I’ve spent my first week taking a look at how we can optimise our Facebook and Twitter channels to ensure each piece of content we release gets as much organic reach as possible, as well as how we can involve more CCs (and eventually community) in the content creation process.
The findings and suggestions here are very much just initial thoughts, so I’d love any and all comments from the wider team if you spot anything that you feel should be added or amended.
You can check out the original WIP Google Doc, including some comments already left by CCs, HERE. Feel free to leave any comments in the doc or directly below
Social Media Publishing Guidelines (WIP)
To best optimise our content reach there are two focus areas we should consider in regards to our content cadence:
- Algorithm optimisation
- Most engaging times for our specific audience
At the conclusion of ETHIndia, Shawn did a deep-dive into our engagement decline. Many of his findings are relevant to our publishing strategy moving forward, including:
- Engagement with our previous content - We published 9 highly technical videos within quick succession across 2-3 days with each individual video getting very low engagement, this impacts the propensity for our followers to receive future posts
- Media Cards - all the videos we’re embeds of YouTube links, Twitter prefers for users to stay in platform and as such if you’re posting 9 YT embedded videos in quick succession you will be punished by this as Twitter would prefer a native experience
- Our own broadcast content (not RT and Replies) exceeded 5 a day. A number of agencies and social media technology companies have conducted studies with clients and have all found that after 5 broadcast tweets (original content) in a day the impressions per post begins to decline.
- _Here’s a more in depth look by Twitter into how the tech behind the algorithm works - https://blog.twitter.com/engineering/en_us/topics/insights/2017/using-deep-learning-at-scale-in-twitters-timelines.html_
- With ethindia we took on the approach of having our community ambassador publish the content and we would simply retweet, this is seen as amplifying UGC and sits outside of the 5 tweets per day
- Outside of that we slowed down the volume of broadcast content mindful of the Rule of 5
- We used plenty of Native media in Twitter, almost all of EthIndia content was Image based which were published natively rather than promoting content housed on a separate platform.
Using Sysomos, I pulled engagement data from a range of dates across 2018 and combined the results to find the times our audience is most engaged. This data was a combination of Status-specific search terms, as well as broader Ethereum terms, and highlighted the following peak times we should be tweeting each day, keeping in mind the Rule of 5:
- 10:00am GMT (2:00am PST)
- 3:00pm GMT (7:00am PST)
- 5:00pm GMT (9:00am PST)
- 7:00pm GMT (11:00am PST)
- 9:00pm GMT (1:00pm PST)
Example screenshot from Sysomos (times are GMT)
For Facebook, it’s the Rule of 2, which most studies suggest is the optimal number of posts per day. Facebook Insights for our Status page indicate the following peak times we should be posting each day:
- 7:00am GMT (11:00pm PST)
- 3:00pm GMT (7:00am PST)
Example screenshot from Facebook Insights (times are GMT)
Publishing strategies for acquisition
You never get a second chance at a first impression. When new potential followers land on our channel for the first time, they naturally will do a quick scan of the most recent pieces of content we’ve published and make the mental decision to hit ‘follow’ or decide that our content isn’t for them. Once the latter decision has been made, it’s much harder to win that user back.
To present our best face to new users at all times, the following strategies are worth considering:
Limit retweets to a maximum of 4 per day, with each retweet spaced in between our original content, resulting in our feed looking fresh and engaging at all times. As an example, this strategy prevents users arriving out our account for the first time, glancing at our 3 most recent tweets being retweets, and deciding to move on without following. Taking into account the Rule of 5, this will result in our content ratio sitting at minimum 55% original content at all times.
Retweets with comments will be considered original content, so whilst they will count as one of our 5 daily tweets, they will not impact the 4 retweets per day strategy.
Ensure there is something for everyone at all times. Our 5 pieces of original content per day should be 5 very different pieces of content, opening up the appeal of following our account to the widest audience possible. For example:
- 10:00am GMT | PeopleOps recruitment tweet
- 3:00pm GMT | Ghost article tweet
- 5:00pm GMT | [Free slot for ad hoc content identified by CM]
- 7:00pm GMT | Town Hall plug
- 9:00pm GMT | Community advocate program plug
Allocate a free slot each day for time-sensitive/ad hoc content
Content Contributions from CCs
To ensure the above learnings and strategies are implemented, and that each piece of content itself is optimised, a minimum notice period would ideally be in place for scheduled, non-urgent content. Wherever possible, and excluding reactive content, requests would be submitted to the relevant Wekan board 48 hours before the requested publishing time.
An example process flow could look like this:
- CC creates Wekan card with their content idea, having a crack at filling out everything, from copy to asset, ensuring empowerment for everyone to contribute. This card would be added to a ‘submission’ list of a dedicated Wekan board and tags Engagement team
- DMM (Digital Marketing Manager) checks current content schedule and, understanding the strategies mentioned earlier in this doc, selects an appropriate date and time to schedule the post, and moves the card over to weekly content calendar Wekan board
- DMM tags the CC, letting them know the date and time the content will be going live, and also has the opportunity to request more info/asset if required
- If necessary, DMM begins the process of content ideation, taking into consideration the objectives of the announcement and the info provided by the CC. In the case that any necessary changes to the Wekan submission are identified, DMM contacts the CC to explain reasons why changes were made, and helps educate the CC on social best practices
- DMM schedules tweet containing final copy and asset
The aim of this process is to two-fold:
- Create a permissionless mechanism for producing content, empowering CCs with as much involvement as possible even with a limited social tool key set, and provides complete transparency for all CCs to see what content is scheduled and for when
- Become as efficient as possible, freeing up time for our CMs to embed themselves in the community and foster meaningful relationships with key evangelists
With the understanding that a lot of teams have a large number of content they’d like published, the above process flow will help all pieces of content come to light under the most optimised circumstances. For example, if PeopleOps have a list of key open roles they want to focus on promoting, we can add them to our daily content mix using the above steps, and a listing can be broadcast each day, with PeopleOps playing a larger role in making that happen.
Again from Shawn’s ETHIndia learnings:
- Engaging more than you broadcast - as we increased our volume of content published this took time and effort away from Replying and engaging with the community. You need your replies to publish ratio to be high in order to be seen as actively engaging in conversation rather than just broadcasting your own content
This is how we’re currently performing in terms of response rates.
Averaged across the past 90 days, we’re currently replying to 6% of tweets we receive.
We’re doing a decent job at replying on the days we do engage, but the large gaps of silence in between these active days leave us with a daily response rate of 11%.
When we do decide to engage with a reply, our average response time is 444 minutes (7.4 hours).
Averaged across the past 90 days, we’re currently replying to 10% of messages and comments we receive on Facebook.
Similar to Twitter, we’re doing a decent job at replying on the days we do engage, but the large gaps of silence in between these active days leave us with a daily response rate of 10%.
When we do decide to engage with a message or comment, our average response time is 797 minutes (13.3 hours)
NOTE: These metrics for Twitter and Facebook should be used as a base for improvement only, and should not be taken on face value. I ran a study which found that 42% of direct tweets we recieve are spam, and 19% don’t require a reply (metrics are an average across a sample of our most-replied to tweets across the past 90 days). Weekends also obviously have a huge impact which is simply a reality we take into account.
Engagement Strategy & OKRs
Improving on the above engagement metrics will greatly enhance our organic reach on both platforms, transforming our channels into a much more meaningful online presence.
By making some key changes, we should be aiming to achieve the following OKRs by the end of Q4.
Twitter & Facebook OKRs
- Response Rate | 20%
- Response Time | 240 minutes (4 hours)
- Implement CM roster based on timezones to ensure that someone is on point to respond to engagements at all times
- Ensure all CMs switch on the relevant alerts when they come online each day
- CMs/CCs to begin signing off on each reply to add personal feel within our community e.g. “-Hutch”
- Enhance our Response Doc. Create a living document that contains all details around replies to queries and comments, as well as a list of the relevant contact people from each team for when the answers aren’t included. This will enable CMs, as well as future community members who want to contribute, to quickly access the guide and be empowered to answer people with the correct and most relevant response.
- Turn on Facebook ‘instant reply’ - draft an automatic message sent to users when they message us for the first time.
- Turn on Facebook ‘away messages’ - draft an automatic message sent to users who message us on weekends
- MORE TO COME!