Most of what gets written about the Facebook algorithm is wrong, or at least useless. Posting time does not meaningfully matter. Hashtags do not move the needle the way they used to. Caption-tweak hacks are noise. What actually drives whether your post gets seen by 200 people or 200,000 is a small set of mechanics that have been publicly described by Meta engineers and confirmed by years of outlier data. This post walks through those mechanics and what they imply for how you should actually be posting.
If you run a Facebook Page and care about reach, this is the framework worth understanding — and at the end we will get into why doing this consistently is the actual hard part, and how One Two Three Post Pro closes that gap.
What the algorithm actually does when you post
Facebook (and Instagram, and TikTok, and the rest) has exactly one business goal: keep users on the platform as long as possible. The longer someone scrolls, the more ads they see. That is the whole equation. Every algorithmic decision flows from it.
To maximise scroll time, the algorithm acts as a matchmaker. It tries to put the post most likely to delight a given user in front of that user, while not delighting them so much they screenshot it and leave. When you publish a new post, the platform runs a multimodal analysis on it within seconds: computer vision over the image or video, audio fingerprinting on any sound, plus all the metadata it has — your caption, your Page history, the time and geography, whether the post mentions other accounts, whether it links externally. All of that rolls into a single internal representation of “what is this post about and who is likely to enjoy it” — call it a topic mapping.
Based on that topic mapping, the algorithm builds a prediction: which users currently active on Facebook are the best fit for this post. That prediction is a ranked list of every active user. The top of the list — roughly 200 people — becomes the initial sample group.
The 200-person sample group is where the post lives or dies
This is the single most important thing to understand about how Facebook works in 2026: most of the people who see your post first are not your followers. The algorithm deliberately seeds new posts to a sample group that is overwhelmingly weighted toward strangers — accounts that fit the topic-mapping profile but have not opted into your Page. The reason is calibration. Your followers are biased: they already like you, so their engagement signal is noisy. Strangers are honest. If a stranger watches, likes, and shares a post, that is genuine algorithmic gold. If 200 strangers shrug, the post is not actually as good as your follower count would suggest.
What happens next depends entirely on how that initial 200 responds:
- Strong engagement from the sample group → the algorithm boosts to roughly 2,000 next, then 20,000, then 200,000, scaling up in 10× steps until the data finally thins. This is the path that produces a viral post.
- Mixed engagement → the algorithm refines its fit-score, picks a fresh sample of 200 with a slightly different profile, and tries again. You might never see the second sample reflected in your view count because it happens within minutes of publishing.
- Weak engagement → the algorithm stops promoting almost immediately. The post stalls at “200 view jail” and never recovers, no matter how good it actually is.
This is why two posts on the same Page can have wildly different reach. The one that landed well with its 200-person sample scales to 50,000 views. The one that did not stays at 180.
The two levers you actually have
If the sample group is the choke point, there are only two things to optimise: help the algorithm find a tight, accurate sample group in the first place, and make sure that group has a strong reason to engage when it sees the post.
Lever 1: Narrow your topic and audience, ruthlessly
The algorithm builds its fit score for your fourth post by looking at how the first three performed. If post one was about tech, post two about health trends, and post three about politics, the algorithm has no idea what post four is about — so it builds a blended sample group made up of strangers from all three audiences. When post four turns out to be about health trends again, two-thirds of the sample group did not want it, the engagement comes back weak, and the post flops. Even though it was the right post for the right audience, the blended sample failed it.
The fix is unglamorous: pick one audience and one tight band of topics, and stay there for months. Every post you publish that is off-topic costs you the next several on-topic posts in algorithmic confidence. The discipline matters more than the topic. A Page that has posted about Edinburgh hidden gems three hundred times in a row gets a sample group of strangers who like Edinburgh hidden gems. A Page that posts about Edinburgh, then about generic UK travel tips, then about a personal life update, gets a blended sample group that engages weakly with all three.
On Facebook specifically, this discipline is hard because the natural temptation is to mix in a quick newsletter promo, a behind-the-scenes shot, or a “Just sharing this!” link whenever you remember. Each off-topic post knocks the algorithm’s calibration loose for the next several real posts. The cleanest fix is to commit ahead of time to a small set of content types and rotate through them mechanically, not opportunistically.
Lever 2: Make the sample group engage
Once the right 200 strangers are looking at your post, three metrics decide whether the algorithm boosts:
- Watch time and percent-completion for video; equivalent dwell time on text posts (how long the post stays on screen before the user scrolls past).
- Engagement rate — likes + comments + shares + saves divided by views. Comments are weighted heaviest because they cost the most user effort.
- Session share — what fraction of a given user’s 60-minute Facebook session is spent on your content. You cannot see this number in any dashboard, but Meta engineers have confirmed it is the most influential of the three. A Page whose readers stick around to read more of its posts gets boosted; a Page that gets a single like and then loses the reader to a competitor’s post does not.
Of those three, engagement rate is the one operators have most direct control over. And on Facebook, two formats consistently outperform on engagement rate:
- Native coloured-background text posts. Facebook gives these visual punch in the feed that a plain link post or even a photo post does not match. Coloured-background posts are statistically more likely to stop a scroller mid-flick, which improves dwell time, which improves the algorithm’s read of the sample-group response.
- Open-ended questions in the caption that invite replies. Comments are the highest-value engagement signal, and the easiest way to drive comments is to ask something the reader actually wants to answer.
If you want to push comments specifically, a few principles work reliably:
- Take a clear stance. Hedged, “on the one hand, on the other hand” captions produce nothing. People comment when they agree forcefully or disagree forcefully. Pick one.
- Lean slightly contrarian. If your stance is the consensus view, no one needs to argue. If it is a minority view, the majority feels compelled to push back.
- Amplify the framing. “This is the best way to cook pasta” gets a polite nod. “Every Italian grandmother is doing this wrong” gets an argument in the comments.
- Anchor around cult-loved brands, people, or ideas. Pre-formed opinions trigger faster engagement than novel categories. A post about Nike will out-comment a post about “shoes”.
- Drive emotion. Outrage, nostalgia, joy, surprise — any of them work. Neutral does not.
And on the link side: do not put external links in the post body. Facebook downranks posts that try to send users off-platform — the algorithm reads them as “this person wants the user to leave, which is the opposite of our goal”. Put the link in the first comment instead. The algorithm does not penalise comments the way it penalises link cards.
The gap between strategy and execution
None of the above is secret. Operators who care about Facebook reach already know these mechanics. The reason most Pages still fail is not strategy — it is execution. If you write out what running the strategy looks like on a Tuesday afternoon, the to-do list is brutal:
- Pick two or three content types and stay rigidly inside them for months.
- Post at least once a day, ideally for the next year, without missing a day.
- Rotate the types so you never post two of the same type back to back (three identical-type posts in a row reads to the algorithm as a stale account).
- Space posts out so they do not bunch up in a single hour (the algorithm reads bursty publishing as low-quality account behaviour).
- Use native Facebook formats wherever possible — coloured-background text posts, not link cards.
- Put any external link in the first comment, never the post body.
- Vary the question prompts so the engagement format does not feel scripted.
- Catch your own mistakes before publishing — an AI-named image filename, a duplicate caption recycled from three weeks ago, a brand mention you forgot is on the Facebook ban list, a winter post you scheduled in October that is about to fire in July.
Read that list again. That is roughly 90 minutes of weekly admin work, every week, forever, before you have written a single line of content. The reason most operators drift back into “post whatever, whenever” is not that they disagree with the strategy. It is that the discipline is incompatible with doing the rest of their actual job.
How One Two Three Post Pro automates the operational layer
One Two Three Post Pro is a WordPress plugin that sits on top of the free One Two Three Post scheduler. It is built around the assumption that you have a WordPress site with blog content and you need that content to keep your Facebook Page running on the algorithm-friendly cadence above. Walking through the list:
Three content types, automatically rotated
Pro auto-generates posts in three formats — only three — chosen specifically because they map to what Facebook’s algorithm rewards:
- Photo posts from your recent blog posts. The featured image plus a short caption that links the reader back to the article. The link goes in the first comment, not the body. This is the consistent-topic workhorse — every photo post is tied to your existing content, so the topic mapping the algorithm builds for your Page stays tight.
- Text questions as native coloured-background text posts. These are the engagement-rate accelerator. The plugin ships a starter pool of 20 open-ended questions you can edit, and falls back to Claude (if a key is configured) to write fresh ones when the pool runs dry. Open-ended questions reliably outperform plain photo posts on comment rate, which directly improves the 200-sample engagement signal.
- Newsletter teasers as native coloured-background text posts with the signup link in the first comment. Punchy curiosity-gap hooks like “Stories the guidebooks won’t tell you, fresh in our inbox”. Capped at 1 per day so the format does not start looking like spam.
You configure the ratio between the three (default 50/25/25 photo/question/teaser) and Pro picks weighted-randomly on each cron tick. The output is a steady stream of varied native Facebook posts on a narrow avatar — exactly what the algorithm uses to lock in a strong fit score.
Cadence held by an hourly cron
Pro’s content auto-fill cron runs hourly and tops your queue up to a target depth (default 12 scheduled posts in the next 24 hours). On a quiet hour it adds one new post; on full hours it adds nothing. The result is a steady cadence you do not have to think about. Set the target depth, walk away, come back in a week to find 84 posts have queued and fired automatically.
Type rotation + spacing validators
Two pre-publish validators directly address the algorithmic-confusion problem above:
- Type rotation blocks a queued post if the last N published posts were all the same type. Stops the “three photo posts in a row” scenario that flattens the algorithm’s fit score.
- Minimum spacing blocks a queued post if another post on the same Page fired within the last N minutes (default 60). Stops bursts that the algorithm reads as low-quality account behaviour.
Both are toggleable, configurable, and silent when nothing is wrong. When they do block, the post is held in pending with the reason logged so you can adjust or wait for the window to clear.
Five more validators catching the “stranger bounces” issues
The 200-view jail happens when the sample group sees something off about the post and skips it. Pro’s other validators catch the most common off-putting issues before the post fires:
- AI-generated image filenames (Facebook is increasingly aggressive about flagging AI imagery — easier not to invite the suspicion).
- Duplicate images via perceptual hashing — the algorithm sees the same picture three weeks in a row and downranks for repetition.
- Duplicate captions — same problem, different vector.
- Seasonal mismatches — a snow post in July reads as scheduled-and-forgotten content, which tanks the sample group’s reaction.
- Banned-word matches — keep brand-safe terms out of the caption before Facebook’s own filters quietly nerf the reach.
First-comment auto-detection
If a queued post’s caption mentions a recent blog post’s title, or matches a configured category keyword, or contains a curiosity-gap teaser phrase like “have you heard…”, Pro auto-fills the corresponding WordPress permalink as the first comment. Operator-set values are never overwritten. This keeps every post link-free in the body, where the algorithm wants it.
Editor pass every six hours
Six times a day, Pro re-runs the validator chain against everything scheduled to fire in the next 24 hours. If a validator would block at publish time, the post is moved to pending immediately so you see the problem hours before the fire window. Avoids the “scheduled post failed at 06:00 because I posted something similar last night” pattern.
The honest summary
The strategy for winning on Facebook in 2026 is well-known: post consistently on a narrow avatar with formats the algorithm rewards, give the 200-person sample group a reason to engage, and keep external links out of the post body. The hard part is sustaining that behaviour for the 12 months it takes to compound. Most Pages do not fail because the operator does not understand the algorithm — they fail because operating it manually is a part-time job, and the part-time job loses to whatever else is on the calendar.
One Two Three Post Pro reduces that part-time job to: install the plugin, set the queue target, pick your content-type ratios, walk away. Posts auto-generate from your blog. Validators stop the algorithmically-bad ones from firing. The editor pass catches problems before they happen. First-comment links keep Facebook from downranking you for sharing external URLs. The cadence stays steady whether you are at your desk that week or on holiday.
The plugin is distributed by email to paid subscribers of this publication. Full feature list is in the docs.