terabytes free, Google Drive is only 15 gigabytes. On top of that, Baidu has access to 300 million users of Tieba (forum) and 200 million cloud users. Tieba alone consists of 10% of Baidu’s traffic. Baidu ranks 1st and 4th for most traffic in China and worldwide respectively.

Mobile will be a growth factor because the CPC gap between mobile and PC is still 20% – the gap will eventually close. The surface area is small on a smartphone, so very few ads can be shown, but they have higher CTR, thus the mobile CPC will grow higher. In addition, Baidu recently added another slot for ads on mobile bringing the total from 3 slots up to 4.



iQiyi is a subscription based video streaming service created by Baidu and Providence Equity Partners (backed by HULU) in 2010. As with search they were not the first video/streaming service in China, but quickly ascended to the top.

The service is an ad-based streaming service for user generated content (a la YouTube) and streaming licensed shows and movies generated internally or acquired via many partnerships with Paramount, Lionsgate, NBCUniversal, BBC Earth and China 3D. As of July 2015, their library consisted of 6,000 movies from all around the world – 800 from Paramount including The Godfather, Forrest Gump, etc. Netflix offers 13,000 titles worldwide cumulatively, but only 5,000 in the US. iQiyi is currently the largest one on mobile with 500 million users and 10 million paid subscribers. Doubling in size since July 2015, this is the largest paid subscriber base among Chinese streaming services and greater than the second and third largest players combined. The subscribers pay an average of $3 a month and get access to the movies as well as an ad-free experience.

iQiyi was created with the purpose of always having a premium and licensed service compared to their competitor’s video services – they currently hold the top spot among the daily time spent both on mobile and PC and is still growing[15]. The strategy has proven profitable and a couple of examples can attest to it:

  1. The success of iQiyi’s original TV series Notes of Tomb Raiders, which received 2.4 billion views over the summer or roughly 200 million views per episode, indicates a profitable model for video streaming sites. The Big Bang Theory, the most viewed TV show in the US, reaches around 20 million weekly viewers.
  2. iQiyi paid users will enjoy access to the series called “Shu Shan Zhan Ji,” for which iQiyi has acquired the exclusive online rights, several months before its TV premier.[16] This is the first time in China that a highly anticipated, star-driven TV series will be aired online prior to its TV premier. In the past, TV series have been either broadcast first on television or simultaneously on both television and online platforms. With “Notes of Tomb Raiders,” iQiyi began testing a new distribution model, highlighting its growing cooperation with cable TV networks for the distribution of premium content in China.
  3. Chinese action film“On Line” broke Internet box office records last month when it became the first film to have online revenues surpass box office takings less than a week after release. Immediately after the film’s run in theaters finished in February, it was released online and within four days, online revenues surpassed its theater box office takings according to the iQIYI Online Video Big Data report.
  4. Working youngsters don’t have time to catch shows or movies on television on a predetermined time. Streaming devices are growing in popularity not only in China, but also globally.

The window from theater/broadcast to online needs to decrease in order to increase the streaming value proposition; in China it seems to be easier since 85% of the theaters’ capacity is unutilized.


Currently, iQiyi dominates in mobile MAU and monthly time spent compared to the other streaming services. There are a few reasons why their dominance will grow:

  1. Tencent and Youkou Toudou focus on user generated content like Youtube and lack third-party premium content
    • Tencent owns the rights to broadcast HBO in China, but the problem is censorship. Game of Thrones without the sex, violence, and profanity is just another medieval show and won’t be as exciting
    • Baidu is focused on content that appeals to the consumer without running the risk of censorship
  2. Netflix could be a major contender except for the fact that they are not present yet in China and they lack Chinese content – very crucial for the target demographic[17]
  3. Chinese crackdown on illegal streaming[18]

iQiyi received a buyout offer of $2.8 billion[19] from Robin Li and the CEO of iQiyi – currently Baidu owns 80%. Robin Li controls 54% of the voting power in Baidu and it looks like the deal will go through. I believe it is a low-ball considering AliBaba acquired Youkou Toudou, third largest video service in China, for $3.5 billion. The rationale behind taking it private is that iQiyi’s expenses are nearly 5% of revenue (mostly for content cost) while revenues are minimal.


Online 2 Offline

Baidu’s O2O initiative consists of three businesses. They are Qunar the travel site, Nuomi for group buying, and Baidu Takeout Delivery (BTD). China is in the early stages of a boom in its on-demand economy, but local commerce remains highly fragmented and happens predominantly offline. Most of the costs are in subsidies and marketing for both Nuomi and BTD to attract suppliers.

Qunar – Travel sites

Qunar was an online travel booking company that was recently (25 October 2015) acquired by Ctrip, the largest online travel booking company in China. Qunar dominated 2-tier cities while Ctrip dominates 1-tier and collectively they control 20% of the total travelling market share. The cumulative online travel market share is almost 66% and growing.


Baidu received a 25% stake in Ctrip in exchange for its 45% stake in Qunar,[20] therefore the stake will appear on the balance sheet as an equity investment at fair value; as of March the stake is valued at $4.5 billion. I will value it at market despite both Oppenheimer and Deutsche Bank having a 12-month price target of $60 and $55 respectively when current price hovers at $45. Baidu is currently the largest shareholder of Ctrip and Ctrip’s management owns some 7.3% – the others are mutual fund like investors cumulatively owning 31%.

Ctrip is an online travel booking company involved in accommodation services, transportation ticketing, packaged-tour, corporate travel and others. The business has been growing organically in the mid 20s pre-acquisition – now they have a strategic partner in Baidu. This will add to the list of current partners Tencent and Priceline.



Nuomi is in the business of group buying similar to Groupon. They were acquired from Renren in 2013 and 2014 in two chunks cumulatively worth $270 million. Baidu has over 650,000 stores from across 400 cities in China and is currently number one by Gross Merchandise Value (GMV) in about 60 of them. China is in the early stages of a boom in its on-demand economy, but local commerce remains highly fragmented and happens predominantly offline; O2O represents a $1.5 trillion market:

Group buying has its roots in China in the early 2000s when merchants offered a discount to a specific product if a certain number of buyers (volume) were achieved. Chinese called it 團購 (tuángòu) or group buying, but it wasn’t until the rise of Internet companies that the efficiency created propelled group buying to another level. Group buyers are intermediaries and make money on the spread of the deal. This can be a powerful business model for the Chinese consumer who is driven by a bargain.

To Westerners the most common and well-known example is Groupon. Google tried to acquire them in 2010 for $6 billion[21] but failed and immediately after launched Google Offers, a similar business which was closed down in 2014. The earliest attempt at group buying in the western world was by Microsoft co-founder Paul Allen, but closed shop in 2002.[22]

Nuomi sells a range of articles such as bookings for hotels, hairdressers, leisure services, online movie ticketing etc. and includes BTD. Internet companies are extremely fast at converting offline to online purchases. For example, in movie ticketing the online penetration was 10% in 2012 and now stands at approximately 50%. On average 85% of movie theater seats are unsold and trucks on the go carrying merchandise are more than 40% under-capacity[23] – Nuomi wants to bring higher efficiency. Over the past 30 days, keyword searches on Baidu search for Nuomi more than doubled year-over-year; that’s multiple times faster than its peers.

There are more than 40 million SMEs in China without websites and Baidu plans to capitalize on this by launching “Local Express” in September 2015. This platform will help Nuomi merchants get discovered by pushing content of their business such as images, locations, distance and promotions in a performance based ad. With one click, the user will be able to land on the merchant’s page and complete the transaction. Baidu expects that this will provide local merchants a very simple way to build their online presence without a domain name or renting servers since Baidu will provide the hosting services. Given consumer preferences towards local stores and the tremendous opportunity in terms of number of merchants, Baidu believes this could drive revenue growth in the future. They are taking a vertical approach to the powerful click-to-action button. Alibaba was involved in similar successful initiatives where they accessed China’s 600,000 villages with the Internet. They use an Internet center for people to reach and purchase goods online as well as provide a platform for many villagers to sell their produce on Alibaba.

Nuomi’s strategy is differentiated from its competitors in that it works more closely with merchants to drive repeat users through the Membership Plus CRM programs such as stored-value card and priority seating, etc. Nuomi is a distant second in terms of market share with 13% while Meituan and Dianping (MD) merged recently[24] and cumulatively control 82%. Alibaba used to be an investor of MD but sold its share shortly after the merger and will pursue O2O on its own.


Baidu Takeout Delivery BTD

BTD connects restaurants with consumers for ordering takeout or to a restaurant reservation. Currently, they fulfill 50% of the takeout orders last mile – not sustainable because profit margins will remain static. From its inception last year, BTD has followed a differentiated strategy by working with high quality licensed restaurants and targeting the priced elastic working age demographic. Their market share in orders is 8%, but in GMV terms is 19%, up from 12% in H1 2015; BTD bills are twice as large as their competitors.

It covers more than 70 cities where they are the leader in half of them, including Beijing and 6 additional provincial capitals. Takeout not only complements the group buying services, but also leans on their technology advantage. The key sustainable edge in take-out food delivery is scale, optimizing logistics and resource allocation. Only 1% of reservations are made online and online take-out food delivery represents only a low single digit percentage of overall restaurant GMV.

The O2O Chinese takeout market was $45 billion in 2015 and is expected to grow to $163 billion and $245 billion in 2017 and 2018 respectively according to Jeffreys. Currently BTD charges between 2%-3% of GMV while competitors charge 3%-5%; the expected charge rate is going to be 5%-7%. The other two big players are Eleme at 33.7%, Meituan at 33.1%, and others 14.2% (BTD’s market share is 19%). We can expect the market to either consolidate or the biggest players to dominate.


R&D: Deep Learning and AI

Just as with Google, R&D tends to be a delicate topic for many analysts due to scepticism arising on the effectiveness of the spending. Google and Baidu are technology companies blessed with the ability to generate FCF early on, only to see them spend enormously on R&D. Despite substantial reading on the matter and to the best of my abilities it remains a black box.

Baidu, despite being 1/8 of Google, employs the same number of people in R&D at approximately 20,000. In monetary terms they spend 1/6 of what Google spends but the difference is not so big considering the wage differential between the two countries.  Therefore, Baidu’s  R&D has a lot of firepower, but what is this firepower doing?

Most of the firepower is being invested in deep learning or machine learning/AI. Baidu’s Chief Scientist Andrew Ng is the leading force behind it. His background is stellar having been 1 of the 3 founders of Google Brain – Google’s AI initiative, founder of Coursera, developing one of the most capable autonomous helicopters[28] in the world, etc. To understand how profitable this initiative has or will be, an analogy is appropriate. In order to have higher ROI on your advertising you can either have more, targeted information or be better at analyzing the information you have at hand. Baidu is well known to have one of the most vast information pools due to search, the app store and Baidu Union. The more steps they take to comprehend their customer, the better they will become at understanding purchasing patterns. Google has claimed that Google Brain has produced in value for Google something close to total costs of Google X,[29] or all of their so-called Moonshots projects.

One of the big achievements was their successful road test of their autonomous BMW car in Beijing.[30] Despite being a bit behind Google on complicated maneuvers, they predict that their project will be used for autonomous shuttle buses in 3 years. Having said all this, the costs will be counted at full value on my valuation rationalizing that they bring intangible economic value.

At their Baidu World Conference in September, they unveiled Duer, the intelligent voice-based assistant that can execute commands for users in the real world from making dining recommendations, finding discounts and ordering food, to buying movie tickets and answering sophisticated questions. Duer represents a kind of sneak peek into the future of search where voice activation, artificial intelligence and services all come together. Duer was named one of the top 10 Technological Breakthroughs of 2016 by MIT Technology Review.[31] Baidu is already seeing a ramp up in users who have switched to voice commands to search the web; this will make it easier for the company to crawl and profile users since voice is similar to a fingerprint in terms of uniqueness.[32]


Baidu is controlled by his founder Robin Li who worked on one of the pioneer projects for search engines. He owns 19% of the outstanding shares, but +50% of the voting power. Many people can be uncomfortable with such control, but Robin has always invested for the company, avoiding dividends and performing buybacks. Sacrificing margins for future benefits in the beginning with sale force expenditures and now with O2O, his moves have paid off.

Executives for Baidu have worked for major Western companies like Google, GMAC, Microsoft, eBay, etc. They have been paid on aggregate less than $1.5 million a year in cash compensation since inception. Last year was an odd year when cash compensation reached $4.7 million. Share-based compensation has been below $5 million cumulatively, but currently the options for many officers including Robin are: price range of $648-$2,245 with a time span of 2020-2024, current price is $187. No executive officer is entitled to any severance benefits upon termination of his or her employment with the company except as required under applicable PRC law – no golden parachutes here.



Baidu competes indirectly with Alibaba (largest e-commerce platform) and Tencent (largest social media platform) for advertising and online consumption. Each will be dominant in their respective market and the market is large enough for them to co-exist as they will leverage their strengths; Baidu in internet traffic and technology, Tencent in social media information and Alibaba in consumer behavior and logistics. Baidu’s search engine makes more FCF than either Tencent or Alibaba, yet these companies are valued 3x higher than Baidu.

All 3 are involved in O2O, but Tencent has a more hands-off approach and only invests as a minority shareholder. It doesn’t look like the company’s focus is O2O, rather investing in a VC-like manner. For Alibaba, they like to get involved on many fronts such as Didi Kuaidi, group buying, etc.

I personally think that calling Baidu “the Google of China” is underestimating and demoting the company. Baidu has around 20,000 in R&D headcount (500 of them in Silicon Valley), almost as much as Google, but is only 1/8 of Google’s market cap. They were able to test drive successfully a driverless car and are one of the top companies with significant operations in machine learning.



  1. The stake for Ctrip and iQiyi will be considered as investments with no effect on P&L
    • iQiyi will most likely be bought out as well
  2. Search will be valued at the margins it’s making on a stand-alone basis, not taking into account Nuomi, iQiyi, or Ctrip margin compression
  3. On Q2 2015, Baidu announced that they would pledge $3.2 billion (20 billion renminbi) for O2O. Therefore, I will subtract that from their cash position – that is equal to the margin compression they will be facing for the next 2 years
  4. I will add to the valuation my estimations for Nuomi and BTD revenue
  5. The terminal value will be 2 years ending December 31, 2017


  1. The search business growth rate has been at least 40% top line for the past few years
  2. Operating margins, excluding O2O, have been stable at 50%
  3. Effective tax rate and Other Income are almost equal, and operating income is a good proxy for Free Cash Flow
    • Note Baidu receives a tax relief since it counts as a technology company in China, therefore paying a lower tax rate for their search business (15%)
  4. Ctrip is being valued at a lower market price for bear case, and sell side consesus $60 for the bull
  5. Nuomi and BTD are being valued at 1x sales, expenses are subtracted from cash so it is essentially 1x FCF
    • Nuomi grew 5x last year
    • BTD grew 12x last year. My bull case assumes the growth rate for takeout delivery estimated by Jeffrey’s, anything below is being conservative.

Why does this opportunity exist

  1. Accounting matters
    • Baidu’s margins are compressed by the investments in the O2O business and partially hide the highly profitable, cash generating search business
  2. Global and Chinese economy slowdown
    • The stock market had gyrations recently in China due to the linking between Shanghai and Hong Kong stock exchange
    • Less than 10% of the population in China has a trading account and they invest minimally in stocks – the effects are minor
  3. Negative sentiment related to Chinese companies in general and sentiment regarding the ability of Chinese tech companies competitive position against their Western counterparts
  4. The shares are very volatile for a $60 billion company, due to lower float


  1. Recently Baidu has decided to segregate the margin effects of iQiyi (5% of revenues) and O2O (25% of revenues)
  2. Sale of iQiyi (soon) and Qunar (last quarter) will provide margin expansion
  3. $1 billion a year of share buybacks, equal to 2% of outstanding shares


The materials to which this disclosure is attached as well as any electronic or verbal communication related to the subject matter of these materials are intended for informational purposes only, are subject to change, and do not constitute investment advice or a recommendation to you.

Any reproduction of these materials, in whole or in part, or the divulgence of any of the contents, is strictly prohibited. These materials are intended for the exclusive use of the designated recipients and may not be reproduced or redistributed in any form or used to conduct any general solicitation or advertising with respect to any investment discussed in the information provided.


[1] Calculated as EBIT/Working Capital + Long-term Capital

[2] “We don’t want to over-monetize the mobile traffic at this time because we think it’s still too early. We still see a lot of growth opportunity in terms of stickiness for mobile search.” – Robin Li from Q3 2014







[9] Atlas lets them measure ad campaigns across screens by solving the cookie problem; and it lets them target real people across mobile and the web.

[10] also known as web cookie, it is a piece of data sent from a website to your web browser in order for the website to recognize you next time you log in.

[11] For example, if a frequent movie consumer tends to search for local cinema times. We’ll include that link on a personal homepage or if a user in Beijing often searches for weather we will include our link on the Baidu homepage to updating weather forecast automatically.



[14] Q4 2014 conference call









[23] Q1 2015 conference call










Posted in Categories | Tagged , , , , , , , , | Leave a comment