2 ms·
What I propose doesn't have to be that expensive. How many active startups are there today? Too lazy to re-open up my old copy of CrunchBase...but let's say the
by danso 11y ago
What I propose doesn't have to be that expensive. How many active startups are there today? Too lazy to re-open up my old copy of CrunchBase...but let's say there's 10,000. The Twitter API let's you perform nearly 18,000 user lookups. With each lookup, you get current number of followers and tweets, and you also get their latest Tweet as an object. If you approach this in the most brute, blunt force way -- paginating through every account's timeline...even if each account was at 3,200 tweets (which is relatively rare for small company accounts that haven't yet brought in a social media team), it would only take you 18 days to collect every single tweet. However...what's the point in doing that? You could also just do a fetch for 200 tweets per user in a single call and extrapolate the tweeting rate for all 10,000 companies on a daily basis...very few Twitter accounts will send out more than 200 tweets in a 2 day period (and for those that did, you can assume that they're still kicking...or are going on a bender).
The analysis doesn't have to be real-time...companies usually implode as a slow burn. If you look at companies like Ashley Madison and Superfish (OK, only the latter is a startup)...they pretty much shut up on Twitter as soon as things blew up. But companies that slowly lose steam...I imagine firing the social media person, or even just having much less heart into social media, is a common symptom. And one that is hard to see unless you're frequently checking on a specific company...but relatively easy to see in a simple statistical analysis. Yes, there would be false positives...but 20-50 false positives a couple of weeks versus 10,000 potential false negatives to check up on is still an advantageous situation.
- minimaxir 11y agoHmm, that math with the rate limits does work. With Twitter's stance on third party data usage, it would be risky, but I think the idea could work.