by Jes Kirkwood, B2B SaaS Marketer & Growth Strategist

Social Media Marketing Insights: What We Learned From 942 Facebook, LinkedIn, and Twitter Posts in 2017

If you’re looking for data-driven social media marketing tips, you’ve come to the right place.

We recently analyzed 942 social media posts to discover what works in 2017. Some of what we discovered will confirm what you already know, but some of it may surprise you. (It certainly surprised us!)

Let’s dive in.

Facebook Marketing Insights

In the first quarter of 2017, we published a total of 132 Facebook posts. Here’s what we discovered:

Screen Shot 2017 04 20 at 10.21.51 AM 1

How character count impacted post performance

Longer Facebook posts (201-500 characters) delivered more impressions and helped us reach more users. Within this range, posts with 401-500 characters performed best, reaching 120.37% more users and delivering 123.97% more impressions than the average post.

Shorter posts (101-300 characters) attracted more clicks, but delivered less reach and fewer impressions, leading to higher engagement rates. Within this range, posts with 101-200 characters performed best, delivering 94.04% more clicks and 13.33% higher engagement rates than the average post.

Our worst performing posts were those with under 100 characters. These post delivered 11.45% fewer clicks and reached 7.80% fewer users.


Like everything else, your best bet depends on your goals. What do you want to achieve via social media?

If you’re not sure, check out this helpful framework:

pasted image 0 12(Source: Simply Measured)

If your primary concern is brand awareness, prioritize impressions and reach by crafting longer posts. If you instead want to generate demand or drive conversions, prioritize engagement and clicks by publishing shorter posts. For a balance of the two, aim for Facebook’s sweet spot: 201-300 characters.

When our Facebook audience is most active and engaged

Posts published on Mondays, Thursdays, and Saturdays got the best results. Saturday posts outperformed every other day, receiving 96.36% more clicks, 40% higher engagement, and 66.62% more reach than the average post. Monday posts came in second place, earning 39.73% more clicks, 39.56% more impressions, and 43% more reach. Thursday posts also performed well, with 10.37% higher engagement and 37.63% more clicks on average.

Posts published on Sundays got the worst results, attracting 41.49% fewer clicks, earning 38.28% less impressions, and reaching 39.96% fewer users on average.

Early morning posts earned more clicks and engagement. Posts published at 9:41 PM PT (our worst time) earned 35.10% lower engagement rates and attracted 30.06% fewer clicks. Posts published at 7:17 AM PT (our best time) 46.36% better engagement rates and attracted 45.52% more clicks.

Early evening posts earned more reach and impressions. Posts published at 1:08 PM PT (our worst time) reached 24.40% fewer users and earned 24.62% fewer impressions. Posts published at 6:41 PM PT (our best time) reached 22.89% more users and earned 18.15% more impressions.


Notice how the days and times that work for our audience differ slightly from the standard best times to post:

pasted image 0 13(Source: CoSchedule)

Like most audiences, ours is more engaged on Thursdays and Saturdays. But unlike most audiences, ours is also engaged on Monday and is not engaged on Sundays. What does this mean?

Well, it confirms that standard best times to post on social media aren’t one-size-fits-all. Every audience is different, which is why it’s imperative that you run your own tests and analyze your own social data to see what works in your specific context.

Having said that, you can (and should) test early morning and early evening posts, as many people are active on social media during their commutes to and from work.

Which post types do (and don’t) work on Facebook

Our top three post types were videos, images, and links:

1. Video posts

Video posts reached 429.22% more users, received 401.58% more impressions, and earned 317.25% more clicks than link posts. In terms of engagement, video posts perform 9.52% worse than link posts but 18.75% better than image posts.

2. Image posts

Image posts reached 28.62% fewer users, received 21.12% fewer impressions, and earned 31.26% fewer clicks than video posts. At the same time, image posts reached 277.78% more users, received 295.67% more impressions, and earned 186.82% more clicks than link posts. In terms of engagement, image posts attracted a 20% lower engagement rate than video and link posts.

3. Link posts

On average, link posts reached 77.95% fewer users, received 77.71% fewer impressions, and earned 71.59% fewer clicks than image and video posts. Despite this, link posts received 20% higher engagement than image and video posts.

Status updates received little to no impressions or clicks, reached only a handful of users, and delivered dismal engagement rates.

Screen Shot 2017 04 20 at 10.58.23 PM 1


Visual media is intrinsically more engaging and has become the medium of choice for many viewers, which is one reason videos and photos perform well on Facebook (and why textual status updates don’t).

Check out this infographic to learn more about why visuals grab and keep our attention:

pasted image 0 10(Source: Column Five Media)

The fact that link posts work well for Autopilot may indicate that our audience is hungry for the longer form content that they click through to: how-to posts and guides for getting started with particular type of marketing, deep dives into specific growth tactics, and real-life case studies that not only reveal potential solutions to their pain points, but also the emails, in-app and on-site messages, text messages, and postcards that helped them achieve results.

Do #hashtags work on Facebook? What about emojis?

Surprisingly, posts without #hashtags performed better than those with. These posts earned 42.31% more clicks, 20.67% greater reach, and 15.83% higher engagement on average.

Posts without emojis also outperformed those with. On average, these posts attracted 49.14% more clicks and a 13.22% higher engagement rate.


After digging into our social data, we can confirm what Buffer’s Kevan Lee and BuzzSumo’s Steve Rayson have already claimed: Hashtags don’t work on Facebook.

Since they don’t improve engagement, we recommend getting rid of them! It’ll declutter your copy and improve your audience’s viewing experience.

Twitter Marketing Insights

During the same period, we published 714 Twitter posts. Here’s what they revealed:

Screen Shot 2017 04 20 at 10.21.39 AM 1

Which post types do (and don’t) work on Twitter

Our top two post types were polls and images:

1. Polls

Twitter polls were our best performing tweets by far. With 448.46% more impressions and 2015.87% more engagements than the average post (including 2400% more retweets and 748.21% more likes), this post format easily outperformed image, GIF, and link posts.

2. Image posts

Image posts delivered 234.76% more engagements on average, including 283.04% more likes, 383.33% more replies, and 137.50% more URL clicks. In terms of engagement, image posts attracted 47.79% higher engagement than the average post.

Our worst performing post type was link posts without images, which earned 20.16% less impressions and 39.71% lower engagement than the average post.


Visual content trumps written content on Twitter, similar to Facebook. But on Twitter, where you have the ability to poll your audience, interactive content seems to trump visual content.

That’s not to say that our visuals didn’t get plenty of love. In addition to getting lots of likes, they were effective conversation starters. But in terms of engagement, our polls blew our photos out of the water. Plus, they were retweeted like crazy, which amplified our reach.


  • If you haven’t already, give polls a try!
  • If you want to get people talking, try posting more visuals.
  • For optimal engagement, avoid publishing posts without a visual element.

When our Twitter audience is most active and engaged

Posts published on Wednesdays and Saturdays got the best overall results. Saturdays (again) outperformed all other days, earning 26.92% more impressions, 66.67% more replies, 38.89% URL clicks, and 34.82% more likes than the average post. Wednesdays also performed well, with 25.32% more impressions, 37.50% more URL clicks, 31.25 more likes, and 31.81% more retweets per post, leading to an average 14.16% higher engagement rate.

Sundays were the best day for amplification, delivering 70.45% more retweets on average. Unfortunately, they were also the worst day for most other metrics, earning 12.47% fewer impressions and 22.15% fewer total engagements, including 49.33% fewer likes.

Top performing posts spanned the day. Out of the 48 times we tested in the first quarter of 2017, posts published at 7:45 AM, 12:24 PM, 3:38 PM, and 9:45 PM PT performed consistently above average. Posts published at 12:24 PM (our best time) delivered the highest impressions (314.06% more impressions, on average). Posts published at 3:38PM earned 303.37% more impressions and 151.89% more engagements than the average post, including 316.67% more clicks.

Other post times require more testing, but showed strong results early on: Posts published at 4:14 PM boasted 93.38% higher engagement rates and 247.22% more clicks than the average post. Posts published at 8:48 AM delivered 240.91% more retweets than average, while posts published at 12:58 PM got the most likes. Finally, followers were 1566.67% more likely to reply to a tweet at 9:41PM than any other time. We’ll continue to test these times in the second quarter of 2017.

Our worst times to post on Twitter were 7:26 PM, 11:06 AM, and 1:46 PM PT.


In line with other research studies, our audience is most engaged on Wednesdays and Saturdays. No surprise here.

Noon to 1:00 PM local time is one of the most popular times to tweet, according to Buffer’s data. Some influencers recommend not posting during that hour, because as more companies tweet, your post inevitably gets buried. But Twitter’s algorithm is set up to show people the best tweets first, so as long as your posts are engaging, they’ll be shown to your audience.

Posting during this period has paid off for us. In fact, 12:24 PM is our best post time! So try it out. After all, it’s a popular time for good reason: Many people browse social while eating their lunch, just looking for something to catch their attention.

How character count impacted tweet performance

Longer tweets outperformed shorter tweets across almost all metrics: impressions, retweets, replies, likes, and engagement rate. In fact, tweets with 125+ characters earned 88.80% more impressions, 213.64% more retweets, 283.33% more replies, and 208.93% more likes—leading to a 24.26% higher engagement rate than posts with fewer than 125 characters.

Having said that, there was one exception: URL clicks. Tweets with 125+ characters delivered 68.05% fewer URL clicks than average.

Overall, tweets with fewer than 50 characters performed worst. On average, they delivered 25.76% fewer impressions, 38.76% lower engagement rates, 61.36% fewer retweets, 50% fewer replies, and 28.57% fewer likes.


On Twitter, there appears to be a positive correlation between character count and engagement. With every added character, comes additional engagements—especially likes and retweets. For this reason, aim to write tweets with a minimum of 100 characters for optimal engagement. But remember: Tweets that exceed 125 characters receive less clicks. So, if you’re trying to drive traffic to your site, keep it under under that limit.

Having said that, it’s super important dig into your own social data to see what length works best for your audience. This is especially important because… what works for Autopilot is abnormal. Most experts recommend keeping tweets between 70 and 100 characters—the opposite of what works for Autopilot. The thing is, optimal tweet length varies greatly by company. For example, Buffer’s optimal tweet length is 80 to 120 characters. Hence why it’s important to figure out what works for your audience.

Do #hashtags work on Twitter? What about @mentions?

Unsurprisingly, posts with #hashtags outperformed those without. These posts earned 200% more replies, 33.34% more retweets, and 26.79% more likes than their counterparts, resulting in a 11.12% higher engagement rate on average.

Posts with @mentions performed significantly better than those without. These posts earned 111.08% more engagements, including 600% more replies and 105.41% more retweets. Interestingly enough, these posts also delivered 106.67% more URL clicks than posts without @mentions.


Hashtags had a positive, but relatively minor, impact on our tweet performance.

Having said that, I know from experience that after a few months have gone by… many of those likes suddenly disappear. The reason behind this is that some people use automated bots to engage with other Twitter users. These bots locate specific hashtags, then like the corresponding tweet on the user’s behalf. For this reason, using hashtags can lead to a false lift in performance.

Recent studies have suggested that hashtags are worthless. For this reason, I don’t typically recommend using hashtags.

Contextual hashtags are one exception. Using branded hashtags (e.g., for a conference) can be an effective way to help users eliminate the clutter and hone in on one conversation.

LinkedIn Marketing Insights

In the first quarter of 2017, we published 96 LinkedIn posts. Here’s what we found:

Screen Shot 2017 04 20 at 10.21.59 AM 1

Which post types do (and don’t) work on LinkedIn

With 122.60% more impressions, 250% more clicks, and 400% more interactions, images outperform links on LinkedIn by a landslide.

Because there are only two different types of posts on LinkedIn, we also analyzed what types of content perform well. Our top three include:

1. Product updates

These posts earned 47.97% more impressions and 312.37% more interactions than the average post.

Screen Shot 2017 04 21 at 7.32.23 AM 1

2. Posts promoting job openings

These posts attracted 89.53% more clicks than average, resulting in a 152.17% higher engagement rate.

Screen Shot 2017 04 21 at 7.36.17 AM 1

3. Growth case studies

These posts attracted 68.48% higher engagement than other content types.        

Screen Shot 2017 04 21 at 7.34.37 AM 1

With an average of 44.64% more impressions, quote visuals and eBook promos also performed well.

Screen Shot 2017 04 21 at 7.28.50 AM 1


Some people mistake LinkedIn for being the stiff sibling or cousin of Facebook and Twitter. But on LinkedIn, we’ve found that visual content again beats out the more business-as-usual link posts. The takeaway here is to not limit yourself. Experiment with different types of content, including photos and videos, to see what resonates with your audience. LinkedIn may be a business-focused social network, but its users are still people… and people find visuals engaging.

Unlike on Facebook and Twitter, many (and perhaps most) of your LinkedIn followers are your customers and employees. For this reason, content centered around your company—whether that’s job postings or product/service updates—tends to be well received. In other words, LinkedIn is an effective channel for communicating with your customers (as opposed to just leads).

The last thing to keep in mind about LinkedIn is that its users tend to be older and hold more senior roles. Case studies are popular with our LinkedIn audience because they focus on problems, solutions, and results. This content resonates with senior-level marketers, who are ultimately responsible for driving growth.

When our LinkedIn audience is most active and engaged

Posts published on Thursdays and Fridays got the best overall results. With 64.40% more impressions, 149.92% more clicks, 73.16% more interactions, and 43.47% higher engagement, Fridays outperformed every other day of the week. Thursdays came in second, earning 25.60% more impressions, 21.48% more clicks, 64.95% interactions, and 26.09% higher engagement than the average post.

Posts published on Tuesdays performed the worst, attracting 34.95% fewer impressions and 57.94% fewer clicks than average. Similarly, Wednesdays had the fewest interactions per post, and Sundays earned the lowest engagement rates.

In the first quarter of 2017, we tested post times between 10:00 AM and 1:00 PM PT. Posts published at 12:42 PM performed the best, with 85.97% more impressions, 570.10% more interactions, and 44.4% more clicks than the average post, resulting in a 56.52% higher engagement rate.

Posts published at 10:11 AM performed the worst, earning 22.57% fewer impressions, 42.78% fewer clicks, 31.44% fewer interactions, and 20.65% lower engagement on average. (On a related note, the number of interactions per post climbed steadily as we tested later and later post times.)


Like other studies have shown, posting around lunchtime works. Our mid-morning posts largely went ignored, but when we shifted our publish time to 12:42 PM, we saw an immediate uptick in engagement. Right on!

pasted image 0 11(Source: CoSchedule)

With that said, the days of the week when we see the most engagement are not in line with industry standards. Most companies report better performance on Tuesdays, Wednesdays, and Thursdays, with dismal engagement on Mondays and Fridays. For Autopilot, Fridays actually delivered the best results.

The lesson? Even if you start off following best practices, experiment to find what days and times work best for your audience.

How character count impacted post performance

Like the other networks, lengthier LinkedIn posts (201-500 characters) tend to earn more impressions—to a certain point. It’s almost like an inverted funnel:

Inverted funnel

Below 200 characters and above 500 characters, impressions begin to decrease.

Longer posts also earn more clicks, interactions, and engagement. Within this range, posts with 201-300 characters performed the best, delivering 257% more impressions, 545.31% more clicks, and 157.73% more interactions.

Screen Shot 2017 04 21 at 8.54.08 AM 1


Our ideal character length on LinkedIn is (again) not in line with industry standards. Whereas OkDork found that LinkedIn 40- to 49-character LinkedIn posts got the most views, we found that 201- to 300-character posts got the most impressions. That just goes to show you how different it can be from company to company!

Do #hashtags work on LinkedIn? What about emojis?

LinkedIn isn’t known for being a network where hashtags are appropriate, which is why we were shocked to discover that posts with #hashtags earned 24.16% higher engagement and attracted 93.70% more impressions, 208.57% more interactions, 135.52% more clicks than posts without hashtags.

Screen Shot 2017 04 21 at 7.03.00 AM 1

We were also surprised to learn that posts with emojis perform better on LinkedIn than those without, earning 52.97% more impressions and 284.62% more interactions on average.

Screen Shot 2017 04 21 at 7.06.55 AM 1


Hashtags and emojis improved the performance of our LinkedIn posts, which is quite surprising.

Emojis may simply stand out because so few people use them on LinkedIn. If that’s the case, using emojis may be an effective way to catch the attention of your followers. Try adding them to a few posts and see what happens.

The thing is, hashtags are no longer “linked” on LinkedIn. So why are they driving engagement?

I can only speculate that either A) users got used to searching for content via hashtags on LinkedIn and continue to do so (which I doubt), or B) LinkedIn’s algorithm is wired to boost posts with hashtags.

Either way, I’d use hashtags cautiously and keep an eye out for any changes that make their way down the pipe.

What other variables impact performance?

We were able to dig up two more insights by analyzing our Q1 2017 social media post data:

1. Catchy copy performs better than clear copy

What’s more effective: Copy that’s clear (i.e., easy to understand) or copy that’s catchy (i.e., more engaging)? It’s an age-old debate between copywriters.

For Autopilot, catchy copy earns 49.14% more impressions, 187.95% more clicks, and 58.54% higher engagement than clear copy.

2. Including numbers in your copy improves performance

We all like seeing concrete results, such as how much revenue a campaign produced, how fast a startup grew, or how many tips and tricks an article covers.

We suspected including numbers in our copy might give us a boost in performance, so we segmented our posts into two groups and ran the numbers. Turns it, we were right! While posts with numbers earned just 15.22% more impressions than those without, they received 46.01% more clicks, 111.08% more interactions, and 42.9% higher engagement.

Are these findings in line with your social media results? Let us know in the comments.

Author Jes Kirkwood, B2B SaaS Marketer & Growth Strategist

Jes Kirkwood is a results-driven marketing strategist with expertise in both demand generation and revenue expansion. She develops minimum viable playbooks that drive predictable growth for early stage B2B SaaS startups. Follow her work (and her journey) on Twitter.

  • Your journey to sky-high growth starts here.

    Subscribe to join a savvy crew of journey marketers and learn the latest trends, tactics, and tools.

    • Guy Marion

      Mind blown- gr8 post @Jes! ??

    • fio @ Hotjar


      Question from a fellow content marketer who may or may not want to do something similar in the
      future — our of curiosity, how long did it take to collect + analyse all this data?

      • Jes Kirkwood

        Hey Fio,

        Great question! In short, longer than a typical blog post ?

        If you already collect the data on a monthly basis, it doesn’t take long to aggregate it once per quarter. Let’s say… 30 mins?

        The analysis took a bit longer, though. For each social network, I spent a few hours working with the data to reveal core insights related to each variable. All together, I probably spent 6-8 hours on analysis and another 2-4 collecting secondary research.

        As for the writing, it really depends on how fast you write. The post is 3,300+ words, so if you average about 250-300 words per hour (for writing/editing/polishing), you’re looking at 10-15 hours.

        In total, that comes to about 24 hours. But I spread the work out over 3 weeks, which helped make it manageable. It might seem like a lot, but those insights will also help inform changes to our social strategy, so it’s a good investment.

        Hope that helps! ?

    More from Marketing
    How To Rally Around The Customer And Skyrocket Growth (Webinar Recap)

    How To Rally Around The Customer And Skyrocket Growth (Webinar Recap)

    Today’s webinar featured Slack Director of Marketing Brad Morris, VentureBeat Analyst Wendy Schuchart, Autopilot CEO Mike Sharkey, and Autopilot CMO  …

    Get Big, Fast with Strategic Partnerships

    Get Big, Fast with Strategic Partnerships

    Over the last four years, the food and beverage industry has grown an average of 2.3% annually. Only 0.1% of  …

    A Decade’s Worth of Insights From Marketo’s First Sales Hire (AKA Autopilot’s New VP of Sales)

    A Decade’s Worth of Insights From Marketo’s First Sales Hire (AKA Autopilot’s New VP of Sales)

    We’re happy to announce that Scott Edmonds is Autopilot’s new VP of sales. He’s a veteran of the marketing automation  …