A week ago, myself and 5 other EAv’ers participated in the Facebook Five Experiment.  Here are the details, data, observations, and conclusions.

Hypothesis: The most recent changes to Facebook give more prominence to those Facebook posts that have more likes and comments.  The hypothesis is, if a group of people was intentional about giving each others FB updates more likes and comments, those updates would be seen by more of their friends and fans and get even more likes and comments.

Experiment: I and 5 other participants agreed to check each other’s Facebook page and profile each day for 1 week and make an effort to like or comment on 5 updates each day.  The result would be 5 participants x 5 actions per day x 7 days = 175 additional Facebook actions on each of our pages and profiles for the week.

Challenges: The experiment was fraught with challenges that made the impact of the experiment difficult to measure:

  • One participant did not begin participating until the 3rd day.
  • One participant posted to their FB page in a non-English language.
  • One participant did not post to their FB page or profile for several days.
  • One participant had 750 likes/comments on their FB profile the week before the experiment, making the impact of 175 likes/comments from the experiment difficult to observe.
  •  I noticed inconsistencies between the number of posts, likes and comments EAv reported on my main profile page and what my “advisors” Sage and Monte reported.
  • We started the experiment only 1-2 weeks after EAv made a major change to their algorithm for calculating dividends.

Raw Data: For each participant I recorded the share price and dividend plus the number of posts, likes, comments, and score for their FB page and profile.   Here’s the raw data.

Anne Thomas (e)ANNET

  • Before: share price 141.07e, divs 1.27e/share
  • After – share price 146.54e, divs 1.43e/share
  • FB Profile – 161 FB Posts, 36 Comments, 154 Likes this week (EAv score: 50)
  • After – 80 Facebook Posts, 58 Comments, 148 Likes this week (EAv score: 53)
  • FB Page – 104 FB Page Posts, 64 Likes this week (EAv score: 13)
  • After – 86 Facebook Page Posts, 40 Comments, 117 Likes this week (EAv score: 13)

Robin Good (e)ROBINGOOD

  • Before: share price 126.13e, divs 0.96e/share
  • After – share price 124.64e, divs 1.12e/share
  • FB Profile – 9 FB Posts, 39 Comments, 30 Likes this week (EAv score: 59)
  • After – 20 Facebook Posts, 63 Comments, 106 Likes this week  (EAv score: 60)
  • FB Page – 24 FB Page Posts, 53 Comments, 37 Likes this week (EAv score: 30)
  • After – 28 Facebook Page Posts, 205 Comments, 83 Likes this week  (EAv score: 30)

Carece H Slaughter (e)CHSLLC

  • Before: Share price 148.54e, divs 1.92e/share
  • After – share price 161.44e, divs 1.81e/share
  • FB Profile – 43 FB Posts, 12 Comments, 120 Likes this week (EAv score: 68)
  • After – 48 Facebook Posts, 55 Comments, 196 Likes this week (EAv score: 69)
  • FB Page – 7 FB Page Posts, 1 Comments, 36 Likes this week (EAv score: 17)
  • After – 22 Facebook Page Posts, 31 Comments, 281 Likes this week (EAv score: 17)

Warren Whitlock (e)WARREN

  • Before: share price 130.26, divs 1.55e/share
  • After – share price 133.78e, divs 1.34e/share)
  • FB Profile – 12 FB Posts, 46 Comments, 22 Likes this week (EAv score: 61)
  • After – 11 Facebook Posts, 70 Comments, 68 Likes this week (EAv score: 60)
  • FB Page – 2 FB Page Posts this week (EAv score: 8 )
  • After – not displayed

Peter Santilli (e)PEOPLEHUBUSA

  • Before: Share price 142.93e, divs 1.47e/share
  • After – share price 147.61, divs 1.23e/share
  • FB Profile – 275 FB Posts, 271 Comments, 498 Likes this week (EAv score: 83)
  • After – 211 Facebook Posts, 141 Comments, 413 Likes this week  (EAv score: 85)
  • FB Page – 11 FB Page Posts this week (EAv score: 3)
  • After – 16 Facebook Page Posts, 1 Comments, 12 Likes this week (EAv score: 4)

Paul Steinbrueck (e)PDSTEIN

  • Before: share price 112.93e, divs 1.20e/share
  • After – share price 119.03e, divs 1.12e/share
  • FB Profile – 29 FB Posts, 24 Comments, 35 Likes this week (EAv Score: 41)
  • After – 31 Facebook Posts, 51 Comments, 111 Likes this week (EAv Score: 42)
  • FB Page – 23 FB Page Posts, 6 Comments, 25 Likes this week (EAv score: 7)
  • Now – 28 Facebook Page Posts, 31 Comments, 86 Likes this week (EAv score: 8)

Analysis

  • Warren’s participation in the experiment was minimal, so I excluded his page & profile from the analysis.
  • Of the other 10 pages/profiles, 7 had a significant increase in likes/comments compared with the previous week.  I decided to focus specifically on those 7.

The main thing I wanted to measure was the number of likes/comments/shares by non-participants on each page/profile  to see if the experiment increased visibility and engagement among non-participants.  However, measuring this would have meant checking every like/comment/share to see if it was made by a participant or non-participant and tallying them all and I just did not have time to do this.  Additionally, if I had measured this and found increased engagement by non-participants, it would have been impossible to tell this was due to the participants likes/comments/shares or the increased number of updates we were making.

Beyond the impact on the engagement in Facebook, I also wanted to try to measure the impact the experiment had on Empire Avenue.  Here are a few things I found interesting…

  • 3 FB profiles experienced a significant increase in likes/comments.  All three also experienced an increase in network score of one.
  • Of the 4 FB pages that experienced a significant increase in likes/comments, only 1 experienced an increase in network score of one.  The other 3 were unchanged.
  • For what it’s worth, the dividends of 2 participants increased during the experiment while 3 decreased.

Latent Results

After the experiment ended last Wednesday, I compared the number of posts, likes & comments shown on my profile to the numbers reported by my Facebook “advisors” and noticed the advisors were reporting numbers that were about half of what was on my profile. Then after the experiment was over and the number of posts, likes and comments each day went down, the numbers reported by my advisors continued to rise. The advisors’ data seems to lag by a few days.

So, I went back yesterday and checked each participant’s Facebook page and profile network scores. On average those 7 pages/profiles that experienced a significant increase in likes and comments during the experiment had their network scores increase by another point after the experiment was over. Whether that is because of experiment or not is hard to say.

Additional Observations

A few other things worth noting:

  • Going into the experiment, I estimated it would take about 20 minutes a day to participate, but because many of the items posted by participants to the FB pages/profiles were videos & articles it actually took me around an hour a day.
  • Some of the comments made by participants seemed forced. One of the participants told me friends & family asked him what was up with all the odd comments.
  • I found it difficult to like/comment on content because I just didn’t share the same interests as some of the participants.
  • On positive note, I came across some excellent content posted by the other participants which I wouldn’t have otherwise seen if it hadn’t been for the experiment.
  • I also enjoyed getting to know Anne, Robin, Carece, Warren, and Peter a little better through the experiment.

Conclusions

  • Did the experiment increase engagement on our Facebook profiles/pages?  Inconclusive.  It may have but there were so many inconsistencies in participation it’s impossible to say for sure.
  • Did the experiment increase our Facebook network scores?  Probably.  The 7 pages/profiles that experienced a significant increase in likes/comments also experienced a combined 12 point increase in network scores during the week of and week after the experiment.
  • Did the experiment increase dividends?  Nope.  There’s no way the experiment would have caused the decrease in dividends some experienced during the experiment period though, so there were obviously other factors involved in the changes in dividends participants experienced.
  • Would I do it again or make this an ongoing part of my social media strategy?  No, not in this way.  However, I do think something like this with 5 people who share more similar interests could work and be worth the time.  If I were to do another experiment like this, I would also do it for 2 weeks in an effort to overcome what seemed to be delays in the data by EAv.

Thank you Anne, Robin, Carece, Warren and Peter!  I appreciate your help with this experiment.

Comments?  Questions?

Advertisements