Tren’s Advice for Twitter

Jack Dorsey very recently asked for feedback on Twitter. The focus of nearly all of the comments he received was on ways to make Twitter’s service qualitatively better for consumers. If Twitter does not have a great product for consumers, nothing else matters. But having a great product for just one side of a three-sided market is not enough to make Twitter into a successful business.

In a deeper analysis below I will argue that Twitter needs to become more of a platform and not less. Twitter’s drive to become a media company reminds me of this joke:

Late one night a police officer found a drunk man crawling around on his hands and knees under a streetlight. The drunk told the police officer that he was looking for his keys. When the police officer asked if he was sure this is where he dropped his keys, the drunk man replied that he believed he dropped his keys across the street. The police officer asked: “Then why are you looking over here?” The drunk explained: “Because the light’s better here.”

Twitter devoting resources to offerings like Moments in an attempt to become a media business is like the drunk looking for his keys under the streetlight, when his keys are across the street where the building blocks of a much better platform business can be found. In short, it is easier for Twitter to try to be like Yahoo than it is to be a better platform. It is also easier to open a bakery or bar than it is to maintain and expand a platform. But because it is easy to do does not mean it is the right thing for Twitter to do.

I will argue below that to be a better platform Twitter must let go of its worship of the monthly average user (MAU) metric and go full speed ahead on ending abuse, bots and spam. MAU will drop for a while, but will emerge will be a more genuine MAU and better advertising targeting and user data.

In terms of suggestions to make the Twitter consumer offering better, why not focus on doing more things for existing Twitter creators? The best creators on Twitter get paid zip to make Twitter more valuable. They don’t get *any* love from Twitter whatsoever. The best thing about Twitter are the people who create Tweets. Twitter should do a far better job of making these people more discoverable and happy. For example, why don’t they reach out and offer a blue check mark to the most valuable creators on Twitter. If you ask to get the check mark and you are not Beyonce, Twitter more or less treats you like a possible criminal. The blue check mark rejection letter, which makes you feel like you have been rejected by Harvard Divinity school for questionable character, is a missed opportunity. YouTube has taken advantage of the opportunity to create a far more positive relationship with its creators. Instead of giving some attention and love to the people on Twitter who create the value for Twitter, Twitter is hiring more people to write first party content about the Kardashians on Moments? That. Is. Just. NUTS.

Blue check marks are just one idea for making the service more valuable for Twitter’s most important creators. Medium has this same problem, but it is even more important for them. In a Tweet his week about Medium I said: “Content can only be consistently good if its creators can make a living from it.” Twitter creators don’t need to make a living on the service, but they will do more work and contribute more if there are greater incentives to do so.

Twitter is a poster child for a critical point that many people do not understand about innovation. Many businesses like Twitter deliver huge amounts of value to consumers via innovation but do not generate a corresponding amount of profit for shareholders from that innovation. That Twitter is a very important innovation for society does not necessarily mean it will deliver a significant profit. That Twitter may have a moat from network effects is a different point than the size and value of the market that the moat protects. Moats are far from generic. Some moats protect very valuable markets and some don’t. Twitter’s objective should be to expand the value of what its moat protects. If it does not expand the value of what its moat protects, it will not grow into its cost structure.

More detailed analysis:

Twitter is a multi-sided platform linking advertisers on the first “side” of the market with people like me on the second side who use Twitter to communicate. Another third side of the platform is composed of businesses which pay for access to a Twitter API that gives them access to data that Twitter collects about interactions on the platform.

twit-plat

The key to delivering value to Twitter’s advertiser customers (Side 1) is having highly relevant data about potential buyers on Side 2 so the advertisers can generate a high and measurable return on investment (ROI) on their Twitter advertising. Advertisers on Side 1 will compare their Twitter ROI to the ROI they can achieve on other social media platforms. This of course is capitalism at work. Every advertiser and buyer of access to the Twitter API must do an opportunity cost analysis.

quart-rev

Twitter must capture a much bigger share of this:

msss

One important objective for Twitter is to make targeted advertising so valuable that when consumers want to buy something that the advertisements are considered by consumers to be valuable information rather than spam. Delivering advertising to consumers as close as possible to the time they are ready to buy is the optimal result for Twitter. How is Twitter doing in targeting is advertising? When you see an advertisement in your Twitter feed is it highly relevant to you? Have you ever clicked on a Twitter Advertisement? Or does a Twitter ad usually make you say: “this is annoying. Why are they sending this to me” The magic of search advertising is that when you are using a search engine you are often in the mood to buy. When you are on Twitter reading a post on a sporting event are you really in the mood to buy just anything? Is an advertisement for anti-itching cream valuable then?

To make the targeted advertisement valuable Twitter must “know you” in a deep way. What are the impediments to that? That should be the focus of work at Twitter.

Understanding the urgency of what Twitter must do is bought into focus by looking at the unit economics of Twitter. We may not have enough data to do the type of LTV calculation that I have written about recently on this blog, but Twitter does. We do know enough to make some educated guesses about Twitter’s Unit Economics. As always, especially for an investor, it is better to be roughly right than precisely wrong. Of course, Twitter can be far more than roughly right. They have the nonpublic data. They can do the math and use the math to create key performance indicators.

Most people who look at Twitter’s numbers focus on the need for more user growth. The cry from analysts is usually some form of: “Twitter must increase MAU and DAU!” This results in thinking and products like the awful Twitter Moments, which is essentially trying to recreate Yahoo’s declining business model and diverts resources that can be better deployed elsewhere. Is it a good thing is MAU and DAU grow? Sure. But the more important question is: what is the highest ROI for Twitter on new investments?

ugs

Over-fixation on growing MAU and DAU also makes Twitter fearful of fixing important but broken aspects of the service like abuse, spam and bots since that would result in user count drops and less “activity.” Unfortunately, that confuses genuine real consumers with counts inflated by fake users. Twitter needs to fix the abuse problem regardless of what it does to negatively impact fake MAU. Spammers and bots must go. No one is going to be fooled about MAU claims when the company can’t generate a profit. The clock is ticking.

Some people argue that this focus on growth MAU makes the Twitter usage data less useful too both in targeting advertising and for buying of access to the Twitter API. Trip Chowdry of Global Equities Research argues:

“If data quality is bad, Ad targeting is bad, and if Ad targeting is bad, Advertisers are not happy, and hence monetization will remain challenging for TWTR. If data quality is bad, then performing prediction on these data sets will also be wrong, hence 10% of revenues that TWTR gets from selling its data will also suffer.”

In my post on Chamath Palihapitiya he makes clear that the for key Facebook in the early days wasn’t MAU and DAU.

“After all the testing, all the iterating, all of this stuff, you know what the single biggest thing we realized [at Facebook]? Get any individual to seven friends in 10 days. That was it. You want a keystone? That was our keystone. There’s not much more complexity than that. It’s not just top-line growth. It’s acquisition, engagement, ongoing product value. It’s understanding the core value and convincing people that may not want to use it.”

What’s the core value of Twitter? What key performance indicator (KPI) drives that value best? It isn’t activity generated by abuse, bots and spam. I suspect the right KPI is something that captures real users sharing and creating content. Something like more than X Tweets shared or created over period of time Y.

Anil Dash in a thoughtful post on Medium argues that Twitter has made a mistake by focusing on the wrong metrics:

“Your relationship with Wall Street investors (and, to some degree, with advertisers) is fundamentally broken because you’ve gotten trapped into using the wrong metrics to measure the success or progress of Twitter. New signups are flat, and they’re going to stay flat, and every desperate flailing attempt to change that just reminds engaged users that they’re not seeing any progress and they don’t believe you can ship features they care about. Meanwhile, do you know how many new video creators joined YouTube this quarter? Me neither! You know why? Because all the good videos are on YouTube! What percentage of people who visit YouTube each month are logged in? What percentage ever uploaded a video? Answer: Nobody gives a shit. Because YouTube inarguably drives culture, and people (and advertisers!) want to be part of that.”

The more important tasks for Twitter and other platforms are less fun than writing Kanye entries in Moments. The focus of Twitter should be on increasing average revenue per user (ARPU) which is driven by the ability to better target users.

As I said, it is useful to look at unit economics

ARPU:

Twitter’s total most recent quarterly revenue: $616 million a quarter or $205 million a month

Monthly active users:

tmau

There are also ~500 million monthly average users (MAUs) who are not logged in, but let’s ignore them since it would make Twitter’s ARPU per user even worse.

Monthly ARPU: 317 million logged in users are generating $205 million a month or 64 cents per user per month.

Gross Margin (revenue less cost of goods sold or COGS):

tgm

The problem with the new “we are a media company” approach being adopted by Twitter is that even if they do grow users it is not a high gross margin path. Being a “media company” has significantly worse margins that a platform business and it scales far less well.

Churn (customers lost):

This variable and the next are the hardest to know if you are not Twitter. What is Twitter’s customer retention rate? Twitter knows but we must make an educated guess. We know net growth is 1-4% (including bots, etc.). Within that envelope there can be a lot of churn that is killing growth or not.

Survey Monkey has some rough data that should be viewed with caution.

smchurn

This chart looks at weekly churn, which refers to people who used the app one week, but didn’t use it again in the following week. Some of those churned users will probably return to the app in the future, but generally speaking, Twitter retention rates do not look very good at all, especially when compared to Facebook’s apps.

I’m going to make an educated guess Twitter’s real churn is 4% a month since that is typical in a consumer business with no annual contract. How did I make that guess? Well I know what their loss is as a company, I know gross margin and ARPU and I see their income statement which generally reflects sales and marketing costs. And I know many churn “comparables” from other companies over a period of decades. With that data, you use pattern recognition and make your educated guess.

CAC (customer acquisition cost):

CAC is also a hard one. In the case of a company like Twitter, a lot of the COGs is really CAC. I use the same reverse engineering process I used with churn to come up with my educated guess. Twitter, as I said, does not need to guess.

Looking at Twitter’s SG&A and the losses (see below) I, going to guess based on losses and SG&A that Twitter’s direct and indirect CAC is about $8.

So how does this net out? Well, Twitter’s unit economics might look roughly like this:

roughy

Unlike me Twitter does not need to guess what the inputs variables into the unit economics calculation are. But we do know that if Twitter can get ARPU up to $3 a month the operating leverage gets lots better. Adding a lot of new COGS to the math makes things worse.

There is some urgency if you look at the financials:

mcap-2