Rockets, Data, and DMPs

cartoon-rocket-6I am amazed at the recent ascent of RocketFuel, which has grown faster than any ad network I have ever seen. Budgets in the display media business are guarded by large agency holding companies, and agencies typically discourage clients from scaling this fast with a specific vendor. It looks like Rocketfuel has found a way around the industry stalwarts.

The secret is that Rocketfuel has done a good job defining success on each campaign. This involves a direct conversation with marketers to lay out goals, and afterwards performance can be measured directly against any other vendor, including an agency trading desk. To the extent a client wants low-cost clicks, Rocketfuel delivers, and to the extent a client wants email sign-ups, Rocketfuel can deliver those too. There are murmurs about the methodology underpinning all of Rocketfuel’s success, but at this stage they are arguably the best performance vendor in the market.

The one knock on Rocketfuel, which is actually a knock on every adtech company, is that they are not achieving the big vision of advertisers using their platform directly to buy media. Like every other ad network, Rocketfuel is currently bought campaign-by-campaign using insertion orders. The public market expects them to transition to a software model that is used in-house by clients. Unfortunately I think this vision is still 5-10 years off.

One hard part about buying media is the optimization that occurs mid-campaign. Typically the first dollars on every performance campaign are used to locate the audience, and then incremental dollars perform better once the system hones in. There are algorithms that automate parts of this puzzle, but it is largely customized for each client. Think about this problem from the perspective of a financial trader – if every stock broker used the same platform with the same algorithms, a smart hedge fund would swoop in and take advantage of that information to make a profit.  The same thing would happen in the media business if advertisers tried to take these systems in house today without investing in teams of data scientists.

A more realistic near-term vision is that marketers will bring data science in house over the next five years.  And only after that occurs will they venture into buying media. The market for these new “data science systems” is still emerging – the category includes a hodge-podge of acronyms like DMP (Data Management Platform), MDS (Marketing Data System), and TMS (Tag Management System) – but eventually a standard vocabulary will emerge to encompass it all.

Greycroft is keeping close tabs on the race to build the ultimate client-side data platform.  We have a few bets here already, such as Performance HorizonResonate Insights, and Tagman, and will likely make more over the next year or two.  We believe that this market is big enough to support a number of billion dollar outcomes.  Most importantly, the winner of the data war might also win the bigger fight to control media spend, and if played right someone could run the table here.  Based on what I am seeing in the market, my bet is that the ultimate winner is not going to be Rocketfuel.

A New State Of The Union

Welcome to my first political-themed blog post (with a mix of social science):

Every election cycle, research firms contact voters and ask questions like “How important is reforming Medicare and Social Security?” or “How important is reducing greenhouse gas emissions by 80% over the next 20 years?” I get these forms in the mail too, and on a couple occasions I even filled them out.

I always assumed that my party’s agenda would be the issues that registered voters feel strongest about. But I was wrong.

A few months ago I was introduced to a group called No Labels (you can check them out here) as well as the core research team at Pew. They shared the inside scoop about how political agendas are crafted in the US. Even though I don’t typically discuss politics I thought the math and psychology behind election results was interesting.

As it turns out, politicians use surveys to create “distance” between candidates. For instance, rather than focus on the issues that registered Democrats care most about, candidates will use a formula like: D – R + I. This takes the weight of the registered Democrat base, subtracts the weight of the registered Republican base, and then adds back Independents. Republicans do the same thing but superimpose D and R. As you can imagine, the resulting agenda is very polarizing.

The crazy thing is that this strategy wins elections. Candidates get support from their base and independents, and they distinguish themselves from their opponents on key issues. Julius Caesar called it divide and conquer.

The obvious downside of this approach is that nobody gets what they really want. The not-so-obvious downside, which many people are finding out right now, is that the winner is left with an ungovernable mess. It is virtually impossible to move the country forward when our President’s strategy is predicated on just how much he can make half of the population dislike him.

I included a snapshot below of the top 5 issues for each party when they employ the “distance” strategy. This is from my friends at Pew and No Labels. You will see how eerily familiar these themes are for current candidates in office.

Republican agenda (R-D+I):

1.) Balance the federal budget by 2024.
2.) Democrats and Republicans strike a grand bargain to reduce the deficit by reforming the tax code and entitlement spending by 2016.
3.) Solve Social Security and Medicare financial problems by 2020.
4.) Reform Medicare and Social Security so it is secure another 75 years.
5.) Cut taxes across the board by 2020 to generate more opportunity and jobs.

Here is the Democrat agenda (D-R+I):

1.) Raise the minimum wage to 12.50 and provide a minimum wage that provides a living wage.
2.) Cut income inequality in America by enacting tax reform that raises taxes on the wealthy and lowers them on the poor.
3.) Make the latest cures and medicines available to every American who needs them by 2020.
4.) Cut the number of Americans living in poverty in half over the next 10 years.
5.) Ensure that every child has access to high quality early education by 2018.

The team at No Labels is trying a different approach. They used the same body of research, but instead of creating “distance” they tried to find the issues that most voters agree upon. These are the results when you take D + R + 2 x Independent:

1.) Solve Social Security and Medicare financial problems by 2020.
2.) Reform Medicare and Social Security so it is secure another 75 years.

3.) Balance the federal budget by 2024.
4.) Create 50 million new jobs over the next 10 years.
5.) By 2018 bring back the 2 million manufacturing jobs lost during the recession.

If you are looking for a true national agenda, this is it.


Just A Little Bit of History Repeating


As we approach the end of 2013 I am feeling a little anxious.  If you are a gambler you might know the feeling, it is the same sense you get during a good run.  The dice are hot and the numbers keep coming.  Like a good sport you keep pressing.  But at some point you look up and see that you have an uncomfortable amount of money on the table.

At holiday parties my VC friends keep asking the same question – is it 1999 or 1997?‎  How much time do we have left on this bull market?  If you are a young partner at a venture capital firm this is not an academic question because almost all of your net worth is tied up in start-up stocks.

The good news is that I think we have at least another year.  There is even an outside chance that the market could get hotter.  The big unknown is how the market will react once the Fed stops stimulus spending next year. ‎

In hindsight, this all started with the financial crisis.  The Fed lowered interest rates to zero, and like many investors I moved money out of my zero-return bank account and into investment-grade bond funds.  The yields on those funds quickly went to zero because everyone else was doing the exact same thing, so I moved money into riskier bond funds.  The yield on those funds went down too.  After that I put money into equities.

As they say, “You can’t fight the Fed.”

The S&P 500 was up 23% in 2013.  Many people predicted that would happen. But who would have guessed that Best Buy would be the third best performing stock this year, up 230%?  The market is willing to overlook that Best Buy is shrinking, losing money, and every comparable company (Circuit City, CompUSA, Tweeter, etc) went bankrupt.  Similarly, many of the hot IPOs this year never made a profit before going public.  So much for the axiom about “four consecutive quarters of profitability ahead of an IPO”.  That just didn’t matter in 2013.

This is the hazard of Fed stimulus.  Returns flatten out on safe investments, encouraging investors to take more risk.

If you are an institutional investor sitting on the sidelines, 2013 must have been a painful year.  Even if you are right in the long run about the value of equities, you under-performed the market by a lot in 2013.

The reason I am bullish about 2014 is that the Fed inadvertently created a pressure cooker.  Money managers can’t afford to have two bad years in a row – that means they lose their job – so they will take more risk in 2014.  Maybe they invest in some IPOs. ‎ Maybe they dip into the private markets, putting money into “Pre-IPO” growth companies.

This is where it starts to get interesting for VCs like me.

The amount of money on the sidelines today dwarfs the markets from the 1990s.  McKinsey wrote a report in August of 2011 that showed the global financial markets grew by $100 Trillion since 2000, to a total of $212 Trillion, and almost all the growth was in the public and private debt markets.  All it takes is a few investors to make small adjustments, maybe take a little more risk in 2014, and the good times come again.

‎It is just ‎a little bit of history repeating.


Hello Walled Garden

Picture1When Twitter upgraded their mobile application this year, to a version called “New, New Twitter”, there was one significant change.  When I clicked on a link, rather than open my Chrome browser to load the web page, Twitter loaded a built-in browser.  This made a seamless transition between Twitter and the web, without ever leaving the app.

In the past few weeks Facebook updated their mobile app as well.  And guess what?  It now does the exact same thing as Twitter.  That means Facebook has a browser too.

So what is all the fuss about web browsers?

Back in 2010, Time Magazine wrote that browser technology was an altruistic pursuit.  If a user stayed on all Microsoft products – Windows OS, Internet Explorer, and Bing – the web would somehow work better for the consumer.

That is just nonsense.

The truth is that the browser captures data, which has become an important asset in the digital age.  Information about what websites people visit, commonly called “behavioral data”, provides the core insight to make online advertising work.  Owning a consumer’s web browser is the best way to see all of this behavior. (And, in case you were wondering, even the privacy advocates at Firefox get all their money from Google.)

Now that Facebook and Twitter come with mobile web browsers, their users are completely invisible to Google.  There is no Google Chrome activity and, most importantly, no search revenue.  It is a black hole.  And you can multiply that across a billion users.

This should get your attention.  “Facebook Browser” market share is not being measured by any official sources, but the company has so many page views that their mobile browser is likely #3 in the world right now.  The only mobile browsers with a larger footprint are Android and Safari because they are bundled with  your phone.

In addition to the data, another benefit of owning the browser is that you can determine what happens when a consumer mis-types a URL.  In the case of Google, if you type Nike into the browser (without the .com) they take you to a page filled with pay-per-click Nike ads.  If Facebook had a working search engine they could profit from this as well.

As a start-up investor, I think Facebook and Twitter’s new browser is good news.  Google has become dominant in too many categories for my comfort.  I don’t even know all the ways that Google is making money, but whenever I browse the web I know the meter is running.

When I look across the tech landscape there are very few companies that can mount an attack on Google. In this new iteration, Facebook and Twitter are using their mobile products to keep traffic for themselves. This is reminiscent of the walled garden strategy used by Prodigy, Compuserve, and AOL in the early 1990s.  And who knows, maybe this time around it will even work…

The VC And The Art World

In 2010, while the financial crisis was still unwinding, I traveled down to Tampa with my wife to visit her family.  On a back road not far from our hotel was a recently built housing complex.  These were beautiful, low-rise brick buildings, with landscaped lawns and tile roofs that reminded me of Florence.

The sign out front read:

2,000 square feet, 3 bedroom, 3 bathroom condos.

Starting at $349,999.



$120,000 or best offer.

I turned to my wife and said that we are moving to Tampa.

For the price of a one bedroom in New York City we could have 27 bedrooms, 27 bathrooms, and 9 kitchens. Our own Versailles!!

I often think of this Tampa measuring stick.  It is a reminder that there is a world outside of NY and SF where buyers are scarce.  Let’s call this the “Real World”.  In the real world there aren’t 47 people showing up at every open house and making all cash bids.

Someday I want to host a TV show called “The Price is NOT Right” where Midwesterners like me guess how much menu items cost at NY restaurants.  My favorite is the $7.50 toast (and yes, it is just toast) on the breakfast menu at E.A.T.

But that is small potatoes.

This past week in New York a Francis Bacon painting sold at auction for $142MM.  It is an oil painting on canvas, painted 44 years ago.  My wife has a Masters in Art History and, after a long explanation of the importance of the artwork, she told me that the price is crazy.

Paintings are not income-producing assets, so the only reason a painting is worth $142MM is that someone is willing to pay $142MM for it.  The art market is tricky.  Sometimes buyers lose interest in an artist for no apparent reason and values take a nosedive.

Earlier this week Snapchat turned down an offer from Facebook for $3BN.  Like the Francis Bacon painting, Snapchat is not an income producing asset (at least not yet).  I consider myself an expert in valuing start-up companies and I am struggling to figure this one out.

I have recently heard a new rationale for valuation.  It goes like this:  “Google will pay billions for this company.  Just look at Waze.”  You can also substitute Facebook for Google and Instagram for Waze.  Whenever I hear this it reminds me of the Francis Bacon painting.

This leads me to one final thought:  If a company doesn’t make money, is it art?

The Unicorn Effect

[Title borrowed from Aileen Lee’s essay “Welcome to the Unicorn Club.” It is worth a read.]


[Editors Note:  I don’t know the guy in this photo.]

This past February we flew the entire Greycroft team to Utah for a ski weekend, which doubled as our firm-wide offsite.  In addition to the usual stuff, every partner had a chance to lead a discussion on a topic of his or her choice.  I picked a review of “Thinking, Fast and Slow” by Daniel Kahneman.

My goal was to help the firm identify decision-making bias.  My theory is that investors often focus on recent outcomes, which anchor us to an unrealistic notion of exit value.  I have circumstantial evidence that other VCs do this, but I mostly find myself falling into this trap all the time.

This subject is top of mind because in the past few weeks we have passed on, as well as outright lost, a number of competitive investments based on valuation.  It is not the first time we have gone through a spell like this, and it won’t be the last.

In my experience VC valuations generally fall within a close range – firms have the same information and we underwrite investments with similar return targets – until something changes in the market and valuations suddenly rise or fall.  We are in a phase right now where other firms have bid 200-300% higher than our proposals.  The only conclusion is that other investors have expectations that are very different from our expectations.

I am sure we are not alone.

In the past 24 hours both Fred Wilson and Aileen Lee have written blogs about the number of billion dollar outcomes, so big deals are top of mind.  If you read their posts you may even think that there are a lot of billion dollar outcomes out there.  This is the unicorn effect:  just because someone once saw a unicorn doesn’t mean that you will see one too.  If Aileen’s figures are correct, an early stage investor who makes 2 to 3 investments a year could invest for 400 years and still have less than a 50% chance of funding a single billion dollar outcome.

I also looked at the most recent data from my friends at CB Insights and PWC, and, in spite of the strong IPO market, the mean and median outcomes for venture-backed companies have not changed much year over year.  Based on their data, the median venture-backed outcome is still hovering around $55MM.  This number excludes both write-offs and companies that were acquired without a disclosed purchase price.  If you were to include those investments you would have a median outcome of zero, because fewer than 50% of VC-backed companies reach a successful exit.  On a similar “upside-only” analysis, the mean outcome is around $120MM.

Whenever we lose an investment on valuation I can’t help but think to myself that maybe the other guy is right.  Maybe we have missed the next Facebook or Google.  On the other hand, maybe a better strategy is to sit out a hand or two and see what the future brings. Time will tell.

Winner Take All


I recently realized that I have given a lot of entrepreneurs bad advice when it comes to software sales.  Unfortunately I have been doing it for many years as an investor.

I used to think that the goal of a Series A software investor was to build a great team, build a great product, and prove that customers were willing to pay for your product.  I thought that if I could de-risk the sales cycle and get some proof points, that would be adequate to get other investors to follow on later with capital for scaling.

Almost all my companies are in greenfield market opportunities, where there is no incumbent and the sales cycle is completely unknown, so I put a lot of emphasis on figuring out how to sell early.  I have nothing against brownfield markets, but I find it more interesting to build something for the first time.

Since my companies are evangelizing new areas, I used to tell CEOs that they should price high enough to recoup sales and marketing costs.  With a field sales team that is inevitably a six figure price point.  Customer payments also defer operating costs in the early days, which is helpful because we are a smaller fund.

The mistake I made, which is painfully clear to me now, is that many SAAS markets have a winner-take-all dynamic.  This is especially true if your products are sticky, or if they have network effects.  There is a new trend called “consumerization of the enterprise” that takes this a step further, with software companies that don’t use sales teams at all.  Instead they market like consumer companies and grow virally.

When a company is in a winner-take-all market there will typically be one major outcome, worth billions in enterprise value at exit.  Other companies in the sector either survive as a tuck-in acquisition or go bankrupt.  This is radically different from “non-winner-take-all” markets where there could be a half dozen reasonable exits.  If you are investing in a winner-take-all market the only thing that matters is making sure your money is in the winner.

This “winner take all” strategy thrives in Silicon Valley, where there are a lot of smart programmers who can replicate and improve upon any product.  Moreover, these programmers are not motivated by revenues – they only want the world to use their tools.

This trend arrived at the right time, because we are currently in a market with infinite growth capital.  If you look at the IPO class of 2013, which is heavily weighted towards west coast software companies, the average paid-in capital prior to IPO was $103MM. That is not a typo.  To reinforce the point, almost all of these companies are still losing money, and some are losing more money as they get larger.  All that matters today is growth.

My companies, which were focused on revenues, would go to market and build up a base of $5 to $10 million per year in recurring revenue.  Once that success became known competitors swarmed in, offering a similar product for free.  Some customers would churn and others would renew at 25-50% of the initial contract value.  It is impossible to maintain high pricing against free alternatives.  The net result is that for a period of 1-2 years my companies would struggle to stay afloat as the existing book of business became an anchor to growth.  This is a death spiral for venture-backed companies because nobody funds flat growth.

I have now seen this happen twice, and I have vowed to never let it happen again.

The right strategy in winner-take-all markets is to get as many customers as you can, as fast as you can.  Instead of proving early on that customers will pay, prove that the market for your product is vast and that customer acquisition scales quickly.

My new view is that early pricing is irrelevant.  If anything, pricing high in the near term is a hindrance.  It is better to under-price initially and raise prices over time – increasing dollar renewal rates will amplify future growth.  If you are lucky this will coincide with the exact time that you are looking to exit or raise capital.  That is a recipe for success.

The Second Generation of Social Media Marketing

Five years ago, in a three floor walk-up above an Indian restaurant, the team at Buddy Media unveiled the first social media management system.  I remember the board meeting like it was yesterday.  For the first time, someone could create a Facebook page without using Facebook’s obscure, in-house programming language.  This would be a huge time saver for agencies and advertisers.

Eventually the company added tools for publishing messages to the Facebook stream, as well as analytics and media buying.  Then they tacked on Twitter, LinkedIn, and YouTube. To this day these are the core components of every social media management system.

As much as this sector has been a success, there are still two unanswered questions in my mind about social media marketing:

1.) Do people actually follow brands on Twitter?

2.) Who visits brand pages on Facebook?

I ask because I don’t.

If you do, can you please write in the comments below what corporate account you follow so I can check them out?  Just curious.

In general, branded content gets low engagement within social media.  I know this because a picture of a celebrity’s bowel movement just got more likes in my stream than a great brand campaign.

It is easy to point fingers and say the problem is bad content.  With few exceptions (notably Red Bull, Old Spice, and Dos Equis) most brands are not good at social. Unfortunately the two areas that work in social – controversy and humor – are “off brand” for most advertisers.

But the real problem is that brands approach social with a broadcast mentality.  Social media was never constructed to be a passive, one-to-many medium like TV.  It is inherently an active, many-to-many medium.

Many-to-many communications are hard to engineer.  Some companies pay people to share content into the stream.  Other companies buy distribution from Facebook or Twitter via “sponsored posts”.  In the long run neither approach is great.  Humans have a pretty good bullshit meter, and if it isn’t authentic it won’t work in social.

For the past two years I have been working with a company that set out to create a system for native, many-to-many conversations in social.  They are called PeopleLinx, and we finally funded them this March.

For an advertiser, the arithmetic of many-to-many marketing is simple.  The average person has 200 first-degree connections across social media.  That means a large company, with 10,000 employees, is less than three degrees removed from everyone on the planet. This is the power you can unlock by engaging employee networks on your behalf.

Every brand has content that employees should want to share – whether it is recognition and awards, new product releases, or a specific program that an employee worked on.  This is the highest quality approach to social media marketing:  an authentic message delivered among friends.

It is important to remember that companies are just a collection of people, who collectively possess a vastly broader reach than a marketing department.  This idea has the potential to change marketing, and maybe in the process it can humanize companies too.

Top 10 CMO Mistakes: Site Search

search-366x274We have all done it – you visit a website and start looking for information, but quickly get frustrated because what you want is not on the home page.  Next you try site search, which takes you to a landing page filled with blue, indecipherable links.  Out of frustration you click on the back button a couple times, go to Google, and run a longer search.

There are a lot of surprising facts when it comes to site search.  For starters, a third of all website visitors will search at least once. And the activity I described above is very common.  80% of users who search will immediately leave the website afterwards (technically called a “bounce” in Internet jargon)*.  Put this together, and site search is costing the average marketer approximately 24% of all their site traffic.

This is why site search may be the biggest issue when it comes to online marketing and customer acquisition.  There is no other part of the conversion funnel that costs marketers as much of their traffic, or as much of their money, as poor site search.  The irony is that Google is the major beneficiary because most bounces go to for another search, and meanwhile Google is the world’s largest vendor of site search products.  A cynical person might think that Google does this intentionally.

The good news is that there is a solution to site search, which is a new technology called “Curated Search”.  Marketers are starting to get wind of this technology, and it has been adopted by brands like Apple, Blackberry, Xerox, and about a dozen others.  The leading vendor in this space is a company in Chicago called Elicit that we seed funded about two years ago, although there are others out there as well.

Curated search takes advantage of the fact that search activity on a marketer’s website is highly concentrated.  For a typical marketer, roughly 95% of searches will fall into one of 500 keywords.  In statistics this is called a power curve.  It is almost the exact opposite of the search activity on, where there is a flat distribution of searches across hundreds of millions of terms.

The good news about a power curve is that a human being can look at this data and make informed decisions.  This is where curated search comes in.

A typical marketer using curated search will group terms into a few essential themes like support, product information, location information, investor relations, etc.  They will create corresponding landing pages for these terms that are easily accessible from the search box. This way, users never see a page filled with blue links. Marketers can also inject advertising or promotions in the search experience, which creates a new opportunity to market items based on real-time intent.

Curated search may sound like an obvious best practice, but keep in mind that most marketers are still in the dark ages when it comes to site search.  If you ask a CMO on the street, my guess is that he (or she) doesn’t even know what terms are searched for on his website, or worse, he assumes it is one thing when in actuality it is something completely different.  This is because most CMOs outsource site search to the IT Department when it should be a marketing function.  Until that is resolved site search will likely remain as one of the top 10 CMO mistakes.

*This data is based on a representative sample from 15 marketers, who were using a wide variety of tools including Google Search Appliance, Baynote, and Endeca.

Top 10 CMO Mistakes: Online Video

MistakesLater this month I am speaking at AdWeek with my friend Shelley Zalis on the topic “Top 10 Mistakes that CMOs Make.”  Leading up to AdWeek I am going to publish a handful of items for friends who can’t make the conference.

This post is about online video, and specifically YouTube.  As a consumer I love, and I think they have great video publishing tools.  However, YouTube poses a number of challenges for marketers.

YouTube’s agreement with content creators is that you get video tools for free, but Google controls the advertising around the content.  It may seem like a harmless arrangement, but video hosted on the YouTube player is shared across the entire web, not just on YouTube.

There are a number of ways that this can cause problems for brands.  Just last week I was researching the Nike Fuelband and clicked on a Nike video.  I was immediately hit with a 30 second Reebok pre-roll ad.  I have come to expect this sort of competitive advertising with Google, but this video was playing on a Nike managed website because they were using the YouTube player.

This experience often happens without a CMO knowing about it.  Agencies occasionally use the YouTube player, naively thinking it is cheaper than grabbing a comparable video player from JW or Adobe.  As time passes it becomes harder to undo this decision because the content gets picked up and shared to numerous sites.  Google continues to own the advertising inventory on the video as it becomes further and further entrenched into the web, and if you don’t like this arrangement your only choice is to yank all the content and start over.

When it comes to video, brands have two basic choices:

1.) Purchase a video platform that provides the same service as YouTube (hosting, CDN, streaming, CMS tools, sharing, etc).  This is relatively cheap – $1.00 per thousand streams – and there are a dozen providers out there one could use.  With this choice you control the assets, the user experience, and the advertising.

2.) Use the YouTube player.  The tools are free, but you will likely end paying ten dollars per thousand streams to make sure that competitors don’t hijack your content.

It seems like a clear choice to me.  Google is betting that most advertisers can’t figure this one out, and so far they are right.


2 of 10