Which of our posts actually work, and what should we make next?
Tricks with the can get watched. Posts about the viewer’s friends get shared. We sorted every Instagram and Facebook post from one summer campaign by what was on screen, then scored each kind against the brand’s own account numbers.
1 of 7
Use the arrow keys, or swipe.
Why they asked.
A hard seltzer brand ran a summer campaign on Instagram and Facebook: 39 Instagram posts and 29 Facebook posts over 13 weeks, from mid-June to mid-September. The social team had a feel for what was working. They didn’t have a count.
The platform dashboards ranked posts one at a time. The team needed a ranking of ideas: which kinds of post to brief again, which to try with a new angle, and which to stop paying to produce. A single post is a one-off. A kind of post is a line in next year’s budget.
The next content plan was being written, and the team wanted it built on what people watched and passed on to friends.
How we answered it.
Their own data
The brand’s own social export and posting calendar: 68 posts across Instagram and Facebook over 13 weeks.
What people say and do
How people responded to each post: what they watched, what they shared, and what they sent on to friends.
Pull every campaign post from the brand’s own export
We took the brand’s social media management export for the campaign window and kept every campaign-labeled post: 39 on Instagram and 29 on Facebook.
Watch each post and sort it by what is on screen
Every post was watched and sorted into a kind of post, which we call an archetype: hosting hacks, can shenanigans, workday escape, unconventional hangs, and an “other” bucket. When two or more “other” posts landed together on one subject, we named a candidate archetype. Two came out of it: posts about a big soccer tournament (three posts) and 3D animation (two).
Score each kind against the account’s own average
For each archetype we took median views, median shares and median share rate, and compared them with the period average: 8,148 views and 36 shares per Instagram post, and 3,230 views and 1.2 shares per Facebook post.
Keep the platforms apart
Instagram and Facebook numbers were never added together. Sixteen Facebook posts shared creative with an Instagram post and were watched on Facebook to confirm it was the same video at the same length. Three Facebook photo posts couldn’t be fetched, so they took their Instagram twin’s label on an exact caption match.
Check whether one post is carrying a whole kind
For every archetype we measured how much of its views and shares came from its single best post. Where one post did most of the work, the report says so and shows the median without it.
Write three lines: keep, test differently, stop
The report ends with what to make more of, what to try with a new frame, and what to stop. Each read names something you can see on screen, so a creative team can brief it the next morning.
Everything we found.
2.4×
Hosting hacks get watched, on both platforms
The four Instagram posts that showed a trick with the can had a median of 19,188 views against an account average of 8,148: 2.36 times. On Facebook the same kind led again, with a median of 4,509 views, 1.4 times that platform’s average. The three biggest show the can frozen, disguised or turned into dessert, and each one shows a step a viewer could repeat at home that night. They get watched more than shared: a median of 26 shares, under the account’s 36.
51 per 1,000
The most shared post was about the viewer’s friends
One Instagram post about the end of a big tournament, when everyone goes back to work, took 409 shares on 8,000 views. That is 51 shares for every thousand views, and 11 times the Instagram average for shares. Nothing else on the account was above 12 per thousand. The can was in hand and never pitched.
29,864
A moment people were in can travel without the product
A nine-slide carousel about what people would miss about the tournament had no can in any slide and did 29,864 views, second on Instagram. With the share leader and one weaker post (2,121 views, 4 shares), the three tournament posts ranked first by shares and second by views on Instagram, at a median of 8,000 views and 46 shares. On Facebook the subject did 739, 598 and 415 views, and the carousel never ran there. Three posts, so directional: the test is whether the next big shared moment does the same.
11 posts
Can shenanigans is the workhorse
Eleven Instagram posts where the can is the character, with a median just under the account average (6,107 views, 0.75 times) and three posts in each top ten. Its best post, a photo built around the can’s expiration date, did 25,657 views. When the can is doing something on screen, it also gets shared: the top product posts by shares took 138, 69 and 43.
77%
One post made a whole archetype look good
Workday escape had seven Instagram posts and a median of 3,962 views. One of them, a sign held up from a balcony with a single line of text that reads with the sound off, took 45% of the archetype’s views and 77% of its shares, with 201 shares. Without it the median falls to 3,119, and the archetype is the second weakest on the account. The hit was a permission gag with a sight joke. The posts that played “quit your job” straight, in a hammock or at a laptop, stayed under 2,300 views each.
1.2
Facebook watches. It doesn’t share.
Across 29 Facebook posts the average was 1.2 shares per post, and 24 of the 29 got zero or one. Facebook does watch: the slushie how-to did 37,836 views there, 11.7 times the Facebook average and the biggest post on either platform, with zero shares. Only one Facebook post reached double-digit shares, a still image of cans in hot dog buns, at 11,388 views and 12 shares. It never ran on Instagram. The tournament post that took 409 shares on Instagram took one on Facebook.
0.3×
An unusual place with nothing happening in it is the weakest idea
Unconventional hangs, the can in an odd location, had a median of 2,415 Instagram views, 0.3 times the average, the weakest named archetype on the account. On Facebook the same kind had a median of 418 views and one share across the whole archetype, and a single animal post carried 63% of its views.
175
3D animation sat at the bottom of the account
Two animated posts did 2,103 and 175 views, a median of 1,139 and 0.14 times the average: the lowest on the account. Neither had a moment where the product does something on screen. Two posts is directional, but nothing here argues for more.
80%
Strong archetypes still lean on one post
On Facebook the slushie took 80% of the hosting hacks views; the other three posts did 6,832, 2,186 and 686 views. The hot dog bun photo took 47% of can shenanigans views and 60% of its shares there. On Instagram the top post took 44% of hosting hacks views, 28% of can shenanigans and 48% of unconventional hangs. A kind of post with one giant hit is one idea that worked. Brief it as that idea.
41
Some winners came from outside the plan
Four Instagram posts outside the planned archetypes made a top ten. Beyond the two tournament posts, a coworker asking “what’d you do this weekend”, answered with a cut a second (a bar, cans, a wizard costume), took 41 shares and ranked eighth by shares. A single person walking out of a gas station with a 12-pack, no friends and no text, did 10,364 views and ranked tenth by views. Next year’s archetypes hide in the “other” bucket.
What we told them to do.
- Keep: hosting hacksThey lead on views on both platforms. Make more of the physical trick with the can that someone could copy tonight, like the slushie.
- Test: a friend-group frameShares came from a shared moment or a shared friend, where the next move is to send it to someone. Run can shenanigans and hosting hacks with a friend-group frame on top, and watch the share rate.
- Stop: 3D animationTwo animated posts sat at the bottom of the account, with no moment where the product does anything.
- Stop: workday escape as writtenKeep the permission idea from the balcony sign. Drop “quit your job” played straight.
- Judge Facebook on viewsWith 24 of 29 posts at zero or one share, a Facebook share rate ranks archetypes on single shares. Use views there.
- Name what is on screenEvery read in the report describes something you can see in the post. A creative team can brief that. They can’t brief a theory.
What any brand can take from this.
Sort by what happens on screen before you rank anything
A dashboard ranks posts. A budget pays for ideas. Group your posts by what happens in them, compare the groups by their median, and one lucky hit can no longer carry a weak idea into next year’s plan.
Views and shares are two different jobs
In this account the most watched kind of post was not the most shared kind. Decide which job a post has before you brief it, then judge it on that number.
A share needs someone to send it to
The posts that got passed on were about the viewer’s friends or a moment they had all been in. If you want shares, write the post so the obvious next move is sending it to a specific person.
What this looks like for your brand.
Try it out for freeCould Arcus run this on our own social accounts?
Yes. You give us your social media management export, or access to the tool, for a campaign or a quarter, plus the labels you use for campaigns. We watch every post, sort it, score it against your own account average and send a few slides: the answer on top, a scorecard by kind of post, and what to keep, test and stop.
We post on TikTok and YouTube too. Does the method change?
No. It works the same on every platform: sort by what is on screen, score each kind against that account’s own average, and keep the platforms apart. We can also line your posts up with public conversation about the brand and with your own numbers, like sales or sign-ups, so you can see which content shows up next to results. Showing cause takes a test.
How do we start?
Try it free. Pick a question and we answer it with your own numbers, then walk you through it in 30 minutes. No card. “Which of our posts work?” is a common place to start, and we tell you when to expect the answer before we begin. Once everything is connected, you can ask it again every quarter, with your sales or sign-ups in the same picture.
The numbers behind it
39 Instagram posts and 29 Facebook posts over 13 weeks, from the brand’s own social export.
Sources: The brand’s own social export, The brand’s posting calendar
the account’s average views for one type of post: a trick with the can you could copy tonight
How we checked, and what this doesn’t prove
Every number comes from the brand’s own social export. Every post was watched and sorted by hand. Small groups are marked directional.
- Small groups. Hosting hacks is four posts, the brief’s minimum for a read that is not directional. The tournament archetype is three posts and 3D animation is two, so both are directional. Every Facebook archetype is eight posts or fewer.
- Campaign-labeled Instagram posts start in mid-July, so the Instagram window is nine weeks. Facebook posts start in late June.
- Instagram gives no share count to outside tools. Shares come from the export’s total shares field and can’t be checked against the platform.
- This shows what got watched and shared. It doesn’t show what any post did for sales. That takes lining posts up with your own numbers, or a regions-on, regions-off test.
Names of the brand, its campaigns and its titles are removed. The numbers are exactly as delivered.
Download this report as a PDFQuestions people ask
More in the full FAQ, or ask us on a call.
What is a content archetype?
A repeatable kind of post, defined by what happens on screen: a trick with the product, the product as a character, a joke about a moment everyone shared. Sorting posts into archetypes lets you compare ideas, which is what you brief and pay for.
Why use median views instead of average views?
One viral post can drag an average far above what a typical post in that group does. The median is the middle post, so it shows what you can expect the next time you make that kind of post.
How many posts do you need for a reliable read?
In this report the line was four posts per kind. Below that, a finding is marked directional: worth a test, not enough on its own to move budget.
Can a content report tell me which posts drove sales?
Not from social numbers alone. Views and shares tell you what people watched and passed on. To connect content to sales, tickets or sign-ups, Arcus lines up posting dates with your own numbers, and runs a regions-on, regions-off test when you need proof of cause.
More real answers
Your questionis next.
Pick the question you’re least sure about. We answer it with your own numbers and what people say online, then walk you through it in 30 minutes. Free, no card. Once everything is connected, answers like this one keep coming.