2 ms·
I don't understand - can't one calculate MAU from firehose data? If so then all companies who bought that data can access this elusive MAU metric.
by Lur 12y ago
I don't understand - can't one calculate MAU from firehose data? If so then all companies who bought that data can access this elusive MAU metric.
- minimaxir 12y agoMAU includes users who passively consume content and do not post content. Deriving MAU from tweet data alone will vastly underestimate it.
- bhouston 12y agoYou can likely calibrate past firehose data with past MAU data and then you can use firehose as a predictor. I am sure there are strong correlations between the two.
- patio11 12y agoIn addition to this, the firehose license almost certainly includes 96 pt font printed in blood-on-asbestos specifically enumerating "publishing aggregate statistics" as one of the things you're not allowed to do with the data, on pain of a breach of contract suit with damages measured in percentage points of Google's market cap.
- MichaelGG 12y agoSo why publish it, when you can just calculate and trade on the info? Though the costs of getting all the data might hurt the profitability of this plan. (The resellers charge per tweet, so you'd need to fall back to a sampling scheme, though that might be his enough.)
- joosters 12y agoAn 'active' user doesn't have to be tweeting, so they won't all show up in the firehose. MAU figures are pretty dubious anyway - see the other recode story about how 4 million 'active' users vanished because iOS no longer automatically fetches tweets: http://recode.net/2015/02/05/how-twitter-lost-4-million-users-because-of-apples-ios-8/ http://recode.net/2015/02/05/how-twitter-lost-4-million-user...
- MichaelGG 12y agoSo, in that sense, going by unique tweeters is an even better metric. Also, since Twitter offers limited (1% or 5%) of the firehouse to anyone via the streaming API, wouldn't that be a fairly decent way to estimate things?
- dredmorbius 12y agoThe data can still be correlated. Given a history of known actives and corresponding tweets, the average tweets-per-active can be determined. That may change over time but it's unlikely to do so in a capricious fashion without good cause. Say, onboarding a large class of users for whom Twitter and/or Tweets are of secondary interest. Not that, say, Google might have done that with Google+ or anything. Just because you cannot _directly_ see the number you're interested in doesn't mean you cannot infer it by other means.
- AznHisoka 12y agoTrue, if only stock prices were based solely on quantitative metrics. But it's also based on human emotions and expectations.