Web 2.0: The First 25,000 Users Are Irrelevant

For the past few months, whenever I talk to someone about a Web 2.0 application and hear that they already have “10,000 users”, I’ve been telling that them the first 25,000 users are irrelevant.

Josh Kopelman has a perfect post up today called 53,651.  This is the number of RSS subscribers to Michael Arrington’s great TechCruch blog, and is exactly at the core of the “first 25,000 user” issue.  Since there are 53,651 RSS subscribers of TechCrunch (at least as of 5/12/06) , if something gets reviewed there, it’s likely to get 5,000 to 10,000 users in the next 24 hours “just to try it out.”  As so many traffic graphs of these “TechCrunched” products show, there is a huge spike in use for a day or two, and then it goes right back down to where things were before they were TechCrunched.  For example:


Unfortunately, many entrepreneurs (and many VCs) confused one time “tryout users” with real sustainable users.  As an analytics freak (I’ve invested in a number of web / Internet analytics related companies over the past 10 years, including NetGenesis, Service Metrics, and now FeedBurner), you only have to ask two more questions to know whether (a) the company really understands its traffic / user base and (b) whether they’ve got the “first 25,000 user problem.”

Thanks Josh for the outstanding post and putting the gap between the Web 2.0 geeks and Mainstreet USA front and center.  Remember – the first 25,000 users are the same dudes (such as me) that play with everything.  Oh – and yes – I’ve fallen victim to this also.

It’s time to go La Vache hunting.

  • The first 25,000 users aren’t only irrelevant – they’re potentially poisonous.

    If the developers listen to the feedback of these early adopters (the initial audience) they’ll take the product/service down a path that increases the geek factor rather than down-geeking the offering.

    It’s a difficult place to be. Many of these developers are fully involved with the blogosphere, web 2.0, … and therefore are tapped into feedback loops that distort the reality of what needs to be completed in order to bring the product/service to the mainstream.

    It’s a difficult thing to listen to feedback from your initial users, the first 25,000, and do the opposite of what they recommend. You alienate your “support base” etc etc. Tough situation.

  • Brad:

    per Fraser’s comment, how should a startup reconcile the new product management concept of “Ready, Fire, Aim” with the early adopter poisoning scenario?


  • Excellent point Brad. It’s a testament to what Michael has built at TechCrunch (although I think his new blog design is way too busy and commerce-like), but to think of how many bootstrapped companies (and VC-funded to be fair) may be getting the wrong impression from early usage data.

    As an advisor and capital provider to a number of these entities, how early in the process do you make sure your CEO and executive team are aware of this risk?

  • Great article.

    By the way, did I miss them, or did you not reveal the “two more questions” to ask to see if a startup manifests this problem?

  • Andrew – I’m all about Ready Fire Aim and – in fact – there’s nothing wrong with being on Techcrunch and getting 20,000 people trying your stuff the next 24 hours (other than the inevitable server load and potential performance issue or crash that ensues.) You just have to make sure you have a broad spectrum of early users (or at least are getting feedback from the spectrum of users you expect to end up with.) Many things in life follow a normal curve – make sure that you understand the shape of this and are satisfying the bulk of people within two standard deviations of the mean rather than focusing on the tails.

    Jason: Oh – I tell them early and often. I love to measure traffic, but I love understanding the dynamics even more. The absolute measurement just doesn’t mean that much.

    Dharmesh: I’ve got to leave SOMETHING for another post…

  • That type of audience size is indeed impressive.

    For what it’s worth, the audience size of Adobe Flash Player 8 increases by five million successful installations every single day. Its audience surpassed the Firefox audience within its first month of release, and surpassed Windows XP audience size within its first three months.

    For wanting to reach influencers, then weblogs make sense… reaching 50,000 of the right people is very useful.


  • cc

    make a default web2.0 guest login – like user:guest password:guest the people can try the service and only sign in if they want a real account.

  • PJ

    Never underestimate your audience. How did geeks get to be geeks? Even Joe Sixpack understands the power of Internet marketing. The dive bar that I was at last night had a sign on the wall that said: “Add us on Myspace.”

    You have 80 million Myspace users learning HTML and CSS to customize their profiles. Now they’re getting interested in web metrics, to see who is looking at their profile. These are ordinary people that didn’t care about the Internet before–now they need to care. I predict that Myspace will create a *huge* market for web hosting.

    In a recent New Yorker article on Facebook, there was a quote that said something like: “If your profile isn’t online, you don’t exist.”

    The Internet is not a fad.

    Anyway, Techcrunch is just one website, and that 53K number doesn’t tell you much. Most people I know read blog content in context, not in an RSS reader. How many uniques/day do they get?

  • Jonathan Kolyer

    The take-home point here is, can you sustain the growth?

    One-time spikes are just that. Sure you can get giddy about your popularity. But talk to me in a month. I’ll bust your bubble so fast, it’ll make your head spin.

    So me a steady growth curve, then we’ll talk.

  • Around 10% of our total readership comes in from RSS (note that it is total readership not daily pageviews). A review of a good product on Techcrunch and the knock-on effects generate 30k+ ‘users’, depending on what the product is. I would say the more important effects of being reviewed on Techcrunch are the opportunities that arise from the exposure (funding, partnerships, potential employees, etc.).

    There is a very large gap between this web early adopter market to bigger and broader markets. At Omnidrive we saw takeup stall at 40k beta users, and then take off again when we reached out again. I would estimate from speaking to various people that the next mountain is at 300-400k users. At the moment I can only think of Flickr as an example of a service that has climbed that hill (not sure if del.icio.us is past that point yet). I am sure some of the social networks that had roots in the same audience grew beyond that point as well, the good social networks don’t seem to follow the same adoption curves.

    The other part of the equation is how sticky these users are, and that is all up to how good the offering is (95% of the people we send an invitation to create an account within 24 hours, of our total user base only a very small percentage haven’t come back)

  • Andrew

    The fundamental point is this. Successful sites will do whatever it takes to appear less successful. They block alexa. They keep off the web 2.0 radar. To appear on techcrunch would be a disaster to a company trying to fly under the radar. That’s why a higher proportion of techcrunch companies will not survive, compared to the sum of all sites being launched.

  • johnr

    53,651…Is this # even accurate? I admit I don’t know how feedburner comes up with this number but if it is just the number of RSS readers that subscribe to the feed URL than I’d guess it is well overvalued. I have tried probably 5 or 6 different RSS readers over the past two years (due in part to TechCrunch’s reviews of RSS readers). So if Arrington’s feed is in each of my newgator/rojo/bloglines/myyahoo/newsalloy/sage feed readers, doesn’t that mean I’m subscribed to TechCrunch 6 times? I’m guessing that many of his subscribers have tried out more than one feed reader, effectively subscribing more than once and inflating that number significantly.

    Brad, as an investor in both feedburner and newsgator you would know more than me here…am I way off on this assumption?

  • John – excellent comment that helps prove my point.

    There are actually two interesting FeedBurner “subscriber” numbers. The 53,651 number is the daily unique “subscribers” that pool the feed. However, as you point out, you are counted numerous times since you’ve subscribed to Techcrunch in multiple feed readers. As a result, FeedBurner recently rolled out a new metric called “Reach” which is the number of people that actually look at the post in a feedreader in a 24 hour period. While reach is probably a little overstated (since lots of people will “skim articles” but not actually read them, it’s still a much more accurate view of daily readership.

    For example, on Sunday, FeedBurner tells me I had 5,402 subscribers and a reach of 1,421. I don’t look at this ratio daily, but a 4:1 / 5:1 ratio seems about average.

  • Viral marketing is a great way to generate buzz, but as you’ve noted said buzz does not create a sustainable model.

    What helps you create a sustainable model? Interaction with your first 25,000 users. I think the first 25,000 users are the most relevant because they will help you to create value in your product. Not only do they help you realize how your product is best used, they help you realize essential features that you are missing.

    Now, when you get to user 100,000 you know what your users want and you didn’t have to experiment on the masses.

  • Ben

    I’m intrigued like Dharmesh is. What are “the 2 questions”?

  • Pingback: Be Here Now…by writing about it later()

  • Pingback: HipHop, PHP, and the Evolution of Language | Technosailor.com()

  • The incident was especially painful for Blippy, given that a New York Times breitling navitimer profile of the company appeared Friday morning, highlighting the growth of start-ups replica watches like Blippy that are designed to share personal information breitling evolution with the world.

  • It's a difficult thing to listen to feedback from your initial users, the first 25,000, and do the opposite of what they recommend. You alienate your "support base" etc etc. Tough situation.

  • Pingback: cash advance lenders()