Web Notifications, Part 1: Interactive
Breaking down a dense topic with paths of information.
Interactive notifications, through which we give users the opportunity to self-direct as they move through a narrative using action buttons embedded below a small amount of text, were the first type we worked with, starting with the first jobs report experiment.
In this format, currently only available on Chrome, the user’s response to a prompt within the initial notification informs the interaction they will continue to have with the notifications to come. By using the action buttons below the body text as a navigational tool, the user moves through a thread. We’ve run several iterations of an interactive notification through the jobs report experiment each month.
We set up the interactivity to give the user two possible tracks to follow — good news and bad news — in an order only they determine.
After the user receives an initial alert with directions and options …
… they click on one of the action buttons — the “Give me 😀 news” or “Give me 😟 news” — to decide what alert they will see next.
Eventually one of the options runs out. After all the good news, the user sees this:
And when the users reach the end of the sequence, with both good news and bad news exhausted, they are given the option to read the full article:
We’ve written pretty extensively already about the jobs report experiment so far, here and here.
We think interactive notifications are an interesting option for newsrooms looking to try a new format for an explainer, or to allow users to chart their own path through a series of facts. It’s also a way to test how to break down a dense topic into pieces and maybe even allow for playfulness in the explanation.
One of the reasons we started with interactive notifications was because we were curious about how to use the action buttons in a news context. We’ve used them thus far to support the “good news” or “bad news” paradigm, but depending on what you are trying to explain, other dichotomies could work similarly well.
To aid those looking to explore, here are some limitations we’ve discovered thus far.
- Start small and simple. For interactive alerts, you need to have the entire notifications queue ready to go before you start sending, so start with something that you can easily map out, including all the possible outcomes. In each of the jobs report notification experiments we have run we have had six notifications overall, three for good news and three for bad.
- Difficult to use in breaking news. Because you need the entire path queued up before you can deploy the notification, interactive alerts are difficult to put together in a breaking news scenario where all the facts are not yet known. We had also planned to send a series after the EU referendum, however, news was still breaking very quickly and we found it too difficult to string together a coherent narrative that was also timely.
- To get clean data about user preferences, consider the order of the options. Interestingly, users preferred clicking for good news before bad news at a ratio of 2:1. However, we have also put the good news action button on the left, where the eye naturally starts reading first (at least in English). Before we can make judgements about the type of news our readers prefer, we should randomize the button placement to reduce bias.
More in this series about web notifications:
