Archive | SaaS RSS feed for this section

Predictions for 2016: Self-Driving Trucks, AI, and Brain Monitoring

8 Jan

This post originally appeared on Xconomy here.

Whether we have been in a tech bubble or not is frankly not that interesting. What is interesting is that the foundation for innovation is as strong as we’ve ever seen and entrepreneurs are bringing the future to reality at an amazing pace. Here are a few of my predictions for what we’ll see in 2016:

1. Self-driving vehicles hit the road for real, led by commercial trucks on interstate highways. So far most conversations around self-driving cars focus on personal vehicles. It’s unlikely we’ll take a fully autonomous car for our daily commute for quite a few years, but commercial trucking will see self-driving vehicles emerge far sooner. One company, Peloton, is already making this a reality. Long haul trucking is an easier technical challenge because interstates are generally straight lines, well-marked, digitally mapped in high definition, and generally free of pedestrians, bicycles, and other random obstacles; and the economic productivity of trucks can justify substantial investment in sophisticated cameras, sensors, and computers needed for autopilot systems. The economic need for autonomous trucks is huge due to the high cost and shortage of drivers, regulatory limits on driving time, and the fuel efficiency gained from convoys travelling close together in peloton formation.

2. Artificial intelligence will improve by leaps and bounds, and so will the way we interact with it. At first Siri frustrated us with its faults, and Google Now annoyed us with random cards on our screen, but I’ve noticed that recently both systems have gotten much broader and more accurate. Most of us barely scratch the surface in terms of their capabilities. In 2016 we will become more accustomed to interacting with AI systems of all kinds that are more natural, comfortable, and intelligent. Just as it took time for us to get used to self-checkout stations at the grocery store and the early days of voicemail were profoundly awkward, societal norms will need to adjust, and designers will need to create better user experiences for us to accept without pause that we’re interacting with an AI system and not a human. As we get better at these interactions, the AI technology gets smarter. Our digital assistant will ask nuanced questions (e.g. “did you mean the fruit or the company?”) to ensure it’s answering correctly. Smarter AI will enable automation of the rote components across a huge spectrum of jobs categories including dietary/fitness training, customer service, financial advice, education, and medicine, freeing up humans to focus on the most value added components of these occupations. As our expectations for human to computer interactions continue to grow, AI systems will rise to the challenge.

3. Wearable brain-monitoring devices for mindfulness training will become mainstream. It’s rare to see a person without a smartphone glued to his or her hand in almost any setting these days, though many people are becoming aware of the downsides of digital addiction and its effect on mental and physical health and our relationships. Couple this with growing public interest in meditation, yoga, and digital detox, and we’ll see mindfulness training become front and center in 2016 in wearable device form. Wearable fitness devices like the Fitbit or Apple Watch have become part of everyday life and early pioneers in this space have developed devices to monitor brainwaves, such as the Muse Brain Sensing Headband. The irony of a digital device helping with meditation and mindfulness may make you cringe, but brain training is hard and feedback essential and self-quantification of progress almost impossible until recently. Brain wave sensing devices will improve as our awareness and need for them grows, and they’ll soon become as mainstream as heart rate monitors.

Regardless of what the financial markets do in 2016, innovation will continue at fever pitch and cool new products and technologies will become newly indispensable parts of our work and lives.

Advertisements

Why Beckon Beckoned to Me: The Arrival of Marketing Performance Management

27 Jan

Beckon 1

In today’s online and mobile world customers are educating themselves about products and services long before any flesh and blood sales persons utters a word. Consumers can easily find their own way to detailed product information, user reviews, professional reviews, demonstration videos, user generated unboxing videos, sales rankings, price comparisons, social media sentiment, buying guides, and more. For this reason, marketing is more important than ever and the Chief Marketing Officer has never been more powerful nor controlled more budget, for both media and technology.

At the same time, however, the job of a marketing leader has never been harder or more stressful. The digital landscape has become so vast and dynamic that the marketer must master an endless parade of new channels. Just as they were getting used to Facebook, Twitter, and Instagram, along comes Pinterest and SnapChat, and undoubtedly the next hot channel is just around the corner. With an expanding universe of marketing channels comes an ever increasing volume of data. With all this data available, CMOs are expected to quantify their results with the precision of a CFO. While each new channel begets a host of new adtech and marketing tools to help the CMO manage campaigns, measure performance, and optimize results, each of these solutions produces data and reports in their own unique format.

Beckon 2

It has gotten to the point that when the CEO asks “how is our marketing doing?” it strikes fear in the heart of the CMO. This perfectly benign question usually kicks off an all-night exercise of cutting and pasting data and charts from various marketing execution systems into a lengthy presentation to answer the CEO’s question. When PowerPoint (now in its 28th year) is the tool of choice for marketers to aggregate and translate performance data from their various systems you know the situation is dire. Making matters even worse is the fact that many large brands can’t even access their own campaign data as it is held hostage by their various marketing agencies. Not only does the client get charged a markup by the agency for report requests, but it’s the classic case of the “fox guarding the henhouse” to ask an agency to report on their own performance.

Several years ago, a former marketing leader from a Venrock portfolio company did a stint as an Entrepreneur-In-Residence in our offices. She identified this lack of a single “CMO dashboard” integrating data from various point solutions as a problem she had experienced firsthand. Essentially she wanted a “System of Record” for marketing. After several months of researching potential technical solutions she concluded that, despite the crying need for such a product, building one would be too difficult. She gave up frustrated and joined another best of breed marketing tool company. This unsolved problem stuck in my head.

A few years later I met Beckon. The team at Beckon have been marketers, built marketing point tools, worked in agencies, and have built, installed and used systems of record in other enterprise functional domains such as finance and sales. Having seen the problems facing modern marketers firsthand they have taken a novel approach to building a system which can pull in data from over 100 different marketing point tools. While some of this data is available via well supported APIs, much of the data comes in via spreadsheet imports and email parsing (think TripIt.)   The next thing Beckon does is normalize the data so that marketers can compare different campaigns across different channels with one common taxonomy. They allow the marketing team to add metadata such as geographies or regions, product names or categories, customer segments, agencies, objectives, and so on, in order to put the marketing results in appropriate context. They allow for What If analysis, planning, and time series tracking. Beckons creates beautiful visualizations and answers to plain English questions that don’t require analysts skilled in SQL queries. And because this is not a BI tool, but rather an application built “by marketers, for marketers”, it is loaded with best practices for omni-channel marketing performance management right out of the box with no IT Department involvement.

Beckon 3

Over the past year some of the best brands in the world have adopted Beckon. Coke, Microsoft, GAP, and BSkyB are among the clients using Beckon to manage their omnichannel marketing. The real sweet spot for Beckon are mass marketers and indirect sellers who spend across at least five different channels. While I have written about my keen interest in predictive and intelligent software, the truth is that relevant, advanced modeling is only possible if the data sets the models are built on are comprehensive, normalized and continuously updated. Finance, for example, measures its performance according to GAAP (Generally Accepted Accounting Principles), a consistent, agreed-upon methodology shared within and across companies. As a result, we can understand a company’s financial performance quarter to quarter and compare performance to other companies in a standardized way. Marketers have never had a similar system. That’s what Beckon finally brings to marketing – a strong, united data foundation upon which all kinds of consistent, robust marketing analyses can flow – benchmarking, planned versus actuals, test and control, lift over baseline calculations, econometric (mix) models and more. Beckon gives marketers self-serve access to many of these analyses within its application and can also flow its standardized, merged and continuously updated data sets to advanced analytics teams and tools.

Marketing can finally have its own system of record the way sales has Salesforce, manufacturing has SAP, Finance has Oracle, and HR has Workday. Beckon is Marketing Performance Management. Finally the CMO does not have to hide when the CEO calls (or beckons) them to their office.

Beckon 4

The Future of Software is…Wicked Smaaht

18 Apr

smarter software

Software as a Service and cloud computing has been transformational for the software industry.  But compared to what is coming next, you ain’t seen nothing yet.  First, to appreciate where we are heading a quick review of where we’ve been is in order.  Back in the olden days of business software a software company sold you an application which you installed on your servers and desktops which made business processes more efficient, facilitated workflow, and sped up information retrieval.  As you used it this software accumulated data such as your customer records, financial results and manufacturing statistics.  If you wanted to deeply analyze this data for trends and insights you bought Business Intelligence or Analytics packages from a different set of software vendors so you could slice and dice your data, generate reports for executives, and hopefully decipher interesting trends about your business that you would then go act on.  In the early 2000s Software-as-a Service companies emerged and enabled you to “rent” business applications, rather than buy them, as your employees accessed them through the Internet and their web browsers.  This came with many advantages in total cost of ownership and manageability, but fundamentally most of the first SaaS applications were about workflow and data storage/retrieval just like their on-premise software forefathers.  In the last few years we’ve had a “Big Data” explosion and a host of new open source technologies like Hadoop, MapReduce, and Cassandra packaged by a set of new companies that help businesses manage and manipulate their ever expanding mountains of data.  Also emerging is a new generation of cloud based analytics companies that make it easy to slice, dice, and visualize big data sets.

So what’s the point of this history lesson?  The point is that, for the most part, all of these business applications and more recent Big Data tools have left the burden of capturing real business insight, making decisions, and taking action on the business customers themselves.  In essence, if you wanted real business value you had to create that value yourself by getting your employees to use the applications (which often means manual data input), have analysts mine and interpret the data, and ask managers and executives to make decisions based on what they see in the reports and charts.  For example, if your company used Sales Force Automation, whether it be Siebel on premise or Salesforce.com in the cloud, your sales reps had to diligently input data about their sales calls and management had to be smart about logging in to read the reports, suggest actions for each account and discern broader trends across the data.  A new breed of software company is emerging, however, that combines data science expertise with deep understanding of business problems.  I call them Data Driven Solutions.  These solutions use algorithmic data mining, not only on your own data but often on external third party data sets accessible by cloud ecosystems and APIs.  Data Driven Solutions make predictions about business functions, prescribe what to do next, and in many cases take action autonomously.  Trained analysts are not required to query databases but rather business users get answers directly from the software.  These answers typically feed seamlessly into the flow of business activity, often invisibly.  While this distinction may seem subtle, I believe it is fundamental and disruptive, and represents the future of software.  This is in no way the end of the SaaS, but in fact where SaaS is going next and presents massive opportunity to new SaaS innovators and a potential threat to incumbents who do not adapt.

Data Driven Solutions

8 Suggestions for Building Data Driven Applications

Think Moneyball, for everything.  Billy Beane of the Oakland A’s defied the conventional wisdom of traditional baseball talent scouts by recruiting players other teams underappreciated but whom he believed represented great return on investment.  He did this not by relying on his own brilliant sense of which players to recruit but by letting a math whiz run regression analysis on player statistics to figure out which lesser heralded stats were most predictive of winning baseball games.  The math predicted results and told him which players to acquire, predictions which Beane followed to his team’s competitive advantage.  Opportunities to apply this approach in business are practically everywhere.  6Sense* is a new SaaS company that analyzes B2B website traffic and third party data to predict which prospects are most likely to buy from you, what they will buy, when they will buy, and how much they will buy.  Like Beane, they don’t rely on rules of thumb in scoring prospects, such as whether the prospect downloaded a white paper, viewed lots of product detail webpages, or has “Procurement” in their job title.  6Sense has found that these heuristics yield only about 50% accurate forecasts which is not enough to compel a sales person to trust the results.  Instead, 6Sense uses a variety of machine learning statistical models to uncover the unexpected correlations which drive predictive accuracy up to 85-90% accuracy which definitely gets a sale rep’s attention.   Instead of being a chore to use like Sales Force Automation, 6Sense tells sales reps how to close more deals and earn more commissions.  Infer, Lattice, and C9 are also innovating in the area of predictive CRM solutions.  Use your domain expertise to figure out what problems to solve for your customers, but let the data lead to new and unexpected insights.

Build in Data Learning Loops   Google enjoys a very powerful form of Network Effect.   The more searches they run and resultant clicks they see the better they understand the intent of what a searcher wanted to find.  This makes their search algorithms better which earns them more user searches, which keeps the search quality/volume flywheel spinning.  This notion of a “learning loop” can be applied to many business settings as long as you find a way to “close the loop” and see how your prediction or answer actually fared.  For example, AppNexus* is an AdTech company that operates an exchange where publishers and ad networks on one side are matched by algorithms with advertisers and agencies on the other side to put the right ad in front of the right audience at the right time.  Learning loops are built in to the bidding and optimization algorithms which get the chance to learn from their results more than a billion times per day.  Data Learning Loops are powerful sources of competitive advantage akin to natural monopolies for those who achieve greatest scale.

Don’t Just DescribePredict and Prescribe   Some may ask whether Data Driven Solutions are just a new name for Business Intelligence.  I don’t think so.  Analytics packages mostly describe what is happening by sorting and filtering your data to show sums and averages and trend lines in tabular or graphical format.  Data Driven Solutions go much further by using the data to make predictions and even prescribe or execute actions.  A good example is the retail industry and Point of Sale results.  POS data is the basket by basket, sku by sku, store by store sales results that are collected across hundreds of thousands of retail outlets daily.  Nielsen has been compiling this data for decades and batch processes data sets for retailers and their vendors to study on a monthly basis.  How those vendors and retailers derive value from the data is up to them.  Retail Solutions* is a data driven solutions company which also gets POS data from retailers and shares it with vendors, but on a near real-time or daily basis.  More important than freshness of the data, however, is that Retail Solutions offers predictions and prescribes actions as solutions to business problems.  RSI doesn’t just create reports, they predict when you will be out of stock on a given SKU and sends mobile alerts to shop clerks, distributors and store managers to make sure the shelves stay full.  This is one of over ten predictive and prescriptive applications they provide in order to maximize return on investment for their customers.  Pretty charts are not enough.

Data is not the point, Focus on Solutions   Lots of companies market themselves as “Big Data” companies, but unless you are selling to IT Departments whose problems actually include managing lots of data, most business customers don’t really care about data.  They care about solving business problems.  Athenahealth* helps doctors get paid faster.  Turns out cashflow is really important to doctors and as a result Athena has grown very quickly and is now one of the largest SaaS companies.  Doctors don’t care how Athena actually does what they do, which happens to involve statistical analysis of massive amounts of insurance claims data and heavy use of learning loops.  The team at Athena deeply understands healthcare and doctors and so astutely markets themselves as a complete solution to real problems, and resist pounding their chest about how smart they are at Big Data.  In fact, the word “data” does not appear even once on their homepage.  Smart move.

Horizontal Strategy: Solve New Problems in New Ways   Providing applications for horizontal business functions like sales, finance, or human resources that function similarly across many industries represents very large opportunities because the market sizes are huge.  As a result there are powerful SaaS incumbents, such as Salesforce.com, Netsuite, and Workday, in each of these functional domains.  As you would expect, these players are starting to add data driven application intelligence to their offerings.  Fortunately for startups the challenges businesses face are constantly changing thus creating opportunities to be the first to solve new problems with a new approach.  In the realm of marketing, for example, “Content Marketing” is the hottest new trend and is the digital marketing approach seeing the greatest increase in budget allocation.  Yet marketers are highly confused as to what content to produce, how to produce it, where and how to distribute it, and especially how to measure ROI.  Captora is a young startup that has jumped on this new problem with data and domain expertise and is seeing rapid growth and using their head start to establish a beachhead before direct competition comes at them.  Knowing the experience of the team they won’t be resting on the side of the road but rather racing ahead to broaden their solution in synch with new challenges facing modern marketers.

Vertical Strategy: Feed the Starving   Providing deep solutions in specific industry verticals like healthcare, entertainment or education can be a huge opportunity.  This is especially true in industries where data has largely been non-existent or hard to access as has been the case in the three industries I just mentioned.  If a Data Driven Solution can access, interpret, or create new data and use it to solve a big problem the market reaction can be like a starving person being offering a hot meal.    Castlight Health*, for example, solves the problem that in healthcare it is generally impossible to know what a given service (an office visit or a test for example) will cost until after you’ve consumed the service and you receive your bill 30 days later.  It turns out that the variance in pricing for even a commodity service like an MRI test can be 5 to 10x in a given 5 mile radius.  If one could know the price difference ahead of time they can consume intelligently–as we do in most other shopping situations.    Large employers, who tend to be self-insured, really like the idea of helping their employees spend less on healthcare as those savings drop straight to the bottom line, and as a result some of the largest employers in America have adopted Castlights’ solution.  Customers like CVS Caremark, Microsoft, and Wal-Mart don’t really care about the big data blahdy blah that Castlight uses to come up with their solution, they just know they are starving for ways to lower their employee health care costs and Castlight has an effective solution.

Vertical Strategy   While some industries are just getting their first taste of Big Data, others have been sophisticated handlers and miners of Big Data for a long time, such as the investment industry, airlines, and eCommerce.  In those fields a small incremental advantage afforded by a data driven vertical solution can be extremely valuable.  DataMinr* is a company that transforms the full Twitter stream of public tweets using sophisticated math to discern important news events amid all the noisy babble as quickly as possible ahead of the media.  Investment hedge funds will pay handsomely for incremental advantage and getting a jump on news that might move the market or a particular stock is something they are eager to buy even amidst all their number crunching sophistication and home grown solutions.  On April 23, 2013 when the stock market “Flash Crash” occurred based on a rumor that the Whitehouse was under attack, Dataminr’s algorithms figured out the attack was a hoax a full two minutes before other new outlets and their clients were able to act on the news ahead of the market’s rapid recovery from the severe dip the rumor had caused.  It turns out that news agencies like CNN, which typically rely on human reporters and shoe leather to beak news, have also turned to Dataminr as a solution to their problem.  Dataminr thus serves both a very sophisticated big data segment, investment funds, and an industry at the opposite end of the data automation curve, the news industry, with a solution that simply could not have existed until very recently.

Consumer solutions can be driven by data too  Using Uber is a magical experience.  Push a button on your phone and a car appears within an instant to take you where you want to go—no hailing, no reservations, no need to reach into your pocket for payment, and remarkably little waiting for your ride to arrive.  If Uber simply sent messages to available drivers about customers needing rides the system might still be good but customers would have longer wait times, which wouldn’t be as magical.  Instead, Uber uses statistical analysis on data coming from their drivers and riders to predict where demand will be highest and recommends that drivers congregate there to be ready for ride requests.  Nowhere in their marketing does Uber talk about data or “quantifying” your ride patterns—consumers don’t need to know how the magic happens as long as their ride shows up quickly.  Similarly, Better Finance* makes secured loans to consumers with low or no credit so they can buy smartphones and other high-value items.  Better Finance can do this at rates far less than payday lenders because of their data driven underwriting and feedback loops coming from high loan volumes and thus their underwriting algorithms constantly improve—to the benefit of Better Finance and their customers.  Opportunities to create consumer solutions enabled by big data are everywhere…just don’t mention the word data.

Traditional SaaS and on-premise software will be around for a long time in the future and these vendors will add more and more data intelligence to their offerings.  They will be joined however, and possibly threatened by, a new generation of nimble and innovative next generation SaaS companies that will combine data and domain expertise to add massive business value to their customers.

I look forward to meeting and helping as many of those companies as possible.

*Venrock is an investor in these companies.