Unsolicited data scientists solving your problems without using your data

Here I described how I identified a root cause, and provided a dual solution for a problem impacting LinkedIn and its members. For your reference, I also included links to many other solutions, coming from our data science research labs, applied to a bunch of companies and business problems.

The problem: LinkedIn email blasts deployed to group members are working less and less, many times not at all. I stopped receiving most of them two weeks ago without any action on my part - though I like them as it helps me identify great articles for my weekly digests. Posting an article on LinkedIn is becoming incredibly more difficult as popular bloggers get flagged as spammers, as a result of this email spike featuring the same postings over and over.

The explanation: LinkedIn recently started to send email digests from all groups, to all members, too frequently (or maybe they improved email delivery and these messages started to pop up like mushrooms after the rain, in Inbox rather than Spambox). In these digests, the most popular articles appear at the top. Great bloggers posting top quality content get featured over and over - the same articles from the same guys appearing in multiple blogs (because of cross-posting) week after week (because of the popularity of the postings in question). Eventually, the popular blogger gets flagged, is unable to post, and lower quality content eventually shows up as the replacement.

The solution: If you are a popular blogger, stop posting on LinkedIn, and instead post on your website. Content posted on LinkedIn competes with your content on Google search, and if your content no longer shows up on LinkedIn, people will eventually have to subscribe to your website - long term this is a winning strategy for you, backed by data science evidence. You can hire a few interns or create a few profiles to still continue to post on LinkedIn and comment existing posts, but your strategy should be to stop relying on LinkedIn. For LinkedIn, the obvious solution is to NOT featuring over and over the same great posts across multiple groups in these group eBlasts that feature popular articles, and limit cross-posting to 2 or 3 groups maximum. Don't do a group blast for groups with no NEW great content since last week, that's one of the main sources of same content being hammered week after week to many users. Don't rely exclusively on your "popularity score" algorithm, apply filters, instead design a content diversification algorithm as described below. This will solve all the issues.

Content diversification algorithm: I suggest LinkedIn data scientists to work on this new algorithm  asap. In short, you need to design an algorithm that detects when an article has been over-featured, over-posted, over-promoted, and stop promoting it. It involves collecting and analyzing data from different data silos, to make sure that  the proportion of people seing the same blog post more than 3 times a week in their Inbox, is minimized. Such an algorithm would ultimately deliver rich, varied content to all users, a win-win solution both for users and LinkedIn.

Other business problems that we analyzed

Proposed solutions to some government problems

Views: 3319


You need to be a member of Data Science Central to add comments!

Join Data Science Central

© 2021   TechTarget, Inc.   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service