6/n These nos in some sense define how the company wants to measure its success. Uber cares less about DAU on their App, but rather wants to focus on rides, and now also eats.
Meta wants to talk about how vast they are,while Twitter wants to say they are an advertising business.
Conversation
7/n I also hear this notion that because Twitter takes out suspected SPAM from it's mDAU numbers, its disingenuous. This is straight up BS. Advertisers care about how many real people would see their ads and not how many users you may have. This is an advertiser facing metric and
1
1
8/n ... There are functionally equivalent metrics that other companies use. Eg Pinterest does a similar thing. It removes any SPAM accounts from its calculation of Active users. Surprisingly it seems Meta and Snapchat do not, at least based on my reading( I could be wrong)
1
1
1
9/n This is another example of how definitions of even basic "standard metrics" may vary. Every company will have some idea of how many users it is left with AFTER removing Spam and bots because those are the Only users you can monetise.
1
2
10/n There is also the notion that Twitter is not measuring it right. For eg, the sample size is too small. Not exactly. Getting statistically significant result is not as much dependent on sample size as it is on sample selection.. ..for eg..
1
2
11/n CNN did exit poll of US Presidential election by sampling just 16K people . Twitter samples 9000 accounts every quarter. A minimum 100 Sample size is generally minimum, and anything above 1000 maybe an overkill. It's a broad generalisation but useful
1
2
12/n Another objection is that it uses human raters to mark something spam, No AI / No ML.
Well DUH.
They already have used the AI / ML algos to remove the SPAM bots and now you need humans to do a final pass. A lot of companies rely on human evaluators. Here is what Meta says
1
2
13/n Another thing that some people freak out about is only 5% Spam? Well 5% of spam in mDAU is very different from 5% spam in DAU(See chart). To put into perspective, Meta claims that only 3% of its MAP's are "Violating accounts". Twitter could be lying, but 5% is not absurd
1
2
14/n Based on all this, I hope you can appreciate that mDAU is neither deceitful, nor out of ordinary, nor a useless metric. Sure twitter CAN talk about total spam on it's platform, and total DAUs , but that's a judgement call by the board . mDAU standalone is a useful metric
1
4
15/n Speaking of useless metrics. There are some useless scores though that make way into SEC filings of many many public companies, eg : NPS score (Net promoter score). ....But.I digress
1
2
16/n I do have a newsletter if you wish to read more of my rants about niche things that many people do not care about .
Replying to
n/n The entire thread is also on my blog . But you folks seem to read threads more than blogs...so don't blame me for keeping with with the times :D
1
4
And....back to top
Quote Tweet
1/n A lot of, generally smart, people have been raising a lot of concerns around Twitter's proprietary "mDAU" figure (Monetizable daily active users). I think those "concerns" may be a bit misguided .
A long thread on why its not a bad metric and why its not even uncommon.
Show this thread
1
1
My key takeaway from recent disclosures. Sure there sessions lots of allegations, but it’s not clear how much merit there is to them.
Quote Tweet
My key takeaways
Musk and team do not like mDAU, they thought mDAU is directly correlated to engagement (which it is not), and revenue over users spread follows the standard 80-20 rule ... twitter.com/chancery_daily…
1