I run a Facebook group, Digital Product Mastermind, and recently I’ve been struggling to understand why my posts are not being seen by all members.

Let me start off by saying I really don’t have the answer to Facebook Group engagement yet; that’s not why I’m writing this. I’m beginning to dive into the data and analytics of my group, and I thought that my process could help you do the same. This process would also work with other social media platforms, like your tweets or your Facebook Page.

My Pre-Group Assumptions

When I started my Facebook Group, I had a few assumptions of how things would work:

1. In a group, Facebook wouldn’t play the News Feed game. All my posts would get seen by members, or at least, they would have the highest priority in the group.
2. Following the conventional wisdom of setting up open-ended questions and starting discussions would drive Facebook Group engagement easily.
3. The hotseats would drive a ton of engagement and get a lot of members excited.

These assumptions weren’t wrong, but they were unproven. Now, with some time and data, I can look at these assumptions and adjust.

Building a Spreadsheet Framework

To get started, I opened up a blank spreadsheet. I made the columns that were relevant to the ways I measured engagement:

Posted By: who was the original poster?

Day of the Week: what day Monday-Sunday was this post added?

Time: what time of day (in my time zone) was this post added?

Content Category: what kind of broad category did this post fall under?

Seen By: How many people saw this post? (it’s my understanding that this measure is only available to groups with 250 members or less)

Likes: How many people liked the post?

Comments: How many total comments did this post get? (including the original poster, and multiple comments by the same person)

Buffer Analytics: Did I share this post with Buffer, and what did the Buffer numbers say?

If you’re following along, also add the column “Commenters” and tally the number of commenters, regardless of how many comments they made. I didn’t do this until it was too late, and you’ll understand why further down.

Data Entry

Then, I started manually adding this information to the spreadsheet. (If any more tech-savvy people know of a faster way to get this information into a spreadsheet, by all means, share share share!) It was relatively simple to add the relevant information as I scrolled through the group. To get the day of the week I had to hover over the timestamp, as the posts generally just show the date (September 10 vs Thursday).

My original goal was to find out the best dates and times to post for maximum exposure. But at around 43 posts, I stopped entering the data (I probably had a total of 75 or so. Definitely less than 100). Not because the previous posts were insignificant, but because I started to see a pattern.

I had reached the middle of August, a little less than one month ago. All the posts around this time averaged about 50 views (out of 70 people at the time). Compared to two weeks later, when posts were averaging around 30 views (out of 75 people). And compared to now, when my posts are getting closer to 15 views (out of 78).

This was around the time I invited several people from the second Small Product Lab to join DPM. I came to the assumed conclusion that this definitely impacted the engagement at the time. There was excitement from new group members, and they were posting and commenting frequently. More recently, we haven’t had as many new members join, and thus the excitement and momentum may have slowed.

Analysis: Post Exposure

I sorted my spreadsheet by “Seen By,” because this is the metric I am most interested in. I can’t get an accurate picture of conversion without getting more eyeballs on my posts, right? Here are the top 5 posts in that category (names of other members redacted because we’re basically the CIA):

I am responsible for 3 out of 5 of the posts, which was a little bit of a relief. The top post was mine, and it was a call for discussion topics for our first event. It was all text, with an open-ended question.

Another post of mine was a hotseat post, with long text and a video. Hotseats are a feature of smaller masterminds, where one person opens up their business and their struggles to the group. We do this in DPM by putting someone in the hotseat once a week, for a whole week, and dedicate a thread to it. Interestingly enough, this was the third hotseat we did as a group, and the others we did before and have done since are much lower in the rankings.

And the last one was a discussion about paying for tools. It stemmed from an internal rant of mine. It was awesome to share my thoughts with everyone and hear theirs. This was a long, text only post.

Something that may just be a coincidence or may be significant: all three of my top posts were posted through Buffer.

The second most exposed post was also an all text discussion/open-ended question post, and it was not posted by me. This post got a ton of comments in comparison to any other post, and it could have been considered somewhat controversial, with people taking principled stands on either side.

There is no pattern in the day or time of any of these posts, which bunks the hypothesis I had going into this analysis. However, there was a pattern in the type of content that was seen by the most people: discussions and open-ended questions.

Analysis: Date & Time

Next, I sorted by days of the week. I was still determined to find some sort of pattern here.

An interesting note about this process: as I spent more time with the numbers, I began to look at views in the 30s as solid performance, like a C. When I started, I thought any scores in the 30s were Fs. It’s all about perspective, folks!

I like to keep everything visually neat, so I made a new tab for each day of the week, and then I sorted by ‘Seen By‘ and then ‘Time.’ If you are following along at home, be sure to correctly format your ‘Time‘ cells as Dates, so you can sort them properly. Otherwise you’ll have 4 PM sorted above 6 AM.

I’m apologizing in advance because this is where I show a bit of my spreadsheet eccentricity. I thought it would be really cool and helpful to plot a colorful timeline of my most seen posts. In my head it was going to be this beautiful graph that clearly indicated the best times for me. Once I started though, I knew I didn’t have enough information to complete the timeline. After I filled in the colors, all I saw was the gaping, untested white space. But on the bright side, I now have a chart I can easily refer to as I test new post times and dates.

Green: 50 views or more

Yellow: 35–49 views

Orange: 26–34 views

Red: 34 views or less

The last thing I want to look at is what time my members are posting. This will at least give me a clue as to what time they’re on Facebook and checking the community. I sorted by ‘Posted By‘ and ignored all the posts by me. I charted the other member’s posts on the same calendar sheet, so I could compare the data. I can’t say this was as eye opening as I would have hoped. There is still a lot of white space in the calendar. But now I have a good place to chart going forward!

Ok. I’ve exhausted the analysis that I want to perform on exposure. Let’s turn to engagement.

Analysis: Engagement

A couple things I need to establish about engagement:

Some of the posts went up when there were less members in the group. Unfortunately I don’t have an easy way to see the exact number at that time, so I’ve decided that I’m going to use the current membership total of 78 for my engagement calculation.

My Facebook Group engagement calculation is:

(likes + comments)/group members: This tells me how much interaction a post received.

I also wanted to include:

(likes + commenters)/group members: This tells me how many individual group members are interacting with the post.

Each calculation on their own could be misleading. In the first calculation, one person could comment ten times and the percentage could be flawed. In the second, ten people could each comment emojis and that could also be misleading. With both equations, I can get a fuller picture.

However, I decided I wanted to do the second calculation after I had already collected my data, sorted and parsed it. What I should have done is counted ‘Commenters‘ at the beginning, or at least included the link to the post in the spreadsheet.. It’s a bit more complicated now because I didn’t add the actual post date, just the day of the week and time.

So as I collect more data, I’ll be sure to collect information about commenters as well, and the links to the posts.

This time around, I’ll just use the first engagement calculation. Here are the top five posts based on engagement rate:

Every single one is a text-based discussion post. The hotseat, interestingly enough, has the lowest “Seen By” rate. The others are in the 50s. With the exception of one, they were all posted in the afternoon or evening. Only two by me this time, and one of them was someone else’s hotseat so I’m not sure that counts!

The last thing I did was sort my data by “Content Category.” It confirmed what I thought to be true: articles had low exposure and low engagement. I had felt anecdotally for awhile that sharing articles wasn’t really worth my time, and this confirmed it. I’m better off saving the articles for Twitter (where they do very well), and focusing on driving discussion on Facebook.

So now that I’ve run through my small data set, I want to return to my assumptions from the beginning of the post.

Analysis: My Assumptions

“In a group, Facebook wouldn’t play the News Feed game. All my posts would get seen by members, or at least, they would have the highest priority in the group.”

Oh boy. This one stings, it’s so untrue. Of course Facebook is playing the News Feed game! First of all, the group’s posts show up in members’ News Feeds. And even if you look at the discussion section of a Facebook Group, the posts aren’t in chronological order.

“Following the conventional wisdom of setting up open-ended questions and starting discussions with would drive engagement easily.”

I would agree with the statement about 90%. I’m not sure that engagement is being driving “easily.” Discussions and open-ends work, but they work best when I have other group member participation.

“The hotseats would drive a ton of engagement and get a lot of members excited.”

I think the hotseats have benefited from being novel and an interesting introspective into other group member’s businesses. They haven’t been the engagement magnet I had hoped they would be. People have been reluctant to sign up for a hotseat spot, and not everyone has the chance to see the post when it’s posted.

However I’m really proud of the hotseats feature and any creative idea I’ve brought to the mastermind so far. Even if an experiment bombs, it’s a data point I can use to fuel my future.

So what now?

I know I need more data. I need to test out different days and times of day to post, and I need to do that with more discussions and open-ended questions.

I also need to increase the interactivity of hotseats. I think they could be bigger and better. Maybe adding a personal touch off Facebook, using Google Hangouts or Blab.

The last thing I need to do is cut myself some slack. I’m growing a community, and that takes time. I have a very clear vision for DPM- I want it to be a place product creators want to be. That means their experiences and opinions about the group help define the path, and I can’t plan out every day, week or month to a T.