Solve the Product Maze Backwards

As the father of young children, I can tell you that there is a special place in my heart for restaurants that provide puzzles and crayons for small children to pass the time.

On a recent trip out to The Counter in Mountain View, Jordan (who is 8)  was really struggling with a large maze puzzle on one of these activity sheets. It was a fairly large maze, and he was frustrated by his inability to see the dead ends ahead, forcing him to retrace his somewhat tortured crayon path.

I told him to try to solve the maze backwards.

As you can probably guess, he began at the end, and was able  to find a path back to the beginning in just a few seconds . He was delighted, and a bit surprised, to see how simple the puzzle looked like from a different perspective.

Surprisingly, I find that both entrepreneurs & product leaders miss this important lesson when evaluating ideas for either their company or their products.

Three Questions in Product Prioritization

In my experience, there are three common questions that often come up when product features are being debated:

  1. Should we build this?
  2. When should we build this?
  3. How should should we build this?

Unfortunately, even highly talented teams can become  get bogged down in debate and uncertainty when all of these questions become entangled. As engineers & designers are professionally trained to answer the question of “How,” the worst debates tend to happen around the questions of  “Should” and “When.”

Too often, when debating what feature to work on next, debates around timing quickly devolve into debates about whether the feature is needed at all.

Solving the maze backwards does a fantastic job of disentangling these two questions. Simply asking the question of “If we are successful, will we have this feature in 3 years?” tends to illuminate whether the debate is about “Should” or “When.”

If the answer is yes, you will have that feature, then the question is simple. You are just debating priority.

Avoid the Local Maximum

One of the well known issues with iterative processes for delivering product features is the “local maximum” problem.

The assumption is that where ever you start with your product, your team keeps working on improvements. Each improvement is measured to ensure it is “better” than the product before the change. However, you can reach a point where every change you make hurts the metrics that you measure. The fear is that there is a better version of your product (the absolute maximum), but it requires a change bigger than you can get to from the current design.

It’s called a local maximum problem because of the similarity to the concept in mathematics when you are traveling along the curve. From the local maximum, every move is down, even though the curve ends up higher eventually.

Solving the maze backwards can help.

By asking the simple question about whether or not your product in the far future has a given capability, it can unblock your thinking about what leaps and changes will be necessary. Whether the limitations are in technical architecture or product design, clarity on your long term vision can help your team visualize a future not trapped by their current constraints.

Too often, the real limitation is not related to either technical or design constraints, but rather a lack of clarity and imagination about what might be possible. Just like a maze, it is easy to get lost in the middle. Thinking backwards from the end goal can help the team escape a Zeno’s paradox of minor feature improvements.

Founders Can Solve the Maze Backwards, Too

It may seem hard to believe, but in early 2009 when I took over LinkedIn’s mobile efforts, there was still active debate within the company about whether to dedicate significant effort to mobile. Why? Well, back in 2009, the Blackberry was still hitting record sales, the  app store was a year old, and from a web metrics point of view, mobile views represented less than 1% of LinkedIn’s traffic. Like every hypergrowth startup, LinkedIn had a huge number of initiatives it wanted to pursue around growth, engagement & revenue, and it wasn’t obvious that mobile would move any of these needles for the company in the next few years.

Solving the maze backwards helped.

What was fairly obvious in 2009 was that the growth rate of mobile engagement was compounding at a phenomenal rate. LinkedIn, as a professional use case, might have been slightly behind social use cases for mobile adoption, but it was fairly clear that within 5 years (by 2014), mobile should represent a majority (over 50%) of all visits to LinkedIn.

Thinking backwards helped give us the confidence to invest in both talent and technology that had little short term payoff, but would become essential to engagement over the next five years as those predictions came true.

Fast forward to 2017. I was recently meeting with a founder who was debating whether they should hire a Vice President of Marketing. As he walked me through his thinking, the argument wandered, and became more focused on whether or not the company “needed” marketing.

I asked him if there was any way, if the company hit their numbers over the next three years, that the company would not need marketing, or an experienced marketing leader?

The CEO quickly responded that marketing would be essential to hit the numbers they were looking for in three years. All of a sudden, the conversation changed. The question wasn’t whether or not to invest in marketing, but more a question of when they need to.  Was this a 2017 or a 2018 problem? Is this something they would need to hit the milestones to raise their next round of funding, or something that they would invest in during the next cycle?

It was now a question of when.

Questions of “Should” vs. Questions of “When”

“The essence of strategy is choosing what not to do.” — Michael Porter

Being clear about what your product will and won’t do is a critical element of product strategy. However, because it is so important, even well-meaning teams can turn almost any feature into an existential debate.

Thinking backwards can help differentiate questions of “should” from questions of “when,” and that can be incredibly productive in moving the discussion to prioritization.

This is not intended to be dismissive of questions of prioritization. Phasing decisions are some of the most important decisions start ups make. Financing for startups is phased. Small teams can only work on a few projects at a time. Customers can only absorb so many new features at once. As a result, prioritization decisions are incredibly difficult to make.

Greedy algorithms are very good, but can be traps if you are working against competitors and an ecosystem that is willing to make bets that lie across the gap from your product’s current local maximum. Thinking backwards can help illuminate long term goals that are across the gap.

When you are building a product roadmap, and get stuck on debates about a short term feature that doesn’t move the numbers, I encourage founders to take a moment and try to solve the maze backwards.

It worked for Jordan, right?

Helping People Save is a Job Worth Doing

“Every day stuff happens to us. Jobs arise in our lives that we need to get done. Some are little jobs, some are big ones. Some jobs surface unpredictably. Other times we know they’re coming. When we realize we have a job to do, we reach out and pull something into our lives to get the job done.” — Clay Christensen

In the summer of 1993, after declaring computer science as my major, I got my first high paying software development internship. Over that summer Hewlett-Packard paid me over $5,000, which seemed like an unbelievable amount at the time.

Unfortunately, like a lot of people, I was so excited by receiving this windfall that I promptly spent it. By Thanksgiving, I was shocked to find that my bank account was nearly empty. All that money, gone. It literally sickened me.

That was the moment when I decided to learn as much as I could about personal finance and I got religious about saving.

The Theory of Jobs to Be Done

For a lot of people, there is a moment they can recall when they consciously decided that they wanted to start saving.

When I attended Harvard Business School at the end of the dot-com era, I was incredibly fortunate to spend time with Clay Christensen, who at the time had just recently published the now famous book, The Innovator’s Dilemma. In his class, we studied his new theory of disruption, and how industrial giants filled with smart people would make seemingly smart decisions that would lead to their downfall.

One aspect of his theory, which later went into his book, Competing Against Luck, is the Theory of Jobs to Be Done. Quite simply, Clay believes that companies can go astray by focusing too much on the data about their customers and the features of their product. Instead, he argues they should focus on the end-to-end experience of the job that their product is being hired to do.

In the past few years, I’ve come to believe that saving is a job that a huge number of people want a product to help them do and help them do it well.

Saving Itself is a Goal

Our lives are filled with a large number of small financial decisions and problems, but there are only a few very large financial moments that warrant the creation of an entire companies to support. Spending, borrowing, investing and financial advice all certainly fit that description. I believe that saving belongs on that list as well.

Americans are in a terrible state when it comes to saving. 6 in 10 Americans don’t have $500 in savings. An estimated 66% of households have zero dollars saved. If you are cynical about small, one-off surveys, The Federal Reserve itself estimated in 2015 that 47% of households didn’t have the means to cover a $400 emergency expense.

Saving is a huge problem, so it isn’t really surprising that tens of millions of Americans seem to be looking for something to help them save. Enter Acorns.

Hiring Acorns

Over the past two years, it has been astounding to watch Acorns grow. An elegantly simple product, designed from the ground up for a mobile generation, Acorns has grown to over 2 million accounts in less than three years. In the first half of 2017 alone, Acorns added over 600,000 new customers. Their overall mission is to look after the financial best interest of the up-and-coming, something I personally care deeply about.

It isn’t really surprising to see why so many Americans have decided to use Acorns to help them save. 75% of Americans have a household income under $100K. Acorns simple features like Round Ups automate the process of making sure that as you spend, you save. Acorns has now performed over 637 million round-up transactions for their customers – each one an action designed to help people save more. I believe that on any given day, thousands of people decide to hire a product to help them save, and increasingly they are hiring Acorns.

When I met the founders of Acorns two years ago, we immediately connected over the common ground between their culture and Wealthfront’s (the company I was running at the time.) They are very different services, focused on different problems and audiences, but with a shared belief in the power of automation. This is a company worth supporting, and I feel fortunate to serve on their Board of Directors.

At a time when people continue to grow more and more frustrated with the solutions offered by incumbent banks and brokerages, I continue to be excited about the opportunities for new products that are built around automation and world-class software design.  As an industry, we can and should radically improve the financial solutions that are available to everyone. Acorns is proving that saving is a job worth doing.

Silicon Valley Home Prices, Stock Prices & Bitcoin

I’m writing this post with a bit of trepidation, because talking about Silicon Valley home prices these days is a bit dicey. The surge of the last five years has been shocking, and almost no one I know feels good about how difficult it is for people to buy a new home in Silicon Valley in 2017.

So if you need a trigger warning, this is it. Stop reading now.

The truth is, as shocking as the rise in Silicon Valley home prices has been, there has also been an asset boom in other dimensions as well. Total compensation for engineers is up considerably and stock prices at the big tech companies continue to rise.

To visualize this, I thought I’d put together a few charts based on real market data. As a proxy for Silicon Valley, I pulled the last 5 years of home prices from Zillow, and monthly stock price data from Yahoo.

Palo Alto Home Prices

Two days ago, the Mercury News reported that a home in Palo Alto sold for $30 million.  A quick check on Zillow seems to confirm this.

I chose Palo Alto as a proxy for Silicon Valley home prices because it is historically “ground zero” for Silicon Valley tech companies, and it has relatively close proximity to all of the massive tech giants (Apple, Google, Facebook).

I picked June 2012 – June 2017, not only because it is roughly five years, but also it also happens to mirror the time that Facebook has spent as a public company. For many in the local real estate market, correctly or incorrectly, the Facebook IPO still looms as a transformational event.

As you can see, in June 2012 the average Palo Alto home cost $1.38 million. Five years later, the estimate for June 2017 is up 84.6% to $2.55 million.

Apple (AAPL)

Apple is the most valuable company in the world, as measured either by market capitalization ($810B as of 6/7/2017) or by profitability ($45.7B in 2016).  Thanks in part to this exception financial performance, Apple stock (AAPL) has risen 84.5% in the last five years, from $83.43 per share to $153.93 per share.

84.5%? Where have I heard that number before?

That’s right, the increase in Apple stock over the last five years is almost exactly the same increase as the average home price in Palo Alto over the same time period.

In June 2012, it took 16,555 shares of Apple stock to purchase the average Palo Alto home. In June 2017, it took 16,566 shares. (Of course, with dividends, you’re actually doing a little better if you are a shareholder.)

If you look at the chart, the pink line shows clearly the large rise in price for the average Palo Alto home. The blue line is the number of AAPL shares it would take to by the average Palo Alto home in that month. As you can see, AAPL stock is volatile, but five years later, that ratio has ended up in almost the exact same place.

Alphabet / Google (GOOG)

Alphabet, the company formerly known as Google, may not be as large as Apple in market capitalization ($686B), but it has seen far more share appreciation in the past five years. Since June 2012, Alphabet has seen its stock price rise 240.4%, from $288.95 in June 2012 to $983.66 per share.

What does this mean? Well, it means that if you have been fortunate enough to hold Google equity, the rise in Palo Alto home prices doesn’t look as ominous. It took 4,780 shares of Google to purchase the average Palo Alto home in June 2012, but it only took 2,592 to purchase the average Palo Alto home in June 2017.

Facebook (FB)

Facebook, the youngest of the massive tech giants, already has one of the largest market capitalizations in the world. As of today, Facebook is valued at $443B. Facebook stock has risen 394% in the past five years, from $31.10 in June 2012 to $153.63 in June 2017.

To state the obvious, it has been a good five years for owners of Facebook stock. Not many assets could make owning Palo Alto real estate look slow, but 394% growth in five years is unbelievable. In June 2012, you would have needed 44,412 shares to buy the average Palo Alto home. In June 2017, that number had dropped significantly to just 16,598 shares.

Bitcoin (BTC)

While I realize that Bitcoin is not a stock, the original idea for this post came from a joke I made on Twitter recently given all of the buzz about Bitcoin, Ethereum and ICOs over the past few weeks.

I couldn’t resist running the numbers.

For the small number of readers of this blog that haven’t been following the price of Bitcoin, the increase in value over the past five years has been unbelievable.The total value of all Bitcoin outstanding is currently about $44.5B. Since June 2012, Bitcoin has risen approximately 4,257%, from $6.70 per Bitcoin to a current value of $2,858.90.

You can see why there has been so much buzz.

In June of 2012, it would have taken 260,149 Bitcoin to buy the average home in Palo Alto. In June of 2017, that number is now down to 892.

Needless to say, anyone who sold Bitcoin to buy a house in 2012 is likely not loving these numbers. But to people who have held Bitcoin for the past five years, Palo Alto is looking cheaper by the day.

Silicon Valley Is Seeing Significant Asset Inflation

To be clear, I’m not attempting to attribute causality to these charts. I believe the real driver of home prices in Silicon Valley is the lack of sufficient building of new supply at pace with the economy, combined with a significant increase in compensation for technology employees and historically low interest rates.

But the fact is, if you are fortunate enough to have equity in one of the tech giants (or in Bitcoin), houses might actually be looking cheaper now relatively than they did five years ago.

I always find it enlightening to look at real data and compare it to intuition. Hope you find this data and these charts as interesting as I did.

Back at Greylock

Today, Reid Hoffman shared the news that I’ve rejoined Greylock Partners at an Executive in Residence. I couldn’t be more excited to be back.

This is an unusual step for me, as it is the first time in my twenty-year career that I’ve decided to come back to a firm. Then again, Greylock is an unusual firm.

When I look around Greylock, I recall Warren Buffett’s famous advice* on what to look for in people:

“Somebody once said that in looking for people to hire, you look for three qualities: integrity, intelligence, and energy. And if you don’t have the first, the other two will kill you.”

Early in my time at LinkedIn, I remember Reid Hoffman modifying this a bit:

“You really want to look for people who are smart, trustworthy, and ambitious. Having just two of the three is a problem.”

Deciding who you want work with is one of the most important decisions that you make in your career. Not to embarrass John Lilly, but I think his advice from the New York Times is spot-on:

“…find your tribe. You should look around and figure out whose team you’re on and whose team you’re not on. And for the people whose team you want to be on, you need to invest in those relationships and treat them well and spend time with them.”

When I look around Greylock, I see nothing but people who are smart, ambitious, and trustworthy; people whose team I want to be on. I can’t think of a better environment for me as I explore and look for the next big thing.**

(*) If you are interested in the history of this flavor of advice, this post on Quote Investigator is fascinating. Goes back more than a hundred years to a German General named Kurt von Hammerstein-Equord.

(**) Might be time to re-read my Executive in Residence series from 2012…

Spend Time Thinking About The People Who Don’t Use Your Product


This is an extension to my original three post series on user acquisition.

Today, AirBnB announced that it had reached a settlement with the city of San Francisco on how to effectively register and monitor legal listings in the city. I am a huge fan of the company, and it seems like a positive outcome for both San Francisco and AirBnB.

For many, the issues around many of the sharing economy companies, including AirBnB, are examples of regulators trying to find a way to both control and incorporate rapid, disruptive innovation.  There is, of course, some truth to this point of view.

However, as a product leader, there is another important takeaway that seems to be too often forgotten. Most of us spend too little time thinking carefully about the people who don’t use our products. 

The people who don’t use your product often won’t show up in your core metrics. But if you don’t spend time understanding them, you will eventually feel the negative effects in your growth and your brand.

It’s Natural for Companies to Obsess About Their Users

When a startup launches a new product, it is natural to obsess with every user it touches. Every click, every tap, every piece of data is precious feedback about your features. The data is one of the most objective sources of information about what your users are doing with your product and when they are doing it. In the early days, before finding product/market fit, a huge amount of time tends to be spent on the people you touch but who don’t convert. In fact, that may be where most people at the company spend their time.

As consumer products find product/market fit and hit escape velocity, more and more engineers and designers spend a disproportionate amount of time on users. The people who work on growth & marketing will still often continue to look at the data on leads, trying to find ways of converting those non-users to users. However, as a percentage of the company, fewer and fewer engineers, designers & product managers will be looking at data from non-users.

This makes sense, of course, because as your product grows, almost all feature development is focused on your users. In 2008, when we established the Growth team at LinkedIn, we discovered that of the hundreds of features on, only three features reliably touched non-users. (For those of you who are curious, those features were the guest invitation (email), the public homepage (, and the public profile (in search.))

Customer obsession, of course, is generally a good thing. But as we learned at LinkedIn, if you want to grow a viral product, you have to spend a considerable amount of time thinking about the non-user, where they touch your brand and your service, and find ways to both reach them and convert them to users.

You Have More Non-Users Than Users

Few brands and products could ever claim that their conversion rate for everyone they touch is over 50%. It is even possible that Facebook, with nearly 2 billion users, still has more people in the world who have heard of the company than who use it.

In 2011, I remember talking to the great founders at CardMunch about a new email they were proposing to add to their service. CardMunch was a wonderful app that made it effortless to scan a business card and then have it automatically entered into your address book, with almost no errors. The proposal was to add an email so that the person whose business card you scanned (non-user) received an email from the CardMunch user with their business card in electronic form.

The team was ready to whip something together quickly and test the idea, and the concept was good in principle. But given some of the experience of Plaxo a decade before, it was prudent to ask the simple question: “How many people will see this new email?” Within a few minutes, we figured out that the number of people who would receive this email within the first three months would be 30 to 50 times the total user base of the application.

Some of you are probably thinking, “sounds like a great growth feature!” Others are likely venting about why we have so many emails cluttering our inboxes. Both reactions are fair.

The guidance I gave the team, however, was to consider the fact that, once they launch this feature, most people who have ever heard of CardMunch will have only heard of it through this email. The product and the brand. I asked them to spend a bit more time on the design on the email, in that context, to ensure that all of their hard work on a wonderful product wouldn’t be drowned in an avalanche of poor experience.

In the end, Sid Viswanathan & team did a great job brainstorming ways that they could show the value of a connected addressbook in the email, including LinkedIn features like people you know in common. Once framed properly, it was simple to think about what they wanted non-users to think about their brand and their product.

Non-Users Matter

Marketers, of course, have known this for decades. It is a brand marketing staple that it takes at least three touches of a brand before it will stick with a potential customer.

Somewhere along the way, software companies lost touch with the basic idea that every piece of content that contains their brand is a potential touch. It is not just the users of the core product that matter for long term growth.

Market research and customer development are often essential for discovering and understanding new potential users for your product. The case can be made that viral systems can, in fact, spread to these new pockets automatically. However, truly viral products are few and far between, and in most cases these new markets will not be in the data sets that your product & engineering teams are focused on.

Brand will also impact your company well beyond new user acquisition. With AirBnB, we now know the many ways in which their service and brand touch non-users. Neighbors, for example, have natural questions and concerns when a house or a unit near by is available on the platform.

Software companies, especially successful ones, tend to have passionate and talented designers and product leaders who are eager to find clever solutions to real user problems. Given the right data and focus, there is no question that these teams can also design and build features that address non-user concerns.

Tesla spends time thinking both about the feeling a driver has in the car, as well as the experience of a non-Tesla owner who is watching that car drive by.

Spend more time thinking about all of the people who touch your product & your brand, not just your users.


Product Leaders as Curators & Editors

Gallery Show

A few years ago, I wrote a few posts to outline the requirements for exceptional product leadership:

While I have been gratified that people continue to find utility and value in these posts, I’ve come to believe that product leadership, particularly the issue of prioritization and phasing of a product roadmap, remains daunting and challenging for most teams.

In particular, the need for organizational scalability and speed of innovation has driven the widespread popularity of small, independent teams building product and features. Unfortunately, the side effect of the explosion of small teams has also amplified user-experience fragmentation and the haphazard quality of many web-based and mobile software applications.

As a result, I’ve come to believe that there are two facets of  product leadership that have become increasingly important for delivering a high quality product experience: curation & editorial.

Curation Amplifies Your Product Experience

Around 2014, I remember first being struck by a product management job description at Pinterest which incorporated the concept of curation as a core responsibility of product management.

The dilemma of product prioritization is always simple to understand: most software teams, filled with talented people, have more ideas for great features that the capability to execute. As a result, there has to be some process for filtering down the ideas to answer the question of “what do we build next?”

Prioritization on metrics, customer requests and delight is not hard to operationalize, but it still leaves open critical questions:

  • How does the product & experience come together for the user after we ship?
  • How does the product communicate the changes to the customer in way they can easily understand and utilize?

I believe curation is the key to answering these questions.

Curation is an under-appreciated skill in software design. In the world of art, curation is a critical and valued function. A curator ensures that the pieces of art not only combine to amplify each other collectively, but also gives thought to the experience a viewer will have when engaging with the collection.

Users need some level of coherence in new versions of your product. With proper curation, features and changes amplify each other, and lead to a greater customer appreciation of your efforts through a product experience that is more coherent and easier to communicate.

Without curation, software feature prioritization tends to devolve purely into the line-item value of a given feature, rather than how it fits in general with the whole product, or the product release. Great curators won’t think twice about cutting a piece that doesn’t fit the theme of the show, even if it is exceptional.

Designers, not surprisingly, tend to intrinsically understand the value of curation, and valiantly attempt to connect features together into a coherent product experience. Unfortunately, they often are forced to incorporate together a hodge-podge of features that have been prioritized independently by different small teams.

This is not an argument against constant enhancement and iteration of code, or the constant shipping of bug fixes and small feature enhancements. But for user-facing features, teams need to be wiling to hear from product leadership that a great idea for a new feature is not enough to qualify it for immediate prioritization. Customers cannot endlessly absorb a haphazard array of changes and feature enhancements. The perceived quality of the product drops, and customers fail to perceive the value in the features that are shipped.

Every Creator Needs an Editor

Understanding the value of editorial comes easily to professionals who have worked in content & design.

In my experience, many otherwise talented engineers and product managers balk at receiving critical review of their work. Sure, most software engineers understand the value of pair programming and code reviews. But for some reason, when it comes to overall feature design, the sentiment almost always shifts to stubborn independence.

Unfortunately, just like in writing, having a great editor is essential for the overall quality  and consistency of the finished work.

Even the best writers benefit from having a great editor. J.K. Rowling may have written all seven Harry Potter books herself, but she had a team of editors ensuring everything from line level quality to the plot consistency of the overall series.

Why editors? In general, editors provide three levels of assistance to writers: proofreading (spelling, punctuation, grammar), copy-editing (phrasing, style), and developmental-editing (plot, character development, pacing, tone, and effectiveness.)

Most writers at first balk at the idea of an editor. They are professionals, after all, and incredibly skilled. Why do they need someone in between them and their readers?

The answer is two-fold: first, editors provide a more objective “second-pair of eyes” not affected by the sunk cost and confirmation bias inherent in any creative process, and second they are typically individuals who are exceptionally talented at finding errors and issues that will be perceived by the target audience.

The same applies to software products.

Even exceptionally talented engineers & designers become blind to their own work. While each function can have their own version of an editorial process, my experience has been that if product leadership doesn’t actively engage in the editorial process, the quality and the coherence of the product suffers.

Product Leaders as Curators & Editors

Most software companies have moved to a bottoms-up, distributed organization process for their engineering, design & product teams. Amazon, of course, is famous for their two-pizza team concept. As a result, the need for curation and editorial to keep the product experience coherent has become critical.

If you look critically at organizations that have a distributed culture, but still ship high quality product experiences, you’ll find that there is an accepted culture of curation & editorial in their product process, connecting all the way to the CEO.

If you are a product leader, think carefully about how you can incorporate curation & editorial into your process as you scale.

Forget the Turing Test. The Key to Conversational Engagement Might Be Trampoline Moments


In 2016, voice-based interfaces exploded into the imagination of the startup community as a potential new consumer platform. Amazon deserves much of the credit for this radical shift, as the Amazon Echo seemed to jump the chasm from early adopters to a more mainstream market. Of course, voice has been a hot topic now for years, as Apple & Google both leveraged their ubiquitous mobile platforms to launch Siri & Google Now, and Microsoft & Amazon have demonstrated incredible technical progress with Cortana & Alexa.

Unfortunately, as the excitement around voice shifts into practical execution, there is an uncomfortable consensus growing that there is something amiss with these new conversational platforms. The issue? The engagement numbers just aren’t as strong as expected, or even as strong as engagement numbers for traditional web or app-based interactions. One of the biggest issues? Retention.

I believe the issue is real, and will be a persistent problem for developers and designers looking to create the next generation of conversational interfaces. But if I had to give one piece of advice to those creative professionals, it would be this:

Deliver trampoline moments.

Lessons from PullString

Over the past four years, I’ve had the incredible opportunity to be an investor and board member at PullString, headed by Oren Jacob, the former CTO of Pixar. This company set out with the audacious goal of reimagining conversational interfaces designed for entertainment, rather than for utility. With a bit of that unique Pixar magic, this incredible team believed in two things that even to this day seem quite at odds with the conventional wisdom of Silicon Valley:

  1. Conversation is a fundamentally new medium for creative content, and would expand beyond the pure utility of a search engine interface to a platform for engagement & entertainment.
  2. A platform to deliver truly engaging entertainment through conversation would require the combination of both technical and creative contributors to the content creation process.

Over the past few years, Pullstring has delivered a wide range of industry-firsts for voice-based engagement for a wide variety of audiences, ranging from young children to adults. Large brands, like Activision’s Call of Duty, Disney’s Marvel and Mattel’s Barbie trust Pullstring’s platform because of its unparalleled scalability and its unique ability to integrate content from creative professionals with expertise in sound, voice, character and dialogue. Even Amazon counts on Pullstring when they want to deliver high quality conversational content.

However, one of the key insights about conversational engagement came early on, during one of their rigorous rounds of user testing & prototyping. After session after session with children, who would use, but not deeply engage with a conversational application, they found it. A trampoline moment.  

Child: Hey
Pullstring: Quick! Name three things you like that are outside.
Child: I think please I’m Chris taxes and jumping on trampolines
Pullstring: W-w-w-w-w-w-wait…you mean like, a real trampoline?
Child: Yeah
Pullstring: Do you think I could go on it sometime? I’ve been using your bed up until now and I think the springs are worn out…
Child: Are you really able to
Pullstring: My oh my, what a day I’ve had…It was so strenuous I can barely remember what I did…Ellington? What have we got in the log?
Pullstring: Right. We sat on the bed. Ellington needed a little rest time from our usual forays.

A couple things you’ll note here:

  1. Speech recognition for children’s speech was very imprecise at the time. The text is not actually what the child said, but the text fed back from the best speech recognition engine of that time.
  2. The child’s willingness to “believe” in Winston (the virtual character, with his friend Ellington) changes dramatically when he demonstrates active listening around one of her favorite things, the trampoline.

This session went on not just for a minute, not just ten minutes, but over 30 minutes. The child had clearly decided to engage, and continued to engage, despite a huge number of imperfections in the interaction.

Why? The trampoline moment.

Turing Test or Trampoline Moment?

For decades, the high bar in artificial intelligence has been the Turing Test, invented by Alan Turing in 1950. The test was fairly simple: an evaluator (human) would have a conversation with two entities, one human and one artificial. If the evaluator could not reliably tell the human from the computer, the machine would “pass” the test.

While there are a number of criticisms of the Turing Test, there is no question that it has profoundly affected the way many evaluate machine-generated conversation.

The insight from the trampoline moment was different, and takes more of its heritage from the world of fiction. The question can be reframed not whether or not the consumer believes the character is human, but instead are they willing to suspend their disbelief long enough to immerse themselves in the experience.

Most people don’t believe that Iron Man is real, or that they are witnessing an accurate portrayal of Alexander Hamilton. They know that the actors in their favorite romantic comedy aren’t really in love, and they forgive plot holes and shallow character development. Even highly critical audiences of science fiction often can and will forgive obvious scientific flaws in the technology presented. (Well, not all of them)

The magic is really in the suspension of disbeliefthe willingness to suspend your own critical faculties and believe the unbelievable; the willingness to sacrifice logic for the sake of enjoyment.

Is it really surprising that a critical insight to human engagement might stem from the arts, where creative geniuses have spent thousands of years attempting to engage and entertain notoriously fickle humans?

Focus on Trampoline Moments, not Intelligence

The progress in artificial intelligence, voice recognition and conversational interfaces has been astounding in the past few years. There is no question that these technologies will reshape almost every facet of our economy and daily lives in the coming decade.

That being said, in Silicon Valley, it is sometimes too easy to focus on the hardest technical problem, rather than the one that will bring the consumer the most delight.

The reason Pullstring spends time talking about finding “trampoline moments” is likely the same reason talented product leaders talk about finding “magic moments” in their product experience. If you can connect with your customer emotionally, you will inevitably find that engagement and retention increase.

Trigger their suspension of disbelief. Find your trampoline moments.