Continuing a long-standing trend, conservative content is performing much better than progressive content on Facebook in terms of both reach and engagement. It is critical that progressives improve their strategies and optimize their content for the platform in order to inform and persuade the public about key policy ideas.
In this report, the Center for American Progress Action Fund uses data from its Facebook social listening database, which spans more than 500 Facebook pages that have been identified as amplifiers of progressive-leaning issues or conservative-leaning issues, 1 to assess engagement on the platform in the second half (H2) of 2021. The data set includes 471,664 posts from 570 distinct pages between July 1 and December 31, 2021.
Trends over time
From July through December 2021, the story was largely consistent with previous months and years: Conservative-leaning pages held a significant edge on Facebook. Figure 1 charts the total engagements by week for the top 50 pages in CAP Action’s database in this period among both progressive and conservative pages.
Even though progressive and conservative pages had roughly similar average engagements per post (6,169 vs. 6,410), conservative pages averaged 20 million more engagements per week over the period because they posted 1.8 times more frequently. As seen in Figure 1, in some weeks, the disparity rose as high as 35 million to 40 million engagements.
Fig.1: Total engagements by week, top 50 progressive- and conservative-leaning pages
There are a few noticeable highlights over these six months, including a dip in conservative engagement throughout September, followed by a gradual return to normal in October. As was chronicled in a previous memo, this trend was largely borne out by declines in engagement with link posts, most notably from The Daily Wire. It is still not known whether that trend was driven by a few individual accounts or by a larger systemic change in Facebook’s algorithm. Either way, it quickly reversed itself, and conservative post engagements, and link engagements more broadly, climbed back up by November.
Other trends include:
- A significant spike in engagement for conservative-leaning pages at the end of August, corresponding with the United States’ withdrawal from Afghanistan
- Increases in both conservative and progressive page engagement throughout the first half of November, with many posts reacting to Glenn Youngkin’s victory in Virginia and the Kyle Rittenhouse trial verdict
- A decline for both progressive and conservative page engagement in December around the holidays, with both fewer posts and lower average engagement per post
Broadly, spikes and dips in engagement for progressive- and conservative-leaning pages often occurred concurrently as the news cycle ebbed and flowed, with major events tending to draw increased attention and engagement from both ends of the political spectrum.
Top pages
Below, in Table 1, we look at the top 10 progressive- and conservative-leaning pages by average engagement per post in H2 2021. The list is led by the page for Pete Buttigieg, which only posted four times in the database’s six-month window—a strategy that helps keep average engagement high—and whose new-parent photo post was one of the most popular posts of the year. Other than Buttigieg’s page, the pages of former President Barack Obama, Candace Owens, and Franklin Graham were the only other pages to average more than 100,000 engagements per post, and they got double to triple the average engagement of other pages on the top 10 list.
There are some patterns among the names. Top progressive pages are a mix of pages for individuals (nearly all of whom hold or have held office in some capacity) and pages that amplify progressive policy messaging with memes (such as Occupy Democrats, The Other 98%, and Call to Activism). On the conservative side, top pages are almost all for individuals, and most of these individuals are media personalities rather than current or former elected officials.
Table 1: Top 10 progressive- and conservative-leaning pages by average engagements
CAP Action also looked at the top pages by total engagements, which accounts for post volume. Ben Shapiro’s page posted an astonishing 16,000 times in the second half of 2021—equivalent to roughly 87 posts per day. Occupy Democrats averaged 39 posts per day but came close to matching Shapiro’s total engagement numbers with higher engagement numbers per post.
Several conservative media outlet pages also made this list—including Fox News, Breitbart, and Newsmax—while no progressive or liberal equivalent came close. More than anything, this reflects some of the long-standing work that has gone into building up conservative media institutions and investing in their Facebook presences.
Pages that promote conservative policies have a strong edge when it comes to total engagements, with the 10th-ranked conservative page generating more engagements than the fifth-ranked progressive-leaning page. There is only one conservative page that is also on the highest average engagement list, reflecting more pages dedicated to volume over per-post quality. Five progressive pages appear on both lists.
Table 2: Top 10 progressive- and conservative-leaning pages by total engagements
Top posts
When it comes to the top posts in this period among progressive- and conservative-leaning pages, CAP Action looked at two different metrics: reactions and shares. Table 3 shows the 10 posts with the most reactions in H2 2021, eight of which were from progressive pages.
These posts were generally not focused on policy priorities. They happened around holidays, anniversaries, births, deaths, and other personal milestones. They also tended to draw relatively few shares and comments: Reactions made up 81% of all engagements for the posts, and shares made up only 3%.
These post topics make sense from a behavioral standpoint: When public figures acknowledge events in their own lives that their audiences can relate to, it is easy to imagine huge numbers of people giving a quick like or love reaction in response to that relatability and then moving on. More often than not, there does not seem to be enough of a substantial hook to merit a share.
The most political posts came from the pages of President Joe Biden and Franklin Graham, the former calling out impressive job numbers and the latter congratulating Virginia Republicans for their recent electoral victories. Interestingly, both posts celebrate political victories, but neither appeal to deeper emotions, such as shock, outrage, or humor, that might drive shares.
Table 3: Top 10 posts by reactions2
Looking at the top 10 posts by shares (Table 4), there’s more of an even split between conservative- and progressive-leaning posts, as well as a more even distribution of types of engagements. Shares comprised around 37 percent of all post engagements, on average, among these posts.
Nearly all conservative top posts by shares are videos,3 while nearly all progressive top posts are photos. The one progressive top post that was a video was not politically focused, even though it was from NowThis Politics.
Posts from progressive-leaning pages were about such divergent topics as missing persons, whales, and hiking advice. The only substantive progressive policy they mentioned was a living wage, although the most-shared progressive post did highlight inequity in media coverage between missing persons Jelani Day and Gabby Petito along racial lines.
Top posts from conservative-leaning pages included pushback against mask mandates, praise for the military, and broad-ranging tirades against liberal influence or governance.
Table 4: Top 10 posts by shares4
Post topics
This section takes a closer look at what types of posts perform well for progressive- and conservative-leaning pages. The data set for all analyses in this and the following sections is restricted to the top 50 conservative- and progressive-leaning pages by total engagement from July through December 2021, in order to create a balanced comparison. This results in a data set of 287,885 posts.
Table 5 compares average post engagement by post topic, which CAP Action defined by searching the text of the posts for a series of keywords. (This means that any post with no text was excluded from this analysis.) A post can be tagged under multiple topics.
Some interesting patterns emerged. Progressive-leaning pages most outperformed conservative-leaning pages on a series of economic-related topics but underperformed in some policy areas:
- Despite getting fewer engagements per post on average, progressive pages did much better than conservative pages in the following areas: Build Back Better, care infrastructure, economy (a broadly inclusive category that includes keywords such as “jobs” and “inflation”), income inequality (which includes keywords related to taxing the ultrarich and billionaires), and infrastructure.
- Economy, income inequality, and infrastructure were similarly common topics among both groups of pages. However, Build Back Better and care infrastructure were among the least-referenced topics by conservative pages and appeared much more frequently on progressive pages.
- Progressive pages narrowly edged out conservative pages on posts related to gun violence but performed slightly worse on both climate- and health care-related posts.
- Note: The health care topic does not explicitly include COVID-19-related keywords but could have some overlap with the COVID-19 topic.
Table 5: Average engagements by topic and page ideological lean
Conservative-leaning pages most outperformed progressive-leaning pages on a series of partisan and identity-related issues:
- Posts that mentioned “Democrats” or “Republicans” performed much better for conservative pages, on average. This aligns with the idea that conservative pages more frequently lean into an “us vs. them” dynamic on Facebook.
- It is also notable that posts that referenced President Biden got higher engagement for conservative pages, which may be more likely to use him as a focal point for criticism, than progressive pages, which may be more likely to hold up his accomplishments and successes.
- Conservative pages made 27,000 posts referencing President Biden, by far the most for any topic and three times the number from progressive pages (about 9,000).
- Identity- and human rights-related topics such as LGBTQ rights, women’s rights, abortion rights, racial inequality, immigration, and voting rights all performed much better for conservative pages.
- The topic of COVID-19 also performed much better for conservative pages, as anti-vaccine and anti-mask talking points have become a priority focus among these groups. It was also by far the most common policy-related topic for both progressive- and conservative-leaning pages.
Post types
Table 6 breaks down post volume and engagement by ideological lean and post type. This breakdown reveals both substantial differences in strategy between progressive- and conservative-leaning pages and opportunities for exploration and potential growth in engagement for progressives.
Table 6: Percentage of posts by post type and page ideological lean
Link posts comprised the vast majority of posts by conservative-leaning pages, followed by videos. Photos were not employed often on conservative pages. Posts on progressive-leaning pages, on the other hand, had more variety, with a substantial amount of photos, links, and videos. Interestingly, despite chronic underperformance of videos on progressive pages, they were used more often on progressive pages than conservative pages.
Table 7 compares the average engagement that different post types received, with carousels, photo posts, and status (text-only) posts having the highest average engagement for both conservative and progressive pages. But CAP Action also compares that to the expected engagement for each post type, based on the average performance from the pages that authored the posts, since not all post formats are used by all pages equally.5 For instance, Barack Obama’s page is unlikely to share a lot of link posts, while a news outlet is unlikely to post text-only statuses. A higher expected engagement means a post type was used more often by pages with better performance overall.
By that measure, although carousels and status posts had high average engagements, they also had higher expected engagement, indicating that they were used more frequently by large pages with high engagement. So that success may not be easily replicable for smaller Facebook pages. There are also small sample size concerns with both post types, with carousels only being used a few hundred times in this data set.
Taking those factors into account, photo posts—which can be Tweet screenshots, produced graphics, or actual photographs—were the most reliably successful for both progressive- and conservative-leaning pages. They outperformed their expected engagement significantly for both groups of pages and generated five times the engagement of link or video posts for progressive pages.
Table 7: Average engagements and expected engagements by post type and page ideological lean
Videos see the biggest disparity between progressive and conservative pages when it comes to performance. For progressive pages, they were the least-engaged post type and most underperformed the expected engagement. For conservative pages, videos were among the better-performing post types, actually outpacing their expected engagement.
Finally, links tell a simple story: They are the most common post type for both groups of pages, have the lowest expected engagement, and underperform those expected values. While links can be top-performing posts, most often they are used as fodder to fill out post volume for pages. Sometimes links can go viral, often they don’t, but they are a low-lift way to fill out post volume for pages and build a following. Spending time finding high-performing and/or timely links to share is a good way to boost engagement.
Post length
Another way of looking at post format is to look at the length of the post text. In CAP Action’s social listening work, it has seen posts on either extreme do well on occasion—posts with zero text (usually a graphic with text or a Tweet screenshot) or posts with paragraphs and paragraphs of text (usually accompanying a link, photo, or video).
In Figure 2 below, a bar chart buckets all posts by the number of characters in the text, with the y-axis showing average engagement and the transparency of the bars representing frequency of posts. Interestingly, progressive-leaning pages see much higher engagement on posts with zero text, with nearly double the average engagement per post than conservative-leaning pages on a similar volume of posts (19,000 for progressive pages and 16,000 for conservative pages).
Figure 2: Average engagements by post text length and page ideological lean
Progressive engagement per post decreases as post text length increases before hitting a minimum of 101 to 500 characters, by far the most commonly used text length. Note that there is a potential self-reinforcing effect in these data: If high-engagement pages exclusively use shorter posts and lower-engagement pages don’t, that will skew the numbers slightly. Nonetheless, the trend is noticeable.
Conservative-leaning pages see roughly equal average engagement on any post text length between 1 and 500 characters, with a significant number of posts (40,000 to 60,0000) in each of the three text-length buckets. But there is also a small subset of posts with extremely long text, and within this subset conservative pages see much higher engagement—20,000 engagements per post for captions longer than 1,000 characters. There are only 700 posts among conservative pages and 1,100 among progressive pages that match those criteria, which is much too small a sample to be considered representative of a larger trend. However, it is interesting to see how pages can stretch the boundaries of what are often considered best practices on Facebook (short, attention-grabbing captions) while still generating good engagement.
Conclusion
Social media data does not lend itself to easy tracking or tidy conclusions; it is difficult, with limited public metrics, to isolate variables or compare performance across disparate audiences. For that reason, this report is limited in its scope, looking at only the past six months of 2021 among a defined subset of public pages and not trying to draw conclusions for the entire platform. Nonetheless, we hope that the findings written here are useful, both for observers seeking to better understand what content resonates on Facebook and for practitioners seeking to improve their ability to educate the public about progressive policies through digital channels.