Times Square Strategy Session – Web Startups and Customer Development

One of the benefits of teaching is that it forces me to get smarter. I was in New York last week with my class at Columbia University and several events made me realize that the Customer Development model needs to better describe its fit with web-based businesses.

Dancing Around the Question
Union Square Ventures was kind enough to sponsor a meetup the night before my class. In it, I got asked a question I often hear: “What if we have a web-based business that doesn’t have revenue or paying customers? What metrics do we use to see if we learned enough in Customer Discovery? And without revenue how do we know if we achieved product/market fit to exit Customer Validation?”

I gave my boilerplate answer, “I’m a product guy and I tend to invest and look at deals that have measurable revenue metrics. However the Customer Development Model and the Lean Startup work equally well for startups on the web. Dave McClure has some great metrics…”  It was an honest but vaguely unsatisfying answer.

Union Square Ventures
The next morning I got to spend time with Brad Burnham, partner at Union Square Ventures talking about their investment strategy and insights about web-based businesses. Bill and his partner Fred Wilson have invested in ~30 or so companies with 27 still active.

They’re putting money into web services/business – most without early revenue. It’s an impressive portfolio. By the time the meeting was over I left wondering whether the Customer Development model would help or hinder their companies.

Eric Ries in Times Square
For any model to be useful it has to predict what happens in the real world – including the web. I realized the Customer Development model needs to be clearer in what exactly a startup is supposed to do, regardless of the business model.

Luckily Eric Ries was spending a few days in New York, so we sat down in the middle of Times Square and hashed this out.

What we concluded is that the Customer Development model needs an additional overlay.

Four Questions
Just as a reminder, the Customer Development has four simple steps: Discovery, Validation, Creation and Company Building.  But it also requires you to ask a few questions about your startup before you use it.

The first question to ask is: “Does your startup have market risk or is it dominated by technical risk?”  Lean Startup/Customer Development is used to find answers to the unknowns about customers and markets. Yet some startups such as Biotech don’t have market risk, instead they are dominated by technical risk. This class of startup needs to spend a decade or so proving that the product works, first in a test tube and then in FDA trials.  Customer Development is unhelpful here.

Lean Startup

Use the Lean Startup – When There’s Market Risk

The second question is: “What’s the “Market Type” of your startup? Are you entering an existing market, resegmenting an existing market, or creating an entirely new market?” Market Type affects your spending and sales ramp after you reach product/market fit. Startups who burn through their cash, usually fail by not understanding Market Type.

Market Type Affects Spending and Sales Ramp

The third question (and the one Eric and I came up with watching the people stream by in Times Square): “What is the “Business Model” of your startup?” Your choice of Business Model affects the metrics you use in discovery and validation and the exit criteria for each step.



Business Model Affects Metrics and Exit Criteria

Web-based Business Model Exit Criteria
In a web-business model you’re looking for traffic, users, conversion, virality, etc – not revenue. Dave McClure’s AARRR metrics and Andrew Chen‘s specifics on freemium models, viral marketing, user acquisition and engagement both offer examples of exit criteria for Customer Discovery and Validation for startups on the web.

Eric and I will be working on others.

Add to FacebookAdd to DiggAdd to Del.icio.usAdd to StumbleuponAdd to RedditAdd to BlinklistAdd to TwitterAdd to TechnoratiAdd to FurlAdd to Newsvine

21 Responses

  1. […] This post was mentioned on Twitter by Sachin Agarwal and Ryan Urban, Brendan McManus. Brendan McManus said: Times Square Strategy Session – Web Startups and Customer Development http://bit.ly/1hP3S8 #startup #advice […]

  2. Great stuff. I would love to get your feedback on these thoughts:

    1. “Does your startup have market risk or is it dominated by technical risk?” Let’s not let startups use this as an escape hatch to avoid customer development. Almost all Web startups are dominated by market risk. Even startups that are dominated by technical risk have the customer validation risk of finding positive ROI distribution in a large market.

    2. ““What’s the “Market Type” of your startup?” I love the theory of market type but I think it is an advanced topic. For startups in customer discovery and validation, I would boil it down to: “1. Don’t launch. Don’t spend time and money on getting the word out about your product beyond what you need to do discovery and validation. 2. Your best bet is probably taking an existing market and serving a portion of it better than the competition, through lower cost or a different set of features. End of theory.”

    3. “What is the “Business Model” of your startup?” I’ve never thought validation required revenue. It is the strongest form of validation and the hardest to misread. But not the only form. Bad forms of validation: “They said they would buy it!”. Better forms: “They signed an LOI. They spend an hour on our site every day and we know why because we interviewed them.”

    I like the idea of making a list of types of validation.

    • Nivi,

      Thanks for the thoughtful comments.

      1. Agreed. I put this distinction in place as I often have students starting biotech companies in my class sitting next to ones doing web apps. The market versus technical risk between the two is enormous. While one could argue that biotech startups also face market risk, their first decade is dominated by whether the science will work.

      2. “When to launch” is not related to Market Type, (how to launch is.) It is related to business model. Not sure I would make the same definitive statements about best bet for Market Type.
      Your suggestion for picking a niche assumes you’ve thought through the rest of the Market Type alternatives.

      3. Validation requires revenue/orders if you’re in a business that sells product.


  3. Thanks kindly for your reply.

    1. “While one could argue that biotech startups also face market risk, their first decade is dominated by whether the science will work.” Great quote. Customer development is about squeezing out risks in the right order. This overlay adds another upfront risk (technical risk) that needs to be squeezed out first, if necessary.

    2. I totally grok what you’re saying. But I haven’t seen startups do anything useful with the theory of market type in the discovery and validation stages, except in re-positioning the product to try to get better product/market fit in discovery. Thinking about market type seems to bog startups down and they don’t get anywhere with it (most startups don’t get one-on-one time with Steve Blank). That’s based on my own personal experience and I would love to experience counter-examples. So my solution has been to punt on the topic and say what I wrote above.

    3. Isn’t the point of the third question that the definition of validation should be expanded to things outside revenue/orders? Can’t customer development work well for other effective definitions of validation and isn’t that how the startups that you’re talking about (Tumblr, Twitter, Facebook, WordPress, etc.) stumbled into customer creation and company building? They may not have had revenue but they had validated learnings that got them into customer creation (e.g. AARRR). Then they can run another customer development process on their revenue hypotheses later. I like this post by Fred Wilson on this topic: http://j.mp/26pIED

    • Nivi,
      I think I understand the problem with #2.
      If you look at the diagram in the post, the consequences of Market Type don’t come into play until later in a company’s life. Yet a startup makes the positioning decisions early on. My experience here says that getting Market Type/Positioning right can be the difference between success and failure for a startup. This a hard concept and even harder to implement. One can even make the case that getting Market Type right is the most valuable thing a marketer can do to add value and affect success in an early stage venture. Perhaps the solution to the “punt on the topic” is to try to get a marketer who has a feel for the “Market Type/Positioning” question on a startup’s advisory board once the Discovery data is in.

      As for #3. I agree completely. That’s exactly the point I was trying to make (perhaps badly) with the post. Instead of trying to make explicit exceptions in the Customer Development model, just add an overlay for your business model. This “business model overlay” would contain success/failure criteria for Discovery and Validation appropriate to your business model. Dave McClure’s and Andrew Chen’s are perfect for web-based businesses and completely inappropriate for others. And vice-versa. The goal is to ultimately be able to plug in AARRR-like criteria for each type of business model.


  4. Interesting discussion. Especially because U are 3 out of 4 leaders (Paul Graham is still missing) in developing entrepreneurship in the US, as far as I can judge that here from Germany. So please, go on!

    @steves number 3: What is a different company to a product selling one? Don’t all companies sell a product in the end? Even a service is a product. And selling itself means getting money for giving some amount of anything. Or do I lay too much emphasis on that word? Sorry, I don’t really get Ur point here, Steve.

    • Erik,
      How refreshing to realize that there is still a part of the world where the only rational business model is selling goods or services in exchange for money.

      That’s no longer the case in Silicon Valley. There are variety of web-based business models (see here) where the initial goals are not revenue.

      The point of the post was to observe that for the Customer Development model to be meaningful it has to work for all business models (including these new ones on the web,) not just those that charge money in exchange for a product.

      Therefore the way to solve that is to ask, “what business model” am I using. And then set up the correct metrics in Customer Discovery and Validation.

      Slightly off-topic, there’s a good example of an entrepreneur going through the “Discovery/Validation” process for a web business that charges for its product here. Even though he isn’t couching it in the “Discovery/Validation” language you can see that his metrics for success are radically different than one delivering a physical product.


  5. A startup that does not aim to generate revenue sounds extremely disturbing to me. Back in 1999, it all ended very bad.

  6. […] This post was mentioned on Twitter by Calle Hunefalk and [Startup Digest], [Startup Digest]. [Startup Digest] said: Times Square Strategy Session – Web Startups and Customer Development http://bit.ly/2byCV2 #custdev #startup […]

  7. I can’t keep up with all the useful information you’re sharing here Steve. I’m trying to wrap my head around the bigger picture, and then boil it down to my current project/pre-business.

    Specifically we’re working on building information stream handling tools (reducing large lists of incoming links and text into topical tags that can be searched or manually dug through) as well as connecting ads to a users interests (same tags now used for relevant ads/plugins).

    I know we have to create something users find valuable (the two-way search, games?) as well as a potential revenue model (why not start with a framework for revenue?). But I’m facing both technical (can we do what I think we can do), and market (are they people besides the team who will love the idea) risks.

  8. Hi
    This is really nice article, Thanks for sharing to us.

  9. I think it’s possible you’re being too generous to the revenue-challenged. What your CustomerDevelopment model seems to do is reduce risk by avoiding scale until you have some basis for predicting success. In the past you have focused on revenue because (a) getting real orders at any price over 0 proves whether people value the offering at all, and (b) setting a real price via real orders is much more realistic than market research forecasts.

    The revenue-delayed models seem to go against that risk-reduction rule. For one thing, the ability to convert any non-revenue metric *into* revenue (step 1: collect underpants) seems like the biggest unknown to me, completely unpredictable from company to company. For another, many of these models end up needing *huge* scale, so a small change in growth-rate or point-of-levelling-off makes all the difference in live-or-die.

    Or am I missing something?

    • Nope, you’re not missing anything. I still have a hard time on the “build it and they will come” business models that go for something other than revenue.

      That said, the reality is there are a ton of web-based business models that do just that. Examples here. Given that entrepreneurs are going to execute these models, and VC’s are going to fund them, my best hope is to provide an overlay for the Customer Development model that allows them to keep score.

      Therefore instead of a dogma that says, “if you’re giving it away for free, you are forever banished from Customer Development,” I am proposing a way to bring even the misguided into the fold. And that is: declare what your business model is. Set your exit criteria for Discovery and Validation. See if you can get to scale.

      Does this help?


  10. Thanks again Steve. You say, “Market Type affects your spending and sales ramp after you reach product/market fit.” Based on your discussion in the comments, I think we should add “Market Type also affects your positioning before you reach product/market fit.”

    • Nivi,

      Yes. I was trying for brevity, but you are correct.
      Market Type and positioning go hand-in-hand. You start working on it almost from day one in Discovery.

      Thanks for pointing it out.


  11. […] Steve Blank and I have a good discussion about his and Eric Ries’ new customer development overlay. Regarding the Atlas Developer Beta, John Gruber says that “Access to the beta program is $20″ — this is the new way and I like it. Fake Steve Jobs always has better analysis than WSJ, NYT, TechCrunch, Economist, and your mom combined. I don’t think you need operational experience to be a good VC but Mark Suster’s great new post illustrates the difference. […]

  12. Steve, a great post is judged by the level of discussion it manages to spur, so in that metric, this is a fantastic one. I think it’s important you and eric are starting to tackle the issue of the ‘other’ business models, and I think this path will eventually converge to a good addition to the customer discovery framework. However, I’m not sure it’s as simple as dropping just an ‘overlay’.

    The reason is, that while the revenues metric is a rather hard, cold type of metric, it’s hardly the case for the non-revenue metrics that people will slap on if you’ll give them half a chance. So founders might be tempted to first shoot the arrow and then draw the target around it, by choosing the ‘soft’ metrics that puts them in the best light.

    So certainly the suggestions above for metrics are good ones, but I would claim that the right metrics can be deduced almost mathematically for every type of business. Just like revenues are the right metric for businesses that aim to sell products directly, one can probably deduce a global metric for other types of businesses. For enterprise software, you taught us, the first step is to get ‘some’ customers to buy, and identify how much they’re willing to buy, not to measure how many are buying. Similarly, I’ll be very cautious about metrics for web apps that just measure traffic, eyeballs etc. I’ll be much more interested in lower traffic pattern but one that is highly clustered which suggests word of mouth and user satisfaction. In fact, there’s a decade old marketing theory that uses such measures of entropy to identify early success of products: http://portal.acm.org/citation.cfm?id=1024172

    Looking forward to seeing what you come up with as this chain of thought develops.

  13. […] Times Square Strategy Session – Web Startups and Customer Development […]

  14. […] Times Square Strategy Session – Web Startups and Customer Development […]

  15. Feels to me like the validation options are to go for:
    – users
    – API adoption (which is to say a bit of inertial lock-in)
    – data (not necessarily driven by getting users, and presumably data you can make some sense of and get more of than other people)
    – deals signed with strategic partners (distribution or advertisers)

    and then there’s always money.

  16. Hi Steve, thanks for the awesome post, but, what was question #4? 🙂

    Keep teaching!

Leave a Reply to steveblankCancel reply

%d bloggers like this: