Engaged Reading Time - Issue #52

Adam Tinworth

Oct 15, 2019 — 3 min read

One of my core principles in thinking about audience engagement is that we have to get away from orbiting around Facebook as the central source of what we do. And that's why I've buried the Facebook stuff half way through today's newsletter.

However, with major elections coming up in both the UK and the US, Facebook's role in politics will come under ever-deeper scrutiny, so it'll probably be a recurring theme for the next year, at the very least.

But let's start with something a little more encouraging…

Engaging Read

The business case for listening to your audience is still murky (but early results are promising)

"Newsrooms generally reported increased levels of engagement and higher-quality conversations with audiences. Nearly all of the grant reports we reviewed reported increased engagement." That (still) doesn't yet translate into money.

Visual learning

Hold onto your hats, I'm gonna get a bit geeky here. Let's talk about how a couple of social platforms are starting to use image recognition.

Learning a unified embedding for visual search at Pinterest

Visual search is critical to discovery on Pinterest. This is a description of how it uses machine learning to deliver that.

LinkedIn's AI automatically generates photo text descriptions

LinkedIn details an AI tool that automatically generates alternative photo captions using machine learning. A good accessibility move - that also provides data for the service.

Facebook and Politics

It's hard not to make these emails all about Facebook sometimes, as it's become such a centre of attention. However, here are three conceptually linked stories that explore how the platform is interacting with the outside world.