Data Beats Math –Why Apple’s Maps Failure is a Big Data Problem October 4, 2012Posted by John Vrionis in Uncategorized.
add a comment
At Lightspeed, we’ve been investing in the theme of Big Data over the last four years. We like to say that at the core of the theme is the simple idea that “data beats math” or put another way that predictive algorithms working with data samples will never beat analysis where ALL of the data is evaluated.
We started by investing in building block companies, DataStax and MapR, two pioneers building enterprise class platforms for “big data” workloads by leveraging Cassandra and Hadoop. But we’re also big believers that with these innovative technologies maturing almost every industry, from healthcare to retail, will benefit from the power of “big data” solutions.
There are already some industries benefiting by harnessing the power of these new technologies and pointing them at specific problems where better answers could be derived by looking at MORE data FASTER – companies like ZestCash* that is using big data to provide customers with better options in financial lending and Boundary* which is using big data to revolutionize the IT monitoring space.
The recent chatter about how Apple Maps pale in comparison to Google Maps – ignited by all the iOS 6 users who are now forced to use the Apple product is a perfect example of “data beats math” in action.
Why? In addition to harnessing data from across the network (Google Earth, Listings, etc.), Google Maps has the potentially insurmountable advantage of using collected, historical data about what routes users actually take. They use that information to drive and prioritize recommendations the next time somebody asks for the same directions or the same location.
Conversely, Apple’s data set is infinitely smaller and therefore less accurate. So until Apple builds up enough historical data to compete with Google, the product will be inferior because at the risk of over simplification it has to “guess” about the answer. And as many have pointed out on Twitter, blog, etc. right now guessing just isn’t cutting it.
The big question is will people endure the inferiority of Apple Maps long enough to let Apple capture enough data to provide comparable answers to the Google Maps product? Time will tell, but for my part I’ll be seeking out alternatives for my iPhone.
*Lightspeed backed companies
The Future of Data Storage September 7, 2012Posted by krishparikh in 2012, big data, enterprise 2.0, enterprise infrastructure, startups, Storage.
Tags: Data Storage, Event, Flash Memory, Nimble Storage, Nutanix, Tintri, XtremIO
1 comment so far
In the so-called age of “big data”, enterprises will need to contend not only the sheer volume of data they generate (ranging from hundreds to thousands of terabytes), but also to manage the velocity and variety of these new data streams. (1) To put these numbers in perspective, imagine each enterprise storing and analyzing data equivalent to the volume of information catalogued by the US Library of Congress every year! (2)
Recognizing that this explosion of storage growth cannot be managed by legacy infrastructure, both investors and storage vendors are betting on flash memory as the technology to keep pace with the growing data challenges faced by enterprises. Incumbents EMC and IBM have recently made strategic acquisitions in all-flash storage companies XtremIO and Texas Memory Systems to augment their legacy storage solutions. Meanwhile startups Pure Storage and Nutanix have raised large rounds of growth financing, further validating that investors are also bullish on the flash storage trend.
We at Lightspeed were early believers in the disruptive power of flash memory in next-generation storage systems. (3) The decreasing cost of flash memory driven by widespread adoption in consumer devices, coupled with data access and retrieval times 10-100x faster than rotating disk, and a power and physical footprint 10 times smaller than disk well positioned flash to be the transformative storage technology in the datacenter. Our early investments in component technologies (Link-a-Media, Pliant Technology, Fusion-io), systems companies (XtremIO), and software technologies (IO Turbine) centered around flash memory have validated that hypothesis.
To better understand the role of flash memory and its impact on performance, capacity, energy usage, and cost in next-generation storage systems, I invite you to join me at the Future of Data Storage event on September 18 in San Francisco. Hosted by BTIG and moderated by Andrew Reichman, principal analyst with Forrester Research covering infrastructure and storage technologies, the event will bring together five leading companies focused on driving innovation around data storage in the enterprise:
- Nimble Storage is creating hybrid storage systems that converge primary storage, backup storage, and data protection technology in a single appliance.
- Nutanix is creating converged storage and compute appliances that allow enterprises to build Google-like, scale-out datacenters
- Pure Storage is creating all-flash enterprise storage arrays focused on delivering high performance at cost effective price points.
- Tintri is creating storage systems optimized for virtual machines, improving the manageability and cost-effectiveness of virtualized workloads.
- Virident is creating PCIe flash accelerator cards that allow frequently used data to sit closer to the CPU in servers.
As we look toward the future, startups will continue to innovate around flash memory, creating next-generation storage systems stitched together with intelligent software to disrupt existing markets based on disk architectures.
If you are interested in joining us at the event please email eventRSVP@lsvp.com along with your name and contact information. Webcasting will also be available.
I look forward to exploring these trends further during the Future of Data Storage event from the lens of five emerging startups – hope to see you there!
(1) McKinsey Global Institute Report “Big data: The next frontier for innovation, competition, and productivity”
(2) Library of Congress Website, January 2012 Data: As of January 2012, the Library has collected about 285 terabytes of web archive data growing at a rate of about 5 terabytes per month.
Follow us on twitter at @lightspeedvp for more information on the future of storage and events like these.
Tags: startup advice, Summer Program
1 comment so far
As a part of our Summer Fellowship Program, we bring in influential speakers from around the valley each week to share their insights, lessons learned and tips with our teams. The program has now been in place for six years, so with recent fellowship classes I have been fortunate to pull from our list of alumni when curating the speaker list. One of those alumni, Pinterest CEO Ben Silbermann, was generous enough to join us this past week for lunch with our fellows and alumni from past years.
During lunch, Ben shared details of his background and thoughtfully explained the journey of how he came to be CEO of one of the hottest startups in the consumer internet space. He also shared a number of insights and lessons which I think we can all learn from:
Hire Great People, regardless of if you have a defined role for them: Ben shared that one of the things he is thankful he did in the early days was to hire people that he thought were great people even before he knew exactly what their role would be. Great people, he explained, can add value in various roles and often provide key solutions to problems that arise throughout your lifecycle.
Learn from No: Whether you are seeking funding, making offers to potential employees or trying to build partnerships, as a startup you are going to hear the word no a lot. What makes Ben a great entrepreneur is that he recognizes that most of the time, people are saying no for a good reason. He had the patience, self-awareness and intellectual honesty to evaluate the situation and make the necessary changes. Whatever the reason for No, Ben stressed the importance of using it as opportunity to learn and to correct so that you are moving your company into a position where you can start getting some yeses.
Decide what will make you happy and commit 100% to doing it: One of the things Ben said he learned early on was that while being an entrepreneur meant that he had control over what he was building and doing, it also meant that he lost control over a number of things like a steady paycheck or the resources of a large organization. But, ultimately, the tradeoff was worth it for him to keep going. His advice to the group may seem simple and obvious, but it can be hard to follow! He was convincing – you have to find what makes you happy, because ultimately, that is the person you have to answer to first. Building a startup is really hard, but if you are doing something you love or building a product you are passionate about, it is one of life’s greatest rewards.
Foster your co-founder relationships: Like any relationship, you are going to have some ups and downs as founders so it’s important to foster a good, highly communicative relationship with your co-founder(s) so that you can make it through those rocky days. Again, it may seem fairly straightforward, but it is one of those things that requires consistent attention and can make all the difference.
Recognize what you don’t know and tackle it head on: This was less of a tip and more of an anecdote that Ben shared, but one that I thought was worth mentioning. Every weekend, he reads a different business book in an effort to hone his business, marketing or technical skills. Having a ready appetite to learn and grow as a person and a leader is no doubt a part of Pinterest’s secret sauce and something I encourage any entrepreneur to foster throughout their careers.
It was tremendous to have so many alumni back at Lightspeed and thanks again Ben for your time and thoughts.
Tags: 2012, appliance, converged, data center, datacenter, enterprise IT, funding news, infrastructure, Nutanix, storage, virtualization, vmware
Over the past two years we’ve seen a lot of disruption in the enterprise storage market with everything from the game-changing performance of flash to next-generation storage architectures required to support the cloud and virtualized data center environments
And notwithstanding some early wins from companies like Fusion-io, we believe the underlying data center compute, storage and networking transformation is still in early innings of playing out. A case in point is Nutanix, a company worth paying attention to and one where we recently led an oversubscribed growth round of financing.
So why are we so excited about Nutanix? We believe the company represents the next-generation of IT infrastructure – a CONVERGED storage and compute platform uniquely able to cost-effectively power the datacenters of today and tomorrow.
Nutanix combines enterprise-class compute and storage resources into a single, inexpensive x86 system. It also incorporates elastic scale-out technologies that have historically only been available to some of the world’s largest, and most technically sophisticated companies like Google and Facebook. Now, for the first time, this revolutionary computing paradigm is being delivered to mid-range and large enterprises that are also looking to ride the disruptive economic wave afforded by cloud computing and large-scale virtualization.
The Nutanix magic is in its’ software which is highly sophisticated and delivers the world’s first SDS (Software-defined Storage system), similar to Nicira, also a LIghtspeed portfolio company, which built the world’s first SDN (Software Defined Network system). The combination of SDS, inexpensive compute, and a radically simplified appliance form factor which is easy to deploy and manage has customers excited and highly engaged. They are calling Nutanix Complete the world’s first “datacenter in a box.”
Beyond the technology, we have been hugely impressed by the team at Nutanix. We’ve had the privilege of working with them from the earliest days of the company when Lightspeed originally lead the Series A financing more than two and half years ago. The founders came to us with an extremely bold vision to redefine datacenter storage and computing and we’re incredibly excited to see how emphatically the market is now embracing this vision.
If you found this post useful, follow me at @rmtacct and follow Lightspeed at @lightspeedvp on Twitter.
People still can’t do math – and what it means for pricing August 19, 2012Posted by jeremyliew in Ecommerce, pricing.
add a comment
Definitely worth reading and experimenting with if you’re in ecommerce.
Tags: LTV, model, retention, subscription
We’ve posted before about how to estimate lifetime value (“LTV”) for an ecommerce business and for a subscription business, and have provided a sample cohort analysis for each (ecommerce and subscription). This is one of the most important factors in understanding unit economics.
Recently, Eric Liaw sent us a very interesting May 2006 paper entitled “How to Project Customer Retention”, authored by marketing professors Peter Fader (Wharton) and Bruce Hardie (London Business School) and published in the Journal of Interactive Marketing in 2007. In it, the professors explain how previous attempts to project retention rates using line-fitting regression models failed, even after introducing quadratic or exponential functions. Since we had advocated essentially using an exponential line fit for subscription LTV estimation, we figured it was worth reading. The authors show that exponential form fitting is too conservative and underestimates actual retention rates.
Professors Fader and Hardie decide to start from scratch with a simple assumption: what if each customer has a fixed probability of renewing his or her contract at the end of each period? So if I’m a big movie fan, let’s say I’m 80% likely to renew Netflix each month, but you’re caught up on Breaking Bad and only 30% likely to renew each month going forward. Probability varies by customer, but each customer’s rate remains constant over time.
It turns out that, based on probability theory, this simple assumption implies that the distribution of renewal rates can be characterized by a statistical model. Over time, the difference in each individual’s probability to renew suggests that individuals with lower renewal probabilities will generally drop out before those with higher probabilities. Incidentally, this also explains why incremental retention may appear to improve over time, when it’s actually a likely side effect of the remaining customer mix.
After some mathematical gymnastics, the authors unveil the model they’ve derived: the shifted-beta-geometric distribution. The authors tested the model by using the first seven years of data from a given sample to project renewal rates at the end of the final five years in the sample. The model proved to be quite accurate, within 3% of actuals, and much better than linear or exponential form fitting.
A few quick caveats: this model is appropriate only when the data reflects a discrete renewal period, such as a defined monthly or annual cycle. Also, the model should be reserved for projecting behavior in contractual settings, such as subscription renewals and other observable customer exit points, rather than ecommerce or other businesses where the customer can remain dormant for long periods between orders.
We’ve uploaded a spreadsheet here, along with directions for how to use it yourself.
Hope this is helpful. We look forward to hearing from you regardless, but especially if:
1) You use the model and have any feedback on results
2) Your company uses any other methods to capture, analyze, and project customer retention
3) Your innovative company achieves valuable unit economics. As previously mentioned, we like to see LTV / Customer Acquisition Cost > 2.5 and payback periods under 12 months.
If you found this post useful, follow us @lightspeedvp on Twitter.
Investing in Loyalty August 2, 2012Posted by peternieh in 2012, Consumer internet, startups.
Tags: consumer, funding news, loyalty
1 comment so far
“Customer loyalty is the single most important driver of growth and profitability. – Harvard Business Review
Today, we are excited to announce our investment in FiveStars. Founded by Victor Ho and Matt Doka, FiveStars makes it easy, and affordable, for retailers and merchants to reward their most valuable customers. It is the first loyalty offering that integrates directly with over 90% of existing point of sale (POS) systems and is already being used by hundreds of merchants.
So why loyalty and what makes this company and market interesting to us?
In the face of economic pressures, consumers are thinking more carefully about every purchase, and retailers now face more competition than ever. However, smart retailers are facing these challenges and actually growing their businesses, and bottom lines, by retaining their most loyal customers. A recent student by Harvard Business School found that a 5% increase in customer retention yields an increase in profits between 25 – 100%.
The key, however, is how to retain these customers without increasing complexity and costs. As a former loyalty consultant at McKinsey & Company, Victor Ho has a keen understanding of the challenge that retailers face and has delivered a product that not only meets those needs, but also does it in a way that works seamlessly within their existing business structure and is frictionless for consumers to adopt. It literally “slides in.”
FiveStars offers consumers a single card that they can use to earn rewards for everything from picking up coffee to getting a massage without the hassle of keeping track of multiple cards. A consumer registers once by just giving their phone number and then simply provides the card to merchants on checkout. And because the company integrates directly and easily with the POS, retailers can be up and running with FiveStars in literally minutes. No extra equipment is required, like iPads or smartphones which add to complexity and decrease adoption. Furthermore, Five Stars allows merchants to track spending habits and better personalize promotions and rewards.
The proof is in the pudding, and the company has already signed up several hundred merchants in its first several months of selling with very little marketing or advertising. They have developed a winning formula and with the new capital will be looking to accelerate their go-to-market activities.
It’s a great product, built by a great team and addressing a huge market. We couldn’t be happier to partner with them as they think big and move fast.
If you found this post useful, follow us @lightspeedvp on Twitter
A Tremendous Day for Nicira And The Networking Industry July 23, 2012Posted by John Vrionis in Uncategorized.
Tags: infrastructure, networking, Nicira, SDN, virtualization, vmware
With their acquisition of Nicira today, VMware is making a brilliant strategic move that gives them not only the leading network virtualization technology but also a world-class of team of executives and engineers. Congratulations to Martin, Steve, Rob, Alan, JJ, Paul, Denis and the entire Nicira team – for me, and the rest of the Lightspeed team, it’s truly a privilege to have been a part of the effort where an entrepreneurial team realizes a vision and begins to transform an industry.
Nicira is fundamentally changing networking as we know it, much in the same ways that server virtualization changed the datacenter only a few years ago. The company virtualizes the network, separating the logical network from the physical topology, much like server virtualization decouples the virtual machines from the underlying server hardware. With Nicira, networks can have the same dynamic and flexible operational model of virtual machines and they can be programmed and configured without disruption or manual intervention.
So while today’s news will likely center on the financials of the deal (which admittedly are tremendous for all shareholders) and how the two companies will integrate, to me, the real story is the impact that Nicira has had on the industry in just a few years and the team behind it. These are truly some of the best talents in networking and infrastructure, more broadly, in the world. The spotlight should shine brightly and entirely on them as they’ve done all the hard work to make this happen.
On a more personal note, I’ve known Nicira’s co-founder Martin Casado since 2004 when we were both studying at Stanford. He was getting his PhD and I was in Business School; we were introduced by Andy Rachleff and Nick McKeown and have been friends ever since. Martin is a genius, really. That word tends to be overused, but in Martin’s case it’s the absolute truth. You don’t have to take my word for it, Scott Shenker, a co-founder of Nicira and UC Berkeley computer science professor recently told WIRED, “I’ve known a lot of smart people in my life, and on any dimension you care to mention, he’s off the scale.”
Congratulations guys, thanks again for all the effort and for letting me be a part of it, and of course to VMware for a great move.
1 comment so far
I’m a big fan on focusing on getting the “copy” (the words on the page) right to drive behavior. I’ve posted in the past about Cialdini’s great book, Influence, The Psychology of Persuasion, and how the principles outlined in it can be used for structured brainstorming to improve copy to drive the results that you want. Of course, all this needs to be A:B tested, but it provides great ideas to test.
Sunday’s NY Times has a great story on how behavioral science can help drive policy, and how a change in copy helped increase tax collection in the UK by 15 percentage points:
One early success story involves an attempt to collect taxes from people who fail to pay on time. Most British citizens pay their taxes promptly because it is a simple tax system with few deductions, so that most taxes are collected via payroll withholding. (That’s “make it easy” in action.) But small-business owners and individuals with significant nonpayroll income are expected to save up the money to write a check to the government, and some of them fail to pay on time.
In such cases, the government’s first step is to send a letter asking for payment within six weeks, after which sterner, more expensive measures are taken. The tax collection authority wondered whether this letter might be improved. Indeed, it could.
The winning recipe comes from Robert B. Cialdini, an emeritus professor of psychology and marketing at Arizona State University, and author of the book “Influence: The Psychology of Persuasion.”
People are more likely to comply with a social norm if they know that most other people comply, Mr. Cialdini has found. (Seeing other dog owners carrying plastic bags encourages others to do so as well.) This insight suggests that adding a statement to the letter that a vast majority of taxpayers pay their taxes on time could encourage others to comply. Studies showed that it would be even better to cite local data, too.
Letters using various messages were sent to 140,000 taxpayers in a randomized trial. As the theory predicted, referring to the social norm of a particular area (perhaps, “9 out of 10 people in Exeter pay their taxes on time”) gave the best results: a 15-percentage-point increase in the number of people who paid before the six-week deadline, compared with results from the old-style letter, which was used as a control condition.
The tax authorities estimate that this initiative, if rolled out across the country, could generate £30 million of extra revenue annually. And note that sending an effective letter doesn’t cost any more than sending a bad one.
If you’re in product management and you haven’t read Cialdini’s book, go out and buy it now. Nine out of ten product managers already have! 😉
Pretty interesting article in the current edition of the Economist about the psychology of discounting:
A team of researchers, led by Akshay Rao of the University of Minnesota’s Carlson School of Management, looked at consumers’ attitudes to discounting. Shoppers, they found, much prefer getting something extra free to getting something cheaper. The main reason is that most people are useless at fractions.
Consumers often struggle to realise, for example, that a 50% increase in quantity is the same as a 33% discount in price. They overwhelmingly assume the former is better value. In an experiment, the researchers sold 73% more hand lotion when it was offered in a bonus pack than when it carried an equivalent discount (even after all other effects, such as a desire to stockpile, were controlled for).
This numerical blind spot remains even when the deal clearly favours the discounted product. In another experiment, this time on his undergraduates, Mr Rao offered two deals on loose coffee beans: 33% extra free or 33% off the price. The discount is by far the better proposition, but the supposedly clever students viewed them as equivalent.
73% higher sales is an astonishing number that comes simply from positioning the same discount differently. Of course, this only helps if you are making a positive contribution margin on the sales!
This reminds me a bit of Prize-Linked Savings accounts, basically savings accounts with a lottery ticket attached (that is bought by slightly lowering interest rates):
One way to think of these “prize-linked” accounts is that they can offer an expected market return, but in an innovative way. They pay a guaranteed return below market interest rates, but also provide a lottery ticket whose value makes up the difference.
To be specific, a lottery-lined savings account could offer a lower rate of interest, but also say a one-in-a-million chance of winning $1m for each $100 deposited. Mathematically, the expected return is the same, but the chance to win $1m makes the account much more attractive.
Britain has historically led the way with these sorts of savings opportunities, starting with the “million adventure” lottery in 1694. Households were offered 100,000 tickets at £10 each, with poorer groups able to club together to buy fractions of tickets. Holders received a 6 per cent annual return for 15 years, plus the opportunity to win a prize of between £10 and £1,000. Historians suggest the programme was popular and successful. More recently, much the same theory was seen in the UK’s Premium Savings Bonds, which offer the opportunity to win a prize but no base interest rate. From Brazil to Germany, Mexico to New Zealand, a variety of other prize-linked savings opportunities already exists.
This is another example where reframing the same economic returns can change user behavior. I wonder if framing a 33% off sale so that people buy at full price but with “Every third purchase free” might increase sales overall. Has anyon had any experience with this?
UPDATE: Recently found a really fun and relevant post on pricing experiments that is worth reading from conversionXL.