by Nima Wedlake on February 5, 2016
We regularly conduct research within our focus areas, both to deepen our expertise and identify trends and opportunities. Today we’re publishing our latest research, focused on mobile advertising.
Unsurprisingly, mobile is top-of-mind across our portfolio of ad-tech companies – it’s by far the fastest growing digital ad format. This growth is the result of a dramatic shift in consumer attention away from TV, radio & desktop PCs to mobile. Adults in the U.S. spend more than three hours on mobile devices each day – playing games, connecting with friends and consuming media. As ad-tech investors, this shift in consumer attention creates opportunities to fund companies that are building the ad infrastructure to reach consumers on mobile.
It’s no secret that advertising dollars follow eyeballs, and the rise in mobile ad spend illustrates this shift. In fact, mobile will drive all of the growth in digital ad spend over the next several years (as illustrated in the chart below). We can look to social networks for a clear example of this shift, where both users and advertisers have embraced mobile. In Q4 2015, Facebook reported that mobile ads made up 80 percent of the company’s total ad business, compared with 23 percent in the same quarter of 2012. Similarly, 86 percent of Twitter’s ad revenues come from mobile.
Our mobile research is designed to provide a broad overview of the mobile ad-tech ecosystem, including market size & growth, emerging ad formats & standards, and key ad-tech vendors. In the report we cover three trends contributing to mobile’s continued rise in 2016:
- The mobile ad spend will surpass desktop spend
We’ve reached a tipping point in digital advertising, where consumer attention is moving primarily to mobile devices (supply) and marketers are becoming more comfortable spending their ad dollars on mobile (demand). As the scale tips, expect to see more media planning dollars being allocated to mobile.
- Mobile real-time bidding (RTB) spend is ramping up
Mobile accounted for more than half of programmatic in 2015, and will rise to over three-fourths of total programmatic spend by 2017, according to eMarketer. As we’ve seen in the desktop world, programmatic brings real benefits to advertisers, including better targeting, efficiency, scale, and ultimately performance. The breadth of data to inform programmatic buying on mobile is also growing – which will help drive incremental mobile spend.
- Brand ad dollars are flowing into mobile
The mobile advertising ecosystem has evolved considerably in the past few years. Much of the early ad spend has been driven by app developers looking to drive installs. The market has exploded in terms of download dollars, with companies like Facebook relying heavily on app downloads to drive early mobile ad revenue growth. While performance-based ad spend was first to the party, we expect brand dollars on mobile to ramp up significantly over the next several years. According to MoPub, 60% of its top 25 private marketplace advertisers are brands (as opposed to performance advertisers). Expect this percentage to grow over the coming year.
It’s an exciting time to be at the convergence of mobile, programmatic & brand advertising. We look forward to continued innovation in mobile, especially as formats more engaging ad formats like native & video grow in popularity.
by Laura Cain on January 26, 2016
Everyone knows credit scores matter; credit scores are the backbone for how lenders evaluate a person’s risk. But should credit scores matter? Are credit scores a fair way to evaluate credit worthiness or are the formulas too backwards looking?
Why credit scores suck for young people
Currently three credit reporting agencies dominate the market: Experian, Equifax, and TransUnion, each of which emphasize slightly different aspects of a consumer’s background when determining a credit score. Methods for determining credit scores take into account payment history, new credit inquiries (how often you request for new credit), credit utilization, length of credit history, and types of credit utilized (credit cards, mortgages, auto loans, etc.). The distribution of credit scores is skewed towards the higher end. The mean FICO score is 687 whereas the median score is 723, implying that 50% of the population has a default rate below 5%.
While credit scores have a strong negative correlation with delinquency rates, they also have a strong positive correlation with age. At face value this is an understandable correlation; income and financial prowess typically increase with age, along with the number of data points credit rating agencies can utilize to determine your credit score. However this also implies that credit rating agencies are inept at gauging the risk of younger populations – the average 18-24 year old has a credit score about 100 points below what he or she eventually will have. Banks are therefore overestimating the risk of younger populations and mispricing credit products to make up for the ‘risks’ that result from their inability adequately estimate default rates.
This gap has a huge impact on the student loan industry, as banks do not accordingly adjust rates for younger populations and relay heavily on a co-signers credit to determine risk. While slight differences in interest rates may seem negligible, over time the effect cumulates and the impact becomes substantial.
Students who do not qualify for sizeable scholarships and grants, yet also do not have a strong credit or potential co-signers, are often left at the most significant disadvantage. Opportunities therefore arise for companies able to accurately determine the risk of a student borrower who would otherwise be charged a higher interest rate.
Is there a better approach to lending?
There are approximately 21 million college students in the United States, each paying an average tuition of $19,339/year and taking out an average of $29,400 in student loans. Between 75-88% of students utilize loans to pay for education, making student loans one of the primary sources of financial aid for higher education.
Both the federal government and private banks originate student loans. The federal government accounts for 83% of the $1.1 trillion of outstanding student loan debt. Federal loan programs offer more competitive terms than private loans and are advised to be maxed out before considering private loans.
However, this doesn’t have to be the case. The federal government blindly underwrites students – failing to consider major, school, etc. and offering everyone the same low rates. Private companies able to predict young students’ future creditworthiness can skim-off the best customers and offer more attractive loans.
SoFi is dominating because they take alumni who already have good jobs and simply reassess their creditworthiness and drop their rates. Ditching credit scores (which would still penalize people 10-20 years out of college) and focusing on a person’s cash flow has been an extremely successful strategy. But there is an opportunity to soak up these customers even earlier, we just need to figure out what signals can be used to identify them.
Thomvest analyzed the top 700 schools with Title IV student financial assistance program. Only schools with over 300 loans during 2009-2011 were considered. Together, the data accounts for around 64% of all federal and private student loans during 2009-2011.
We then divided the schools into 7 categories, referencing Forbes and US News for approximate college prestige rankings. Categories 1 & 2 are comprised of roughly 50 schools each, and categories 3-7 are comprised of approximately 100-150 schools each.
The diagram below compares a college’s prestige to its respective default rate. There is a clear distinction between the default rates of category 1 schools and the rest of the categories. However, if 5 of the colleges in category 2 are ignored – Missouri Valley College, Trinity College, Union College, UC Merced, and Ohio State – the default rates in category 2 would all fall below 6%. Although a 6% default rate is in-line with a sub-700 credit score, as a holistic portfolio the loans of the top 100 schools would serve as an attractive portfolio. The weighted average of all category 1 & 2 school default rates is .0255% – which is an impressively low default rate considering no additional factors such as major, income, or other outstanding debts were considered.
Obviously SoFi’s strategy of cash flow analysis has been successful, so what if we try to predict a student’s mid-career salary? Schools with an average mid-career salary of over $100,000 each have a default rate below 5%.
While you may think this is the same group of students that went to the most prestigious schools (that’s why we try to go to the more prestigious schools right?), it’s not. There is a trend between the most prestigious schools having higher mid-career salaries, however, a good portion of the sub $100,000 earners come from the top 50 schools.
Across the board, if you went to a category 5, 6, or 7 school the average student will make less than $100,000. Prestige appears to be good at filtering people out, but not selecting from within the pool of better schools.
Our (pre) conclusion
The information Thomvest referenced had no data on majors. But we have a hunch this is the key factor behind default rates. Prestigious liberal arts schools had higher default rates than lesser known engineering colleges. Unranked medical schools had astonishingly low default rates compared to other schools in their respective prestige cohorts. While this is not mind-boggling, it is important.
With the current price of higher education, some degrees are simply not worth it. Depending on what you want to do, spending tens of thousands of dollars to obtain a piece of paper (your diploma) may not pay off. The signaling from going to a lower ranked school may prohibit you from getting higher paying jobs, making the forfeited salary from not going to college more economical than taking on the debt.
And for those who don’t default, you’re flipping the bill for everyone who does. It’s the way lending has always worked. But then again, the way the government is underwriting student loans is unprecedented.
Right now the system is a lose-lose. However there is a huge opportunity for the company that can crack the problem.
by Laura Cain on August 25, 2015
The banking industry appears to be undergoing a renaissance driven by changing consumer behavior and technical innovation. Software is eating the industry. In retrospect, we can see how the first wave of innovation came in areas such as online account access and payments. Changing consumer behavior (such as the shift to mobile) and the use of big data has enabled increasingly complex transactions (such as lending and asset management) to move online. Consumers have largely stopped going to retail branches, and reserve the occasional branch visit for major one-off transactions.
Our first investment in the financial services industry came many years ago with an investment in LendingClub. We put both equity and debt into the company, making a sizable purchase of loans via the platform itself. We saw the company’s potential to bring marketplace dynamics and software disruption to the lending industry. The end goal for borrowers and investors on the platform was simple: lower cost loans for borrowers, increased yields for investors, and high levels of customer satisfaction. As a result, LendingClub has grown into a sizable public company. With experience on the platform and a realization of the potentially transformative nature of this model, we’ve gone on to invest in companies across the online lending space: Kabbage (www.kabbage.com), LendUp (www.lendup.com), and SoFi (www.sofi.com).
The renaissance in financial services has drawn in substantial amounts of venture capital. In the past year alone, the number of fintech deals has grown 16% and the capital funded is up 46%.
While many entrepreneurs develop expertise in the specific segment they intend to disrupt, we’ve noticed that startups usually don’t have the time or resources to look outside their niche and understand how they fit into the larger context of banking and lending markets. To help put the industry in perspective, we developed an overview of the banking industry in the US. What’s remarkable is not only the insights this gives into the financial lives of Americans (be it millenials or seniors), but also the perspective this gives us on the large banks we’ve all come to use. Indeed, consolidation over the last several decades has led the four major banks (JP Morgan, Bank of America, Citigroup, and Wells Fargo) to hold around half of the market’s depository assets.
Today we’re happy to provide the first version of this industry overview. We’ve chosen brevity over depth, so as to provide a snapshot of the overall banking landscape. We’ll continue to iterate on this overview and welcome questions and comments. In subsequent posts, we plan to provide deeper dives into sectors that are of interest to both ourselves and others. We look forward to contributing to what feels like yet another opportunity to be at the front door of history-making companies.
by Tweed on September 29, 2014
Our firm has been focusing on investing in ad tech for the past few years. As part of our focus on the space, we’ve brought on board research analysts to take a close look at specific segments of the market and to conduct more general market overviews. In discussing our findings with entrepreneurs, we’ve found that the research we were doing generated a lot of interest, and we thought it might be useful to make some of the more general market findings available. We’ve included below a link to download our general market overview for the ad tech market.
When we really began focusing on ad tech few years ago, the investor sentiment was generally negative on the space. Many firms had placed bets that hadn’t panned out, and there was frustration with the perceived lack of the large exit opportunities for startups in the space. We saw a potential place for a contrarian play that involved going deep into mobile, video, and technology that allowed brand marketers to displace middlemen and connect directly with consumers. The big trends that we have been investing in focus on key themes such as transparency and attribution, the migration of ad dollars to mobile and video formats, and marketers moving towards bringing ad tech into their own hands.
Mobile has evolved considerably in the past few years. Much of the ad spend has been driven by app gaming companies. Burst campaigns of downloads allowed a new app to drive to the top of the app store charts. Once there, these gaming companies hoped that organic downloads would take hold and in turn lead to a lower cost of customer acquisition as consumers discovered and downloaded their apps via the app store ranking. The market has exploded in terms of download dollars, with companies like Facebook relying heavily on app downloads to drive their advertising revenue growth. Trends that we continue to see in mobile aim to develop a better understanding of user acquisition and include measurement, attribution, emerging formats like in-app mobile video, and cross device targeting.
We view video as one of the most interesting categories. In the U.S., the ad budgets for television continue to lead spending among formats. With the proliferation of smartphones, tablets, and connected devices, however, much of the time is being shifted away from watching TV. Brands already spend millions of dollars hiring agencies to produce high quality video content, and that creative can be easily ported to smaller screens. The ability to effectively target specific geographies, demographics, and time periods allows for great messaging. While brand dollars are still relatively nascent in mobile, there has been increased movement to bring video ads, with their potent storytelling capability, into mobile. We feel that the advent of programmatic marketing and real-time bidding will enable these brand marketers to ramp ad spend into mobile video in a dramatic way.
Finally, brand dollars have long gone to agencies or other parties who use technology to buy ads. If you follow a dollar of advertising that goes through this cycle, the percent that gets taken out before the ad is even shown is substantial. Many brands are now realizing that they can use advanced technology from DSPs (demand-side platforms) and other self-service tools to accomplish similar goals. This allows for full control, transparency, and iteration of strategy for brands. As budgets shift to the marketing department within brands, internal teams who harness these technologies will become increasingly important. As we have argued in earlier posts, the middlemen in the industry will begin to be displaced, and there will be ample opportunity for software for use by brand marketers to emerge as a key ingredient in targeting and delivering ads.
We hope that the slides we have compiled are helpful and shed some light on the market. Aside from the general overview provided here, we also have been working on more specific research projects in other areas such as mobile exchanges and real-time bidding, and we’ll post our findings on those topics in subsequent blog posts. In the meantime, we welcome any comments or feedback you might have on the general market summary below.
by Nima Wedlake on April 2, 2014
What will become of the humble cookie? The tiny data files sent from websites to browsers have come under much scrutiny recently, particularly from privacy advocates and policy makers. Even advertisers agree that the web needs a viable alternative that balances privacy concerns with marketers’ desire to target users effectively.
As investors focused on the advertising technology space, we’ve paid close attention to the discussions surrounding cookies and other tracking mechanisms, given their importance to the ad ecosystem. In this post, we’ll summarize these discussions and touch on emerging tracking technologies that may ultimately replace cookies.
Is the cookie crumbling?
Third-party cookies (i.e., cookies set by someone other than the website being visited) have enabled digital advertising to flourish into the multi-billion dollar industry it is today. They are used to run retargeting campaigns, enable real-time bidding exchanges, and reconcile user-specific demographic data across multiple sources. And they’re everywhere–nearly 85 percent of the top 1,000 sites have cookies set by a third party, according to a study by the UC Berkeley Center for Law and Technology.
Yet many industry leaders have grimly declared “the death of the cookie” sometime within the next few years. What’s the cause of their cynicism? Here are some of the most common critiques:
Privacy concerns: Critics argue that third-parties collect and store excessive data on consumers, often without their knowledge. Consumers agree–57 percent of Internet users are either “concerned” or “very concerned” about their online privacy, according to a recent study by analytics firm Annalect. Law makers are concerned as well, and have floated potential legislation to limit the scope of tracking by third-party advertisers. They’ve tasked industry and consumer groups with defining a browser-based “Do Not Track” standard that would allow users to easily opt out of tracking.
Limited reach: Cookies aren’t effective in mobile environments (third-party cookies are blocked by default on iOS devices, for instance). This can be limiting for advertisers, given that we spend more time on mobile devices than we do laptops and PCs. Additionally, many desktop browsers including Firefox, Safari and Internet Explorer have chosen to preemptively opt users out of accepting third-party cookies.
Poor cross-device tracking: Cookies can’t provide cross-device targeting capabilities (i.e. targeting the same user across mobile and desktop devices). As consumer attention continues to bifurcate across devices, the value of desktop-only cookies starts to decline.
Stacking up the alternatives
So what’s next? What’s the magic bullet that balances privacy considerations with sophisticated cross-device tracking capabilities? Some interesting cookie alternatives have emerged, each with their own benefits and drawbacks. We can classify these identifiers into three buckets: known, stable and statistical.
Known identifiers are typically associated with some form of personal information, such as a name or email address. Large consumer internet companies have access to millions of known IDs, across both desktop and mobile. These IDs have some important advantages over third-party cookies:
- Known IDs are highly accurate given that we typically pass highly accurate demographic and interest data to social media companies like Facebook and Twitter. Known IDs have large mobile and cross-device reach (persistent login across desktop and mobile devices)
- Privacy concerns may be mitigated by giving users ample control over the how/when the ID is passed to advertisers
- Both Facebook and Twitter are expected to begin allowing advertisers to use their known IDs outside of their “walled gardens” (Twitter’s acquisition of MoPub has many industry observers predicting that this will happen on mobile in the near future).
Stable identifiers are typically associated with a specific device or browser. Apple’s IDFA (“identifierForAdvertising”) used on iOS devices is a good example of a stable ID. These IDs are typically persistent (don’t expire or erase), anonymous and allow for user opt-out.
Most notably, Google is rumored to be developing a stable ID system, known as Advertising ID. The Advertising ID would be a unique identifier associated with the Chrome browser and Android devices that persistently identifies users. It would be anonymously passed to advertisers approved by Google, while giving users greater control over how they are tracked online (such as the ability to opt-out or block specific advertisers). The Advertising ID could also include “known” data for users logged in to Google products like Gmail and Google.
Details on Advertising ID are sparse, but its implications are vast given Google’s massive reach and scale. See Ari Paparo’s excellent post for some predictions on how the Advertising ID may be designed.
Finally, statistical IDs attempt to bypass cookies entirely by using other attributes to identify users, such as IP address, device type, and browsing patterns. Using these attributes, companies like Drawbridge, TapAd and AdTheorent can probabilistically determine whether two devices are connected (for instance, your phone and PC). The resulting statistical ID can then be used for ad targeting. Here’s a more detailed description of the technology.
Although promising, the technology is still in its early days; most statistical IDs are typically only 60 to 70 percent accurate. Nonetheless, many within the industry are optimistic about the potential of statistical IDs because they allow for cross-device targeting, are anonymous (quelling some privacy concerns) and aren’t owned by a single large player like Google or Facebook.