Posts Tagged ‘growth curve’

Zuckerberg: Online Sharing Is Growing At An Exponential Rate (And Users Are Sharing 4 Billion Things A Day)

July 6th, 2011 07:30 admin View Comments

Today at a special event at its headquarters in Palo Alto, California, Facebook CEO Mark Zuckerberg has taken the stage to unveil some key new announcements (you can find our live notes and a live stream right here). One of the first things Zuckerberg announced: Facebook has observed the the rate that its users are sharing is increasing at an exponential rate.

He explains that if you look at any Facebook user, on average, the amount they share today is twice what they shared a year ago — and in one year they’ll probably be sharing twice as much as they are today. It’s an important trend, and it’s one that Zuckerberg says probably applies to the internet at large rather than just Facebook. But a lot of it is happening on the social network — users are now sharing 4 billion ‘things’ on Facebook every day.

So why is this important? If you look at the charts below, you can see we’re right at the elbow of the growth curve. In other words, sharing may be about to explode. The question now is what people will be sharing. You can see in the graph below that Facebook product launches have helped drive additional sharing, and the social network expects additional innovations from other companies involved in everything from music to communication to help drive this explosion of growth.

Source: Zuckerberg: Online Sharing Is Growing At An Exponential Rate (And Users Are Sharing 4 Billion Things A Day)

Instagram Now Adding 130,000 Users Per Week: An Analysis

March 10th, 2011 03:00 admin View Comments

Editor’s note: The following analysis is written by Robert J. Moore, the CEO and co-founder of RJMetrics, an on-demand database analytics and business intelligence startup. Robert blogs at The Metric System and can be followed on Twitter at @RJMetrics.

Since launching just six months ago, Instagram has quickly become one of the web’s top photo sharing services. The company recently passed the two million user mark and announced the launch of their API.

We sampled data from that API and loaded it into an RJMetrics Dashboard for analysis. This yielded detailed new insights into Instagram’s growth and user adoption.

Please note that these are estimates extracted from a sample set, so there may be a small margin of error (more details on our methods can be found at the end of the article). I provide a full breakdown below, but here are some highlights:

  • Instagram is currently adding 130,000 registered users per week
  • Instagram’s 2.2 million users upload 3.6 million new photos per week (or 6 photos per second)
  • 37.5 percent of registered users have never uploaded a photo
  • 5 percent of users have uploaded over 50 photos
  • 65 percent follow nobody, and only 12 percent of users follow more than 10 people
  • For users who upload at least one photo, there is a 45 percent chance they will upload a photo the following week
  • Those same users have a 25 percent chance they will upload a photo 12 weeks later, representing a significantly stronger retention rate than Twitter.

Users and Uploads

The chart above shows weekly new user additions since Instagram’s launch. The company is registering approximately 130,000 new users per week, and that number hasn’t fluctuated much since the new year.

When we view this chart on a cumulative basis, we can see overall user growth curve looks more linear than exponential (although its slope is quite impressive).

If it maintains its current pace, Instagram will continue to add a million new registered users every two months. Things get more interesting when you look at what these users are doing once they join.


Unlike with registered users, the number of new photos uploaded on a weekly basis is clearly climbing, with 2.6 million new images uploaded in the most recent week.

This is a strong indicator that users continue to use the service long after joining. As thousands of new users join each day, the millions that have joined in the past continue to use the service, creating an increasing flow of photos into the system.

When we view this chart on a cumulative basis we can see this exponential effect in action:

It doesn’t get much nicer than that. Despite this impressive trend, however, not all of Instagram’s 2.2 Million registered users are actively using the service.

Below, we show a distribution of users based on how many photos they have uploaded.

As you can see above, 37% of Instagram’s users have never uploaded a single photo and 65% have uploaded fewer than three photos.

Our sample population indicated that about 35% of Instagram’s registered users uploaded a photo last month. This is down from previous months, which is typical in the early months of a new product’s life as the growth/churn ratio stabilizes. This number is clearly still in decline, and where it ends up will be a critical metric for Instagram. Based on its current base of 2.2 million users, this trend implies that around 800,000 Instagram users uploaded a photo last month.

This glut of inactive users is common in free online services, as we saw in our studies of Twitter Data and Foursquare Data last year. This isn’t necessarily a lack of total engagement, as users can use the app to simply follow their friends’ photo streams.

The distribution of followers (how many users a given user is following) is shown below.

Interestingly, over half of Instagram’s users are following exactly one other user, with another 13% not following anyone. We checked into that, and it looks like the vast majority of users who follow only one other user are following the “Instagram Team” account, which was likely automatically added to their list at signup.

This means that 65% of users effectively follow no one.

Only 12% of users are following more than 10 people. Note that Instagram takes you through a process to find friends and other people to follow upon any new installation, explaining why more users are following someone than have uploaded a photo.

Repeat Upload Rate

We know that only 63% of users ever upload their first photo. We were curious about what happens next, so we used the distribution of photo uploads to approximate a “Probability of Incremental Upload” based on how many photos have been uploaded so far.

As you can see, every time a user uploads photo they become increasingly likely to upload another. 76% of those users who upload their first photo go on to upload a second, 85% of those users go on to upload a third, and so on.

Time Between Uploads

The average time between any two uploads made by a single user is 43 hours (obviously users who have uploaded one or fewer photos are excluded from this number). If we map out the average “Time Since Previous Upload” by upload number, we see another impressive engagement trend.

With each new photo uploaded by a user, the expected time until they upload their next photo decreases.

(Note that Instagram’s short history could make this chart a bit of a self-fulfilling prophecy. Since most users are less than a few months old, users who upload more frequently are obviously the only ones who have been capable of racking up those big upload counts.)

Cohort Analysis

Cohort Analysis is a powerful tool that we use to track user engagement and loyalty over time.

Each “cohort” is a group of customers who joined at a similar time, and we use cohort analysis to compare how different cohorts behave over the same relative periods in their lifecycles. User cohorts are typically grouped by month, but since Instagram has such a short history we used the Weekly Cohort feature in RJMetrics instead.

In the chart below, each line is a cohort of users who uploaded their first photos in a distinct week of Instagram’s history. The chart shows the percent of users in that cohort who came back to upload another photo in each subsequent week of their lifecycle. Note that, by definition, this excludes users who never uploaded a photo.

The decay we see here is typical in a cohort analysis. As each week goes by, a smaller percentage of the cohort’s users come back to upload again. Eventually, the lines stabilize, indicating the percentage of users in that cohort that have become “hooked” and are likely to stick around for quite some time.

These cohorts all appear to behave similarly. Typically, 45% of users are still uploading photos in Week 2, but that number drops down to around 25% by Week 12.

This may sound like a low number, but in the universe of free applications it’s actually quite impressive.

By way of context, Twitter’s weekly cohort numbers were notably lower at a similar point in their history. Instagram’s challenge will be retaining that core group of users and trying to push these cohort lines upward by luring their existing users back as time goes on.

Another interesting cohort perspective is to view the cumulative average number of photos uploaded per cohort member over time. This shows how each cohort contributes to the universe of photos during its lifecycle.

You can see a slight decay in the slope of these cohort lines, but they are pretty close to linear. In other words, despite a decreasing number of users uploading photos (as shown in the previous chart), the total number of photos uploaded by the cohort remains strong as time passes.

This suggests that users who continue to upload are uploading photos at a faster rate as time goes on. This is supported by the “Repeat Upload Rate” and “Time Between Uploads” charts discussed above.

Finally, it’s noteworthy that the “Week of November 7th” cohort is the poorest performer in both of these cohort analyses. If you look back at the user growth charts, you’ll see that this cohort was created during Instagram’s peak user growth period (its initial press wave upon launching).

This is evidence that users acquired during such press flashes are, on average, more likely to churn than new users who came in through other channels. We’ve seen this trend in countless other businesses.

How We Did It

Four things allowed us to conduct this analysis: Instagram’s new API, their use of incrementing ID numbers to represent users and photos, the central limit theorem, and the tools in the RJMetrics hosted business intelligence dashboard.

The central limit theorem tells us, among other things, that a large enough random subset of a large data set will behave like its parent set with a high degree of statistical confidence. Since the API allows us to retrieve content by ID number, we conducted a random sample of the data and extrapolated its properties to draw our insights. Our sample consisted of 9,200 users and their 116,000 photos.

Since user registration dates were not directly available via the API, we used the “first upload date” of a user (and its users with surrounding IDs) to approximate the join date of any given user.

This works on the assumption that at least some percentage of users will upload a photo within a very short time of joining. These earliest “first upload” dates within small ID ranges were applied to deduce approximate join dates, which we estimate to be accurate within a day at most.

We also explored the data to ensure that we weren’t overestimating due to “dead IDs” from deleted accounts and photos (which turned out to be a negligible percentage). Finally, we made sure to properly account for users and photos marked “private” that were not accessible via the API (although info from those accounts is not influencing account-specific insights such as cohort analysis).

Source: Instagram Now Adding 130,000 Users Per Week: An Analysis

Film Site MUBI Secures $2.4 Million To Socialise Cinema Buffs

January 7th, 2011 01:13 admin View Comments

MUBI, the online cinema platform, says it’s hit a significant growth curve recently, crossing one million registered users. It’s also raised $2.4 million from Angels to strengthen its balance sheet and grow the team in a round round led by Eduardo Costantini and which includes existing investors, Martin Varsavsky (founder of Fon), and Alec Oxenford.

Read the rest of this entry »

Source: Film Site MUBI Secures $2.4 Million To Socialise Cinema Buffs

StumbleUpon Sent 700M Pageviews To Other Websites In Dec, Is Growing 20% Monthly

January 4th, 2011 01:59 admin View Comments

Social discovery service StumbleUpon made headlines yesterday when CEO Garrett Camp tweeted out that it had surpassed Facebook in terms of referral traffic on StatCounter and was now responsible for 43% of all social media traffic on the 15 billion pageviews that the analytics service tracks (versus Facebook’s 38%).

When a company with 13 million users is outpacing one with 500 million on one specific metrics service. additional context is needed to understand exactly what is going on behind the recent growth curve.

I spoke with StumbleUpon CEO Garrett Camp earlier today about what exactly this recent news means for the company and why StumbleUpon seems to be killing it since August, clocking in about 1.9 million unique visits in November according to Comscore and serving up over 700 million “Stumbles” (pageviews to other websites) in December 2010 versus 400 million Stumbles in December of 2009, according to internal analytics.

Camp says that the service has been growing at 20% month over month since it introduced StumbleUpon for the iPhone and Android in August and an iPad app in April. Mobile engagement, which represents 5% of all Stumbles, is growing at 40%. Web-only Stumbling, also a new addition to the vast oeuvre of browser plugins and toolbars, is growing at a 25% rate. The site also gains around 500-600K new registered users monthly on average and boasts a 40% DAU rate.

“We’re becoming a little more useful” says Camp on the recent traffic spikes, “Discovery as opposed to keyword search is just a better way to find information.” Camp also explains the discrepancy between Facebook and StumbleUpon’s usership vs. referral traffic as such, “Facebook is by design trying to keep you on Facebook, but with us every time you click we’re sending referrals to other sites,” he continues, “like Google our goal is to take you to a new site.”

Says VP of Business Development Marc Leibowitz:

“The growth curve has also steepened recently because of the introduction of more overtly social features (eg, finding & inviting real-world friends) to complement the less obvious but long-standing social aspects of StumbleUpon (eg, algorithmically determined like-minded users); the improving usability of the service (eg, “show me more from [this domain, topic, user]“; better onboarding); and probably the overall proliferation of content, which has increased the demand for easy-to-use, adaptive curation tools like StumbleUpon (see also Google’s experiments w News & Reader).”

The site is also breaking even according to Camp, and at around 65 employees that means revenue is in the ballbark of 10 million annually (When asked if this guess was accurate, Leibowitz told me “You’re not bad at math.” This is not true.). In any case, beating Facebook at something is a hell of a way to start 2011 strong. Congrats guys.


Source: StumbleUpon Sent 700M Pageviews To Other Websites In Dec, Is Growing 20% Monthly

Don’t Tell Tesla: In 2020, Sales Of Electric Vehicles Will Be Less Than 8% Of The Global Market

October 27th, 2010 10:49 admin View Comments

Given all the apparent excitement surrounding electric vehicles and the billions invested by automakers, financial firms and government incentives, you would assume there was a groundswell of consumer support for hybrid and battery electric cars underlying this momentum.

Unfortunately, according to a J.D. Power and Associates report released this morning, we may be on the electric car bandwagon but once you get us to the dealership, we look the other way.

The firm predicts that in 2010, international sales of hybrid and battery electric cars will total 5.2 million, or 7.3%, of the 70-plus million consumer vehicles expected to be sold. This year, the world is on track to sell 954,500 electric units, 2.2% of all car sales. Thus, in a decade, consumers will only move the needle about 5 percentage points, from 2.2 to 7.3. If the future of the car is electric, it’s not happening in the next decade and— given the sluggish growth curve— not the one after that either.

So why the apprehensiveness among consumers?

According to J.D. Power and Associates, a confluence of factors have discouraged would-be-electric owners. Not surprisingly, the number one reason is money. Although there are government subsidies to encourage green car purchases in the US, it’s still difficult for cash-strapped buyers not to compare the new hybrid models against cheaper, less fuel efficient vehicles (remember, the recession is not over for the 9-plus percent who are still unemployed or the many who are underemployed).

“Many consumers say they are concerned about the environment, but when they find out how much a green vehicle is going to cost, their altruistic inclination declines considerably,” John Humphrey, an SVP at J.D. Power, said in a statement. “For example, among consumers in the U.S. who initially say they are interested in buying a hybrid vehicle, the number declines by some 50 percent when they learn of the extra $5,000, on average, it would cost to acquire the vehicle.”

Beyond sticker shock, the report says other popular reasons included: aesthetics, insecurity in the new technology, unhappiness with how the car performed, and the need to recharge batteries/driving range.

Who is actually buying?

J.D. Power says from their research, a very clear, dominant demographic emerged: “buyers of HEVs and BEVs are generally older, more highly educated (possessing a postgraduate degree), high-income individuals who have a deep interest in technology, or who like to be among the early adopters of any new technology product.”

At least there’s hope for those Tesla dealerships in Silicon Valley.

Source: Don’t Tell Tesla: In 2020, Sales Of Electric Vehicles Will Be Less Than 8% Of The Global Market

Even If Solar Grows 30X, It Will Only Be 4% Of America’s Power Capacity

October 25th, 2010 10:33 admin View Comments

For all the talk about solar, the US market for solar power still has a long way to go before it makes a real dent in the country’s overall power capacity.

On Monday morning, Bloomberg New Energy Finance, a major aggregator of green industry data released a few key projections: the US solar market is on track to grow 30x to 44 gigawatts by 2020 and could make up 4.3% of America’s total power capacity. Of this fraction, the bulk (or approximately 68%) will be in photovoltaics with solar thermal making up the remainder.  Furthermore, Bloomberg predicts that consumer traction will also move noticeably higher, with 2.4% of homes solar-equipped by 2020.

Of course, that path to 4.3% of national power capacity is not cheap. In order to get there, the US market will need to attract $100 billion in investment dollars.

The possible leap from 1.4GW to 44GW is an impressive growth curve, but these figures certainly highlight the simultaneous growth and challenge of solar installation.

As Bloomberg New Energy Finance points out, the surge in solar capacity has been supported by two crucial trends: the drop in prices (the price of photovoltaic modules has tumbled from roughly $300 per watt in the mid-20th century, to less than $5 per watt today) and the heavy hand of government support. And yet, even as solar becomes more affordable, it’s still playing catch up to other sources of energy. That may be a well known fact but it often gets muffled in the bucolic vision for solar panel farms as far as the eyes can see.

“The group’s latest analysis places the unsubsidized cost of best-in-class photovoltaic and solar thermal electrivity generation at just below $200/megawatt hour— nearly four times the equivalent cost for a coal-fired power plant ($56/megawatt hour)— and between two and four times the cost of onshore wind power, ” according to the Bloomberg report.

On the investment front, it will be interesting to see how the solar industry fares in the money race. Overall, the sector has been a major beacon for investment dollars but momentum has recently waned. According to a Mercom Capital report for the third quarer,  VC funding for the solar sector was $169.35 million on 11 deals, versus $922 million for 18 transactions in the prior quarter. On the flip side, there was strength in other funding sources (including credit lines from banks), which totaled a healthy $20.7 billion for the quarter.

(Image: Flickr/Warm N’ Fuzzy)

Source: Even If Solar Grows 30X, It Will Only Be 4% Of America’s Power Capacity