Metrics

Question 1

Hey all, Looking to get some insight in this FB execution question:

> > > You’re the PM of Reactions on Facebook. Your gut tells you that the average number of these reactions per post will be lest than the current number of likes per post. What would success look like in terms of non-like reactions per post at launch and how do you come up with this? Would this number differ by reaction? Why or why not?

I will put my thoughts in a threaded response to this post, and welcome feedback if anyone has it. Thanks!

Answer 1

Here’s how I would go about it. I will split the success into two parts: launch success & continuous adoption.

Launch success are metrics that revolve around getting a start in usage of non-like reactions. Chances are that without a (or even with a marketing campaign), these additional reactions will be missed. Owing to FB’s popularity across the globe, and how people in developing economies are still getting to know like better, a geographic success rate would also be crucial. So I’d like to track these numbers:

• No of non-like usage (per emotion) from launch, with bucketing at WoW, MoM levels
• Average non-like reaction per post, binned by geographies
• Alternatives used for a similar reaction, i.e. words used in comments or emojis that signify the equivalent of a non-like reaction
• Would want to see this in a time-series till a point that non-likes contribute to about 25% of the total reactions since launch

Long term, for continuous adoption,

• I’d look at the like-reaction alternatives, such as comments, and see how those numbers have reduced vs how non-like reactions have grown
• How non-like reactions have affected shares of posts and lifecycle
• Do a preference analysis and retire or promote a kind of reaction depending on the result of above 2 factors

As to whether this number would differ by reaction, I’d say a yes. I see reactions as a quick comment rather than write a reply to the post in question. Its highly possible that people find this form of expressionism much less verbose than writing comments. It could be a proxy, without any replies.

—

Alternative answer

Here are some thoughts I had around this:

Reactions are: love, laughing, shocked, angry, etc

How does it work? Press-and-hold to reveal, then let go under appropriate reaction

By nature of the UX, harder/more expensive to get to reactions + may contribute to them being lower than likes

Of the new reactions, ones that are similar to “likes” are likely to get more action b/c they are “positive”. Users are only used to reacting with affirmation

We may want to look at “non like reactions per post” in aggregate, compared to “likes”, because of this. “Angry” + “Shocked” reactions may be polarizing. People disagree all the time, but the reaction signaling is a bit more revealing (versus just a negative comment)

Metrics Brainstorm

# of non like reactions per post
# of each reaction avg per post
ratio of
# non likes to likes per post, % of non like reactions per post
(maybe this is on a scale)

At launch, it’s probably most important that any reaction other than likes gets used at all

Every post isn’t created equally (text only, long text, text + media, media only, etc.). Maybe we analyze the kinds of content pre-launch that get the most likes on average, and see if “reactions” start to penetrate that. Posts that have gotten much discussion but little likes, might get more “negative” reactions.

10%, 20%, 30%?

of comments per post (does this feature have a negative or positive influence on comments?)

of posts (does this feature have a negative or positive influence on posts?)

From here, I’m a little stuck. I’m thinking this question is asking for a specific # or % as it pertains to this specific metric (non like reactions per post), and I’m having trouble wrapping up these thoughts with a concrete recommendation.