30/10/10

Comments (Archived):

1. I think I’ll go back and reread “Social Media’s Secret Weapon – Email.” That being said, when dealing with mobile, I get excited whenever I receive an SMS or in-app notification as well. Even if it’s just to remind me to take my umbrella to work that day.

2. Hard to understate how important and useful open sourcing a piece of information like this is for entrepreneurs. Transparency helps us build stronger, more informed, businesses. Thanks.

   1. You like buzzwords. Cool…

      1. You like to hate about it… Cool.

         1. Your original comment is hilarious, it’s just meaningless and loaded with a bunch of over-hyped SEO/marketing terms which show how little you know about what you think you do. Carry on…

            1. You can debate and discuss without hating on people. Please work on that if you want to comment here

3. Does the number of dormant users also follow a consistent pattern?

4. This is part of the new Marketing Mix. Great insights.

5. The problem with that 30% number is that it’s tied heavily to the user growth rate. In other words, if your user base grows 30% a month, but everyone stops using the product after 1 month, then you still get 30% of your users having used the product that month. (Obviously growing 30% per month is pretty crazy, but I’m just keeping it stupidly simple.) (Actually, Twitter was growing roughly that fast in 2008.)When I pick a single activity/retention number to look at (internally), I tend to drop accounts added in the past month from my universe.And once in a while it’s good to take a bunch of slices:* what % of 2-month-old accounts came this past month* what % of 3-month-old…* (I do ~6 single-month cohorts, then maybe do a 7-12-month group, then a 12-18…)For any site/app that’s not heading for Twitter/Facebook scale, those retention numbers tend to have a Medusa affect on management.

   1. rightthese numbers are best seen once you have a critical mass of users

      1. What is the critical mass figure that applies generally, or is it different for the each business, model, and market? Thanks.

   2. It’s not hard to remove growth rate as a polluting influence to the retention/engagement data if you use cohorts. You can even get close enough by knowing new user registrations/day.As for 60 day and 90 day looks, this might be controversial, but for an immature/early stage product I personally focus much more on the first 30 days than long term. Really, day 1-3 retention is likely the strongest leading indicator. Put simply, I’d rather be in the business of deepening a product that had me madly in love with it for 30 days, than trying to take a product that I come to once a month over 90 days but feel “meh” about it. If you are trying to form the daily love of a new consumer internet service, then I’d be more interested in a really early service that drive 4-5 days active/week for a month (~45/35/10) than a service that drove 1 day active but had the user come back again the second month (~55/5/5). Of course there is no magic one number for success, so taking a look at 60 day cohort retention trends can certainly inform, but at least for me not my leading indicator really early.

   3. I agree that there is an interesting issue about new user signups in a given time period (day,month), and whether those users are counted as ‘active users’ for the purposes of the active user percentage. One tactic is to exclude the sign-up visit from the count, but include later visits from that user in that time period. So if you’re growing 20% a month, and those 20% only sign up in that month, and don’t repeat-visit during that month, then none of those new users will count as active that month. So you could say an ‘active user’ is someone who visited the site in the relevant time period where the visit wasn’t a sign-up visit. 

6. Another example of the value in patterns of data. Would be interesting if there were an anonymized, shared data set among all web/mobile services. Could be a nice boost to overall performance and innovation.

   1. Igniter, Google Analytics has a feature like that. Website owners can opt-in to the feature that shares aggregate data with other sites in their same business vertical.

7. Great insight. I think we have to consider that 30/10/10 equation largely applies to applications that are heavily built out good product, in a slow growth phase.  

8. By the way… I have shared some of these metrics with an investor as well using Flurry feature to invite someone on an account. Now that discussions are over, Id like to stop allowing him to access our data – but cant find a way to do that and Flurry support is not reactive… Metrics remain sensitive info…

   1. Go to the “manage” Company link in the upper right.  There you can remove a user’s access by changing their role from Viewer to Suspended.  You have to be an Administrator to do this.

      1. Thanks .. I was probably under the “viewer” account. Maybe a “viewer” should not be able to invite other viewers if he is not the administrator ? 

9. Re: email to change metrics:  I remember hearing somewhere that LinkedIn has an algorithm that will determine when a user is most likely to be in front of their computer looking at email based on past clicks throughs.  It will only send that user emails during certain time periods.I’d be surprised if their the only service doing this, but nevertheless, it’s still pretty amazing.  Not only are they bringing you information you’re more than likely going to engage with, they bring it to you at a time you’re MOST likely to engage with it.  Powerful.

   1. Wow, sort of sneaky (in that they are monitoring your behavior), but I guess, not really illegal or invasive; it’s a clever idea, hadn’t come across it before.

      1. Ha, yea not illegal or invasive at all. And, when I think about it, I almost prefer mail to be sent to me that way.Sent wirelessly

         1. I just noticed a patent application earlier this week – written by LinkedIn’s product team that deals exactly with predicting user behavior based on past actions and leveraging that for other tasks.

            1. Wow, thats interesting. Good to know.

10. What is so important in having somebody visit an app/site unless that visit converts into an action that causes them to come back again? That will be an interesting metric by itself. It will be valuable information to know what interaction an user has after they click the email notification.

11. I’m intrigued with this. Hadn’t thought about this unique lens into the industry that this data could provide.Actually a strong value ad for your portfolio companies if you can share this across companies in some way.Measurements are only useful in perspective with norms. You have that view. This should get everyone who is reading this thinking…

12. Thanks Fred. Question: are the 30% of monthly users discrete from the 10% daily users? For example of 100 users, 30 use monthly, 10 daily…does that mean that 30+10=40 use every month or do the 10% daily users factor into the 30% monthly user #?

    1. The 10% is included in the 30%. That’s how these numbers are discussed.

       1.  right

13. Well said!Once you have these metrics, it’s usually helpful to break them down by weekly retention cohorts. I think you wrote about that before… If anyone has questions about this stuff feel free to ask me. I work with metrics all the time.

14. Am I wrong to say that retention of active users is the metric most VCs are looking for? Traffic can be bought, engagement shows love for the product. Love for the product eventually turns into the ability to sell something to those who love it. Best sale is organic within the product. 

    1.  we say that we invest in large networks of engaged users, so yes engagement is ciritcal to us

       1. Fred I have a question in terms of advise. We are just as focused on the business partner side of our platform as we are for our consumer side. Is a better strategy to produce partner features that are frictionless and easier to implement (but also free) so they adapt wide early usage, or is it better to show that these partners are willing to pay for deeper integrations based on revenue and subscription based models. So what are you more interested in the ability to have more partners early or have less but show the revenues you can produce from them? 

          1. i prefer it when our portfolio companies focus on the consumer first and the businesses later

15. Just the kind of information I was looking for and worrying over the last week. Many thanks.

16. Great post, Fred.  I’m glad you called this out.  I was talking to a colleague the other day about how it’s not only important for markets to understand these kind of high level metrics, but it’s critical for employees (regardless of their role) to have these kinds of metrics on the tip of their tongue. I encourage people to always be able to speak to new users (total registered, weekly and monthly registered), and user activity (weekly and monthly users and weekly and monthly users divided by total registered).An employee that has this kind of insight/perspective can look at the business far more intelligently, can talk at a high level with senior management/the board/customers, can benchmark against competition, partners and new products and can quickly identify when there’s a material spike or decline in a certain metric and understand its causes.I think it’s key for all up and coming web services to make these metrics very public in their organization and encourage a culture where employees understand them and obsess over the levers that cause them to increase and decrease over time.

17. Hypothetically, if someone has a new social application, and 1000 people have registered…this means 300 people will use it today…and 30 people are using it right now.  What is the number of registered users where the platform starts to support itself?  Stories from entrepreneurs that have done this are welcome.

    1. I imagine it’s different for every service/application.But it’s a nice baseline to know you should reach to know you will likely achieve great success. If you can get those kind of numbers immediately – you’re doing something really right.

       1. “different for every service/application”…agreed… to consider a Twitter/Facebook style network vs. a Quora/Stack Exchange/ Reddit/ HN vs. a Disqus…Some more a rifle approach to beginning (Tw/Fb) vs. some more a shotgun approach accross a community sector (SE/HN)  

    2. Robert, I think there is an error in your math above.  If 1000 people have registered, Fred’s 30/10/1 rule would mean:- 300 people will use it this month – 100 people will use it today- Maximum 10 people will be on concurrently Fred, I am interested in the relationship between the daily active users and the monthly active users in your benchmark:- If 100 people use the service each day, then over 30 days, there would be 3000 uses.- One extreme is that this use is distributed evenly across the 300 monthly active users, so that each uses the service 10 times per month- The other extreme is something like 93 users used the service every single day (93 x 30 = 2790) and the other 210 users used the service once this month.  I think this makes a big difference when evaluating behavior and trying to move users into a more engaged pattern.  You’d want to segment on the 93 vs the 210 and see the difference in their characteristics and behavior.  This is something that my company, Apptegic, helps other companies do.  Would you be willing to share benchmark rules of thumb here?  in your portfolio, do you see closer to 10 uses/month per monthly active user or are the monthly active users more typically highly bi-modal with a group of every-day users and a group of once-a-month users?

18. I really like that you are calling this out.  Mark Suster wrote about another one of these Universal Laws of Internet Dynamics which is the rule of 90/9/1 related to user generated content.  Here is the interesting thing I see.  Companies tend to be a little over or a little under these laws.  That reality is something that can be seen as either negative of positive in the form of opportunity or a company that is positioned uniquely in the Internet Universe.  Either way, understanding where your service is relative to the rest of the industry is very valuable in terms of whether you should focus more on your product or more on scaling.

19. Speaking of Foursquare, any thoughts on Sonar.Me? I’ve been hearing some positive comments about it on Twitter.

    1.  they were a finalist at TC Disrupt where i was a judge. i like what they are doing

       1. Thanks Fred.  I don’t consume a lot of media these days but this post found me via WOM…and I’m very grateful for it.Re: notifications, email or otherwise, one thing we think a lot about @sonarme:twitter is “how we can we leverage the engaged users of other platforms to provide new value for our users?”  Rather than limiting ourselves to solving for our own small ecosystem, we try to answer the question “what are the people our users care about doing – be it on foursquare, facebook, or twitter – that is worth telling sonar users about?”Of course, notifications are still incredibly primitive.  I look forward to the day when sonar tells you what it just did for you, rather than what just happened. e.g. “We saw you just ‘liked’ Kid Cudi on facebook so we took the liberty of looking up their upcoming shows on songkick at your favourite 4sq music venues.  Have fun!”It’s going to be amazing.

20. What does the ratio look like for companies doing a good job of changing the metrics in their favor? If 30/10/10 is typical, what should companies strive for?

    1. i don’t have access to FB’s numbers but i am confident they see way more than 50% of registered users on a monthly basis

       1. Fred – this rule tends to be generally true for ‘successful’ apps in this space though its hardly a ‘rule of physics’. In our experience in the social gaming space, we have seen that conversions that you mention of monthly to daily to ‘currently active’ are generally in the vicinity. For more on the numbers:1) Monthly Active to Registered Users – I think putting in 30% as a constant number here or even a ‘standard number’ doesn’t really tell you that much since this number is sure to decrease over the lifetime of the app. (unless it is a utility like Facebook, Twitter, or Foursquare). Though if you can get 30% of your registered users to become ‘activated users’ – you are doing pretty well in the social gaming space. Even if you are to convert 30% to active users, you will start losing quite a significant chunk of those activated users in 4-12 weeks (the typical lifetimes of users on social games.) This effect will push down the ratio of MAU / Registered Users over time, irrespective of the quality of your experience. (again, use cases for the app will also dictate what this number would look like, and whether you can say it is good for your category.) 2) Daily Users to Registered Users – I am not even sure that there is a reliable actionable metric here. We generally tend to measure Daily Actives to Monthly Actives – and that varies from 8% to 25% in our space depending on how enjoyable your game is, and how the various in-game incentives and messaging (such as email, etc. work.) 3) Current Active to Daily Active – this number is spot on. 

          1. Great feedback and with real numbers. Thanks so much

21. For a moment I thought this was going to be a post on basketball – e.g. 30 points, 10 rebs, 10 assists. Maybe you could call this the “Oscar Robertson Ratio” 🙂

    1.  done. i love it!!!

    2. I would call it the web/mobile (startups’) vital statistics ratio or the T-shaped ratio:303030   10   10Just kidding 🙂

22. By the title I thought this post was going to about business plan length, slide deck, and pitch time.  I was going to complain that 30 was too many pages to expect a potential investor to read in a business plan.  😉

23. I would imagine this is the ratio for a startup at a certain scale, where they have attracted  mainstream users and gotten mainstream press. Presumably in its earlier days a startup like Foursquare or Zynga would have had higher engagement rates, as the community was smaller and more self-selected.Is there a corollary for the engagement rate a small/brand new startup needs to hit the inflection point needed to keep growing into a Zynga or Twitter? I would imagine it’s a higher number than 30/10/10. For example, it wouldn’t surprise me if Turntable.fm was running at a 50/25/15 for their first couple weeks, and is now gradually reverting to the mean.

    1. You would think, but that is typically not the case except in extremes (say, Myspace in massive decline, or the first 10k users).There is downward pressure on engagement as you get to broader audiences, and as a product ages, but it’s remarkably stable past very small threshold of early adopters. Partly because you have increased capacity to drive awareness & retention especially for a social product, (eg. “your friend just tagged you in a photo). 

    2.  right, these numbers are for services “at scale”

       1. So are you saying that at scale, most services you see are seeing DAU/MAU ratio of ~0.33? Seems high.

24. Past and future prediction of this ratio?shortmail.com/artviarhyme

25. One of the best posts I’ve read about web and mobile metrics. Glad @SteveWoda tweeted it.

26. Hey FredI hope I didn’t read it wrong, but why don’t you write it as 30/10/1 ? If I understand you, your final 10 is 10% of the second 10, so it’s 1% of the overall total. 30/10/1 is easier to explain because they’re all percentages of the same thing (instead of having the 3rd be a percentage of the 2nd as with 30/10/10).  Or am I confused?Thanks – and…. interesting!Terry

    1. I agree.  30/10/1 is more consistent.

    2. I imagine it’s because that the users included each day in the 10% aren’t always the same, therefore the base active users probably isn’t “1%” but that’s how many are current active out of the pool?

    3. you beat me to it – i was thinking the same thing

    4. good point

27. How serendipitous! I just read the Wall Street Journal article about all the wacky new metrics that these news companies like Groupon are using that are essentially counts of their revenue without calculations of their expenses, and counts of sign-ups and not 30-day unique users, and now there’s your article.But…I still don’t think you are confronting the vast fakery here, really, and of course, you have a literal vested interest in doing so.Let’s take “ratio of registered users/downloads to monthly actives, daily actives, and max concurrent users (for services that have a real time component).”Well, yeah. Kind of. But let’s take Linden Lab, makers of Second Life, which used to give us more statistics than any other of these companies so we had a lot of interesting metrics to follow. They might claim millions of sign-ups. But they only had 70,000 concurrency, and only 1.5 million monthly uniques. This dipped and then climbed again with better sign-up procedures and a “basic viewer” option to get all the geeky clutter out of the browser.But what really counts? For them, it has to be server space sales (sims), currency fees, a few other consulting or trinket sales — they get about $75 million in income from the 32,000 sims. We don’t know their costs, but the server farm for those sims (4 per server, 8000 servers), the payroll, etc. — well not sure it leaves much left over, which is why they shed a lot of staff recently.What do *I* count as someone actually running a little business there? I count “number of people who spent more than one dollar inworld per month”. So that figure was 450,000 — and the range of those spending $1 to $1000 was a very steep spike of course. The So more than half of the 1.5 million visitors are just gawkers at other people’s content, they don’t even buy any content but just burn up server energy. Worse, they don’t even buy the fremiums, as there are only like 90,000 monthly subscribers.The 450,000 dipped, went back up…but they ceased to publish the number under a new CEO who doesn’t believe in publishing game god statistics.There was another interesting figure which isn’t perhaps so interesting for you, but it’s the sort of number you should get from Etsy:Number of people who have positive monthly income flow. So let’s say the service costs $9.95 — can they sell stuff that puts them in the clear of that fee and gives them a take-home?More to the point, do you have people who can make $2000 a month in real money and feed themselves and pay rent? They used to keep the positive-flow people numbers too, and you saw there were perhaps 200 people who made $5000 or more, several thousand who made $2000 or more per month, and interestingly, something like 25,000 who made $50 or something. That’s a lot of real money for something on the Internet that normally, on other platform, takes your attention, takes your dollars, and gives you back either a free service that is a time suck, or a lot of digital content you can’t resell. Like Farmville.So unless you are tying those users to dollar income for the platform owners, the metric is fake.

    1. yup, at the end of the day you have to pay the billsmaybe i’ll post about dollars per daily active benchmarks

       1. Huffington and Atlantic only became profitable due to special event fees and lead-gen. I would like to see social media and new media pay out better on real things because lead-gen is stupid and also a bubble and $1000 special events are getting harder and harder to justify.

          1. Tying the engagement to a financial metric is the little thread here that will have a long story because all groups are not equal.  You’ll be able to pay a lot more bills with an exclusive group that’s expecting to spend money as a part of the engagement.  For example, I’d like to see these equivalent numbers for Bloomberg’s services.Also, I doesn’t seem like the 30% takes into account e-mail open rates.  If your content is interesting enough that people open your e-mails, couldn’t the engagement number be a lot higher?  For example, I’d assume people can get e-mails from Etsy waiting for the that just right thing and log in only once every 6 months.

    2. totally agree with this. In a previous life we used to refer to these you speak of as KEY BUSINESS DRIVERS. some mckinsey chump dropped this gem in on us – but to be fair when we actually adopted the approach and got company wide buy in to its importance it gave us an amazing lens in to the guts of a fast moving and evolving company. i learned a few things from being a part of this1. All measures MUST have some intrinsic connection to $ either on the revenue or on the cost side. you say unique users – i say how much to acquire. (marginal costs of acquisition of a new user)you say concurrent users – i say whats the cost (bandwidth hogs? using text etc)2. All measures MUST have continual review. What you think is important today – wont be tomorrow. Every business has phases to maturity – where are you on this line? I’ve seen companies watch metrics to perpetuity and it becomes painfully obvious that they are non meaningful and the boats been missed on whats now important.but at the end of the day – for me at least – if you cant tie a metric to a $ then your head is in the proverbial “bubble” – you are chasing notions like “get big quick” and “first mover advantage” and i’m still unconvinced that this approach wont end in failure.

28. oSlightly tangental: what is it about the nature of humans on the internet that causes these ratios? Knowing what causes them means finding ways around them.

29. Fred, I like the way you think, I like the way you work, and I just generally like ya.  This posts, and the one earlier this year on the 100/10/1 rule (1% create, 10% engage, the rest listen) are wonderful ways for entrepreneurs to be thinking more dynamically about networks.  I appreciate your valuable insights, and you’re authenticity.  Thanks,

30. Was going thru my email inbox just now and realized that more than 50% of emails received recently are notifications. And the trend only seems to be growing. More and more I’ve been noticing that friends have been emailing me not directly but via facebook messages, twitter DMs, linkedin, foursquare or whatever other app they have open at the moment. Maybe soon we will stop calling it email inbox and call it notification clearinghouse instead 😉

    1. there’s opportunity in that insight

       1. My buddy Josh Vickers startup http://www.HubKick.com addresses that pain point – i.e. “not all email is equal” and it works across twitter, facebook, etc.

    2. it is driving me crazy – do you ever feel over-notified?

       1. not yet…but then again, I’m probably behind a good percentage of those notification systems…it’s something I tend to build a lot of because they are things I personally want/use for each new service that grabs my attention…

          1. I’m at the point where I get so many notifications that my new friend is my bulk folder. Which doesn’t make me happy.

             1. If you’re using gmail it’s very easy to create filters.  Take all those notification emails and dump them in “social” for example (make sure to skip the Inbox).  Poke in there if you’re feeling lonely, otherwise enjoy the new sense of freedom.

             2. Have done, then I get more, or the damned things learn to get past my filter (i’m looking at you, facebook)

       2. A little bit yeah but its still quite manageable … However its the fact that the trend is growing means that at some point it’ll cross that threshold. 

          1. It will be interesting to see what happens when that day comes.

    3. Seems like a medium is a good form of communication until it becomes too mainstream.  For example, snail mail used to have a high signal to noise ratio.  Now it’s just dead-tree copies of bills you already e-payed for.Same with SMS, and unfortunately now Twitter to some extent.  Pretty soon we’ll be getting mobile notification spam if the price is right…

       1. Good point. Its almost an unwritten rule … anything popular (mainstream) that can be gamed will be gamed and will have spammers / scammers gravitating to it. The payoff is just too high for them not to try it. And it usually ends up being a losing battle for the medium. And I guess that is true not just for popular communication mediums but also popular platforms. Look at the content farm spam in Google search results or the number of viruses on Windows or the Facebook newsfeed or Craigslist housing ads etcSpeaking of snail mail, the USPS just announced they’re closing down 3700 locations. 

31. Fascinating article Fred. I spent the last couple of months analysis the audience data from a set of international broadcasters. As Internet becomes a more important, or in some cases the only way of reaching an audience in another country, they have to find ways of guesstimating the bandwidth capacity they will need as well as the number of concurrent users (on both normal days and when there is a major breaking story). I’m finding major differences in metrics here. In the broadcast biz in Europe about 1-2% of people listen to streams on line (i.e. in a player) as opposed to people listening or viewing over the air, partly because the way the broadcasters have set up their players, you need a different player for each brand of channels. Imagine having a separate TV or radio in the home for each content provider. Crazy, but true.The interesting part comes when broadcasters stop broadcasting, either for economic or political reasons. Does the audience search out an alternative, does that mean a huge rise in those daily user figures? So far, I can’t find any evidence that it does. In some recent cases, changing medium means a catastrophic collapse in the figures. They just dream of having 30% of their audience using the service monthly.If this site can be regarded as a “service” to the VC community, I wonder what the figures would look like for those who have somehow signed up for a feed of this blog? I’m guessing, a very different daily/weekly/monthly user ratio. Getting an advisory that this blog has been updated drives me here to read the article and cherry-pick from the comments.  

    1. There are 120,000 RSS subscribers to this blog. That’s about 30% of the monthly uniques

       1. Interesting.  If I back into the numbers from 120K RSS subscribers monthly uniques are about 400K.  Total readership should be about 1.3 million if the formula holds true for AVC.1.3M   Total readership400K   Monthly = 30% of total readership133K   Daily = 10% of total readership13K     Concurrent = 1%FWIW, 30% of monthly – i.e. RSS Subscribers @ 120K is almost the equivalent of the projected 133K daily.  RSS subscribers are probably not equal to Daily visitors but with significant overlap.  Fun Stuff.

32. Just wanted to say thank you for sharing this – it’s exactly the kind of data entrepreneurs struggle to guess at in planning or trying to benchmark against norms.

33. A lot to think about.  Pushing any of the percentages up 10% will have an impact, much less if you can raise it a third.The notification side of things is the way to do it, but in the long term, we need to be careful so things do not get too jumbled for the user.

34. Could you use these numbers as a metric to judge the popularity of the site on?  If, for instance, you have 100k users, but only 1k log in each month, does that mean you’re not attracting enough use?

35. We’re going to launch a new service in the next few weeks, and its nice to know that if we can stay at or above that 30/10/10 rule that we’re not doing half bad. Thanks Fred, that puts my mind a little more at ease.

36. Something else just occurred to me – in most markets, 50% (or more) of Facebook users log in every day. Something I’m sure they’ll focus on more if/as growth figures start to plateau.

    1. Bear in mind that that is 50% of their monthly actives that are daily actives. Facebook doesn’t announce the total registered count. 

37. Love your last comment…with the breadth of options available to consumers these days, 30% activity with any site (to registrant) basis is fantastic. 

38. Question on the *mobile app* metrics… does 30 represent total downloads or remaining installs?  I’d be interested to know what the uninstall / delete rate is for these apps…  Bueller??

39. Interesting figures. We’re a mobile social network (think of a better Twitter over SMS) in a part of the world not known to create this stuff: Pakistan. We believe that Social Networking is too important a utility to be limited to a few internet users and Pring (http://www.pringit.com) is determined to make social networking mainstream in Pakistan and similar territories. Over the this year we’ve seen rapid adaption and now have 1m+ users. I was curious to see how we stack up against your 30/10/1 generalization and it comes down to 45/25/NA. The last part is not applicable because there is no parallel to concurrency for SMS.Similarly, would you call any user who receives SMS as active despite no ‘active’ participation? The figures above include these users. Internally we classify users as either Active (sends SMS updates), Passive (receive-only) or Dead (neither). Have you come across metrics that could be comparable? 

40. Great & timely post. I recently made a presenation to the higher management about the stats of our recently published app ( which does follow this rule) & had some explaining to do with different usage patterns etc.