Adrianus Wagemakers is an independent financial consultant in the areas of investment management, financial planning, and data analysis.
When Twitter announced the results for the second quarter 2015, investors were initially happy. Revenue was $US502 million, which was higher than the previously forecasted range of $US470 million to $US485 million. Monthly active users were 316 million, which was up 15% year over year, compared to 308 million in the first quarter of 2015.
The mood turned quickly during the conference call, following the publication of the results, and during the conference call the share price of Twitter dropped significantly in the aftermarket. One of the first remarks by Jack Dorsey, co-founder and interim CEO set the scene for a negative mood:
“…product initiatives we’ve mentioned in previous earnings calls like instant timelines and logged out experiences have not yet had meaningful impact on growing our audience or participation. This is unacceptable and we’re not happy about it.”
In his prepared statements, Anthony Noto, Twitter’s CFO noted that growth in monthly active users (MAUs) did not benefit from the positive factors that benefited the first quarter of 2015. MAU’s grew from 308 million to 316 million, but this was mainly caused by a group of users, the so called SMS fast followers, that Twitter introduced as part of MAUs last quarter. Excluding this group of users MAUs increased only from 302 million to 304 million.
Twitter needs both strong growth in users and increasing revenue per user even to justify the kind of valuation we are seeing today after the effects of the second quarter earnings have been digested by investors. In this post we will take a closer look at MAUs, and the main conclusion is that looking at MAUs and the growth rate of MAUs is not enough. There is a possibility that the quality of MAUs is deteriorating, which will immediately impact the effectiveness of monetisation efforts, and the potential revenue per user.
There is also a risk that MAUs will start to decline, because of the fact that relatively “good” MAUs are being replaced by relatively “bad” MAUs. It is understandable that Twitter is putting great effort in further monetisation, but it may be even more important that the user experience improves, so the massive pool of existing, but inactive accounts can be re-engaged.
The importance of MAU is undeniable. Investors use MAU as an indicator for growth and potential size of the platform. Understanding MAU is not so easy, because there have been some surprising elements in MAU that are difficult to understand, and therefore also difficult to forecast. Twitter defines MAU in the following way:
“We define MAUs as Twitter users who logged in and accessed Twitter through our website, mobile website, desktop or mobile applications, SMS or registered third-party applications or websites in the 30-day period ending on the date of measurement. Average MAUs for a period represent the average of the MAUs at the end of each month during the period.”
In the S-1 statement that was published before Twitter went public it was stated that less than 5% of MAUs are spam accounts, and this is still communicated today. Twitter also made an important remark about automatic MAUs in the same S-1 statement:
“Our metrics are also affected by applications that automatically contact our servers for regular updates with no user action involved, and this activity can cause our system to count the users associated with such applications as active users on the day or days such contact occurs. In the three months ended June 30, 2013, approximately seven per cent of all active users used applications that have the capability to automatically contact our servers for regular updates.”
And in the same paragraph:
“We expect that the percentage of active users that use applications that have the capability to automatically contact our servers for regular updates will decline over time, particularly as usage of our mobile applications increases.”
This is all fine. We learn that there is automated activity that may not have been initiated by a user, but is still counted as an MAU. The percentage is 7% and Twitter expects it to go down. In a revised S-1 statement Twitter revised the percentage to 8% for the three months ending September 30, 2013.
In the following quarters we learn from the 10-k statements that this percentage has increased to 11% at the end of the fourth quarter, and to 12% at the end of the first quarter of 2014. On the presentation slides that were made available by Twitter for the second quarter of 2014, the percentage increased further to 14%. This increase to 14% was also briefly discussed in the conference call for the second quarter earnings. However, based on initial information by Twitter the expectation since the IPO was that this percentage would go down. For the MAU numbers this implied the following:
If we compare MAUs for Q2-2013 versus MAU’s for Q2-2014, the numbers show that from the increase of 53 million almost 23 million, or more than 40% was attributable to an increase in the percentage of automated accounts. It is very likely that this caused the quality of MAU’s, in terms of potential accounts to monetise, to decrease. After the second quarter presentations and earnings call, it was surprising to learn that when the 10-Q statement was released that Twitter had adapted the numbers:
“We have reviewed and refined our processes, however, to calculate a new metric that is comprised of only such active users who have used applications with the capability to automatically contact our servers for regular updates where there was no discernable user action involved. In the three months ended June 30, 2014, approximately 11% of all active users solely used third-party applications to access Twitter. However, only up to approximately 8.5% of all active users used third party applications that may have automatically contacted our servers for regular updates without any discernable additional user-initiated action.”
The refinement of the processes by Twitter apparently did not lead to an adjustment of reported MAU of 271 million for the second quarter of 2014. Since this refinement of the internal processes, no changes have been reported anymore to the percentages.
For the first quarter of 2015 Twitter announced a new group of users to be included in the MAU number, the so-called SMS fast followers. These are users who sign-up and access Twitter solely via SMS. Twitter has stated that these users are important, and were not previously included in the MAU numbers. In the first quarter of this year, 6 million MAUs were added to account for the SMS fast followers, giving a total of 308 million for the quarter. For the second quarter of 2015, Twitter reported 316 million MAUs, an increase of 8 million MAUs. 6 million of these new MAUs came from SMS fast followers.
This means that in just one quarter SMS fast followers increased with the same number of users as in all the previous quarters combined. Twitter explained that the monetisation rate of SMS Fast followers is meaningfully lower than the monetisation rate of other users, so from a monetisation perspective MAUs may be increasing, but potential revenue per user is decreasing, and this effect is accelerating because of the increased number of SMS fast followers as part of MAUs
Total existing accounts
In a recent blogpost Chris Sacca (@sacca), a large and well informed investor in Twitter, mentioned that almost 1 billion users have tried Twitter but did not stick around. Together with the 316 million active users at the end of the second quarter of 2015, this means there must be about 1.3 billion Twitter existing accounts currently. More accounts must have been registered, but have disappeared either through account deletions or suspensions by Twitter. It is not so surprising that there are so many existing accounts.
We should not forget that Twitter is already approaching its 10th anniversary, although it only has been a public company for less than 2 years. Many of the accounts that are the most popular today were already registered more than 5 years ago, and many people have tried Twitter already during its existence.
At the end of June last year Twitter reported 274 million MAUs (including SMS fast followers), while total existing accounts at that time were estimated between 1 and 1.1 billion. Based on this estimate, Twitter added 42 million MAUs over a 12 months period, while the total number of accounts seems to have increased with more than 20 million accounts per month.
Since every new account is a logged in user, even if it is just for the time it takes to register, it seems a logical conclusion that the steady stream of new accounts means that from the total MAU number, always about 20-25 million are made up from the most recent registrations. It also seems logical to conclude that if the only noticeable action these new accounts performed was the registration, they will cease to be counted as an MAU after 30 days.
If there is a constant flow of new accounts signing up with Twitter, but MAUs do not reflect this, it must mean that there is substantial churn taking place. For every 100 new accounts becoming an active user, there must be about 80-85 accounts becoming inactive. Most likely many of these accounts are accounts that just signed up and just did not become active, but there must also be churn from users who signed up in the more distant past, users that were active for an extended period of time before they became inactive. With a significant number of new active accounts signing up, and a significant number of accounts becoming inactive, net growth of active accounts can be volatile.
There is a risk that the constant flow of new accounts dries up, or that the quality of these accounts deteriorates, leading to higher churn, and potentially negative growth. A very important key to further growth for Twitter therefore must be in user retention. The churn has to be decreased, or better, a significant number of the 1 billion inactive accounts have to be re-activated. For these users, the user experience has to be improved. Initiatives by Twitter from the recent past do not seem to have made a big impact. Hopefully upcoming improvements will be convincing enough for a substantial number of inactive accounts to return.
MAU’s going forward
An important reason why the Twitter share price dropped during the conference call may have been the following remarks by Mr. Noto:
“To be clear, however, we do not expect to see sustained meaningful growth in MAUs until we start to reach the mass market. We expect that will take a considerable period of time.“
The remark about reaching the mass market seems very weak given the fact that 1 billion users have tried Twitter already, but did not really start or continue to use the platform. Also later in the conference call in answering a question:
“But in the near term, our organic growth is going to be very low as it was this quarter and as I think about Q3 it’s marginally better, but I wouldn’t want you to or anyone else to expect a change in our growth rate relative to what you are seeing in this quarter. I think you’ll see that for a while…“
So clearly Twitter is pointing at a continued period of very modest growth in MAU’s. Without compelling reasons for new users to stick around or old users to return, it was explained earlier that this could well mean that MAUs start to turn negative, and certainly that the quality of MAUs is deteriorating.
In the conference call Twitter also gave metrics about daily active users (DAUs) versus MAUs. In the top 20 markets in the second quarter of 2015 was approximately 44% versus 48% for the first three quarters of 2014. This implies DAUs were approximately 139 million. For the third quarter of 2014, Twitter reported 287 million MAUs, including SMS fast followers.
This implies there were approximately 138 million DAUs at that time, and therefore DAUs have only grown marginally with 1 million. It can be argued that DAUs are more engaged users, and more valuable from a monetisation perspective. A drop in the percentage of DAUs versus total MAUs therefore would lead to a lower quality of MAU’s from a monetisation perspective.
Another reason for some concern is the lack of growth of tweets per day. Currently Twitter shows on its website that 500 million tweets are sent per day. This is the same number that was reported in the S-1 statement that was filed in October 2013. It is also the same number that was reported by Dick Costolo in October 2012, a year earlier, when he spoke to an audience at the IAB engage in London. In three years, where Twitter MAUs grew from less than 200 million to more than 300 million, average tweets per day did not increase. Maybe this is because tweets are moving to private group direct message conversations, and if this is the case, it would be helpful for investors if Twitter would start giving data about direct messages per day and if these are being monetized. In the conference call Mr. Noto made a remark that the number of direct messages sent accelerated.
It may also indicate that Twitter is gaining MAUs that are mostly consuming tweets, and losing MAUs that were creating tweets. If that trend continues, we may actually see the number of tweets starting to drop. It could mean that Twitter is becoming less social and more a broadcast platform, or indeed a real-time information network, as co-founder Evan Williams said on a conference in Aspen, Colorado. It could be argued that this doesn’t really matter. If more passive MAU’s consume fewer tweets the basis for monetisation can still grow. It does matter in case passive users are less engaged users than active users, and more difficult to monetise.
MAU is still one of the top metrics that investors look at when trying to determine the prospects of Twitter. Guidance by Twitter is pointing to weak MAU numbers for a considerable period of time. The initiatives like instant timelines and logged out experiences have not yet had a meaningful impact, so it is only logical that investors start to take a “seeing is believing” approach to Twitter.
If the situation is indeed that the increase of MAUs is the balance between 20+ million accounts that are registered every month, and churn from existing accounts, there is a substantial risk that, in absence of compelling improvements in the user experience, the balance between these two elements will become negative, and MAU will actually start decreasing. In any case it is very likely that the quality of MAU’s is not improving, and this will have an impact on the potential revenue per account.
Twitter needs to re-accelerate growth and it has to come from re-engagement from the 1 billion accounts that are inactive at the moment, and have been disappointed by what Twitter has to offer. Hopefully Chris Sacca was right when he wrote a blog post last month called “What Twitter Says, and What Investors Hear” where he interpreted statements by Dick Costolo and Jack Dorsey. With respect to strategy and product direction he believes they meant to say the following:
“We are nearly ready to launch the products that we are confident will re-accelerate our user growth. We have been working on this stuff for months, ever since Kevin Weil took the product helm. You have seen that our current team has been launching new products at a dramatically faster pace than ever before. These releases, and those to come very shortly, are all part of a comprehensive strategy that we know will attract and retain hundreds of millions of new users. Why would we change that strategy now?”
Hundreds of millions of new users sounds very ambitious, and probably he forgot for a moment that re-engagement of existing accounts is equally, or maybe even more, important to the growth of Twitter. The task seems enormous, and maybe Twitter has been too ambitious in believing they can do everything themselves. It could be the reason why co-founder Evan Williams wants developers to come back, because that would be really good news for Twitter users.