5 ms·
This may have been pointed out elsewhere, but the questions surrounding the % of users that are bots/spam accounts was addressed by Twitter about 3 years ago. I
by natroniks 4y ago
This may have been pointed out elsewhere, but the questions surrounding the % of users that are bots/spam accounts was addressed by Twitter about 3 years ago. In early 2019, Twitter stopped reporting overall user numbers and instead offered the alternative metric "monetizable daily active users." Without saying it outright, the difference between dau and mdau is bots/spam accounts. Many articles discussed this massive change, which happened THREE YEARS AGO. Here is one such article:
https://www.socialmediatoday.com/news/twitter-reports-another-decline-in-users-shifts-to-alternate-performance-m/547981/ https://www.socialmediatoday.com/news/twitter-reports-anothe...
Q1 2019 was the only quarterly report I can find that included both total users and monetizable users. If we take the difference as representing bots etc, then we can see in the linked article that about 60% of overall Twitter users were not monetizable, ie, 60% of Twitter accounts were bot/spam accounts. So, this is all in the public record.
Doubting how much of Twitter's overall accounts are bots - which is what Elon is pursuing - is massively different from Twitter standing by its claim that only 5% of MDAUs are bots. Elon seems to be talking about DAUs and Twitter only reports MDAUs. This distinction seems to be lost on most people following this debate. I think both parties could be offering accurate assessments: Elon could be right that over half of Twitter's users are fake, and Twitter could be right that only 5% of the MDAUs are fake. While both could be right in their assertions, Elon is in the wrong in the big picture. All of this has been disclosed publicly and Twitter never claims that only 5% of users are bots, just 5% of monetizable users. Elon's claims of malfeasance or obfuscation are wrong.
- mlindner 4y agoThis is an interesting take I hadn't heard before. Thank you. I wonder why this hasn't been getting communicated by Twitter to Elon though (he likely has not have realized that this change occurred).
- natroniks 4y agoI too don't know why Twitter or the press doesn't make this metric more apparent. But I will say, if Musk didn't know about this change, then he already is what he's long aspired to be: a space cadet.
- kcplate 4y agoBecause it’s not necessarily in their best interest. Everything is a negotiation tactic on both sides. Does Twitter want Musk to buy them? Maybe, or maybe not. Less than 2 months ago it was actively trying to prevent it. Now it’s trying to force it to happen? Four months ago, Musk wanted to destroy Twitter by starting a competitor. Two months ago he wanted to buy it, now he wants to back out of the deal. Seems to me a dance is taking place.
- SilverBirch 4y agoIt has been communicated to Elon by Twitter. Repeatedly. It's in the SEC filings, it's in their annual reports. Here's a really long twitter thread by the CEO explicitly pointing out that they are measuring spam accounts as % of MDAUs - literally putting that in asterisks for emphasis: https://twitter.com/paraga/status/1526237578843672576 https://twitter.com/paraga/status/1526237578843672576 Here is Elon Musk responding to that thread with a poop emoji: https://twitter.com/elonmusk/status/1526246899606601730 https://twitter.com/elonmusk/status/1526246899606601730 The problem isn't that no one has bridged this divide, it's that Elon Musk is acting in bad faith to derail the deal that he fucked up by signing. There's no point talking about spam accounts because that's literally not the issue. The issue is that Musk signed a deal valuing Twitter at a premium at the peak of the market with no real escape clauses, and since there are no legitimate ways for him to renegotiate he needs to come up with a pretense.
- ModernMech 4y agoIt’s amazing to me that one party here can literally sling shit and still be taken as a serious, rational, adult person by anyone. A poop emoji is a clearer way to say “I’m acting in bad faith” than literally saying “I’m acting in bad faith”. Money has amazing perception bending properties.
- tlb 4y agoYes, but I think it's reasonable to demand evidence that the mDAU numbers really do include only 5% bots. Laziness and wrong incentives could easily combine to let that percentage creep up over the years. It wouldn't be the first company this happened to. It's hard to have 100% confidence in Twitter's ability to classify bots accurately for mDAU purposes, given their terrible accuracy for anti-spam purposes. It's a slightly different problem (because ads can be counted after the fact while anti-spam has to be immediate) but it raises enough doubt about the mDAU numbers that I'd want to see more data.
- natroniks 4y agoIt is prudent to check all facts when negotiating an acquisition. The time to do this was before agreeing to the acquisition. I would also argue the way he's doing it is wrong, but I'm not a hardball negotiator so maybe publicly venting is right. I think it is important that we all distinguish between DAU and MDAU when debating what the likely number is. Frankly, I do not understand why it matters. Musk bought the company based on the idea that it is undervalued: 1) he wants to move away from ads and 2) he wants to grow the user base. If it turns out there are fewer "real" users on the platform than previously thought, isn't that bullish for the prospect of adding more users? I find Musk's tactics to be abhorrent, but I'm looking beyond tactics at the core argument: I do not think it's clearly bad if it turned out Twitter had fewer than expected users when his case is that he can add more users. So long as they didn't fudge the money numbers - earnings and revenue - I don't see how this metric is considered material enough to break the deal.
- vehementi 4y agoWhat he's saying today is distinct from the MAE implication, he's saying that he has the right to this info now (not that it is part of due diligence). He quotes some parts of the original merger agreement and claims his right to this data falls under that.
- tlb 4y agoPublic companies are limited in what data they can provide to potential buyers. In general, they can't provide any more than what's in their public statements. The SEC enforces this rule to ensure that all investors have access to the same info. Also, when buying public companies, you have to be extremely careful not to let word get out before it's publicly announced. It's easy to imagine a Twitter employee tasked with sending logs to E.M. Acquisition Corp leaking this particular bombshell to the press. So it's normal to make an acquisition bid based only on their publicly reported numbers. The acquisition contract provides for a due-diligence phase that happens between when the deal is announced and when it's final (ie, now) where the buyer can verify that there aren't material misrepresentations in the public filings. If there are, they can back out. Everything about this process is normal, except the hyperventilating press.
- vehementi 4y agoFor a little bit of nuance, those non-m users would consist of: - bots, as you say - "legit" bots e.g. customer service - inactive accounts - accounts that use some alternate client (tweetdeck or whatever) - API users - possibly web accounts using ad block (depending on whether twitter measures literal ad views vs. "users logging in with a client that could show ads" - ... > Elon could be right that over half of Twitter's users are fake, and Twitter could be right that only 5% of the MDAUs are fake. While both could be right in their assertions, Elon is in the wrong in the big picture. All of this has been disclosed publicly and Twitter never claims that only 5% of users are bots, just 5% of monetizable users. Elon's claims of malfeasance or obfuscation are wrong. That's exactly my conclusion as well
- natroniks 4y agoThanks for contributing to my comment with further information. I find it odd (in an unconvincing sort of way) that Musk's only complaint has to do with a company-defined metric, as opposed to something like "revenue" or "cash flow" that is standardized. Twitter can define mdau's any way they choose such that, if they wanted, they could draw up the definition in order to end up with a 5% number.