The essentials
- Lever number one: the hook. The first line, or the first two seconds, decides 60 to 80% of the final engagement; that is where most of the work pays off.
- Formats: in 2026, carousels and short video get 1.5 to 3 times the engagement of a single image on Instagram and LinkedIn.
- Conversation: replying to every comment within the hour frequently doubles a post's number of comments and extends its distribution.
- The method: one change at a time, compared with the 90-day median over at least five posts, before drawing a rule from it.
Engagement is not obtained by asking for likes: it comes from precise choices about the hook, the format, the structure, the call to action, the timing and the conversation that follows. This page ranks these levers by observed effect and gives the protocol for testing them on your audience, the only one that counts. The definition and the benchmark figures are in engagement: definition, calculation and 2026 rates.
The seven levers, from strongest to weakest
| Lever | What it changes | Observed effect | Effort |
|---|---|---|---|
| 1. Hook | First line of the text, first two seconds of the video, first slide of the carousel | Determines whether scrolling stops; up to x2 on engagement, all else being equal | Poor |
| 2. Format | Carousel, short video, long structured text depending on the network | x1.5 to x3 compared with a single image | Moderate |
| 3. Topic rooted in a customer question | Immediate usefulness, saves, shares | Qualified engagement well above offer-led content | Low if the questions are collected |
| 4. Conversation | Quick reply to every comment, follow-up question | Comments often doubled; distribution extended | Daily, 20 minutes |
| 5. Single call to action | One request only, precise, consistent with the content | +20 to +50% on the requested action | Poor |
| 6. Time of posting | Posting when the audience is active, never when nobody can reply | +10 to +30% on early interactions | Poor |
| 7. Styling | Readable text, subtitles, contrast, a human face | Real but secondary effect | Moderate |
The order often comes as a surprise: design, which teams spend the most time on, comes last. The hook and the topic, which cost the least, come first.
Working on the hook
On LinkedIn and Facebook, only the first line is visible before the “see more”; on Instagram, the first 125 characters; on TikTok and Reels, the first two seconds and the on-screen text. The hooks that work:
- a precise, unexpected figure (“62% of our quotes start with an Instagram message”);
- a named mistake (“the mistake that costs a month's delay on a site”);
- a question the customer actually asks themselves (“should you replace your boiler before winter?”);
- a clear-cut opinion on an industry topic;
- a “before / after” or a concrete result.
What does not work: starting with the company name, with “we are delighted”, with a date, or with an empty formula. Write three hooks per post and keep the most concrete one.
Choosing the format to suit the network
In 2026, the most engaging formats per network are stable: on Instagram, Reels of 15 to 45 seconds and carousels of 6 to 10 slides; on LinkedIn, structured texts of 800 to 1,300 characters, PDF carousels and short vertical video; on Facebook, Reels and posts that ask a question; on TikTok, straight-to-camera video with a clear lesson. The video recommendations are in video on social media: formats and lengths. The expected tone also differs; see adapting your tone to each network.
Keeping the conversation going
Conversation is the most under-used lever. Three practices:
- Reply to every comment in the hour following the post, then during the day. Each reply counts as a comment and restarts distribution; on LinkedIn, a reply that adds something often triggers a new wave of impressions.
- Follow up with a question in the reply (“have you come across this?”) rather than with a closed thank-you.
- Comment elsewhere: 15 minutes a day of useful comments on posts from accounts your target follows bring profile visits and qualified followers, with no budget.
This day-to-day work is community management in the true sense; an account that posts without conversing plateaus.
What to stop doing
| Practice | Why it hurts | Replace with |
|---|---|---|
| Explicitly asking for likes or shares | Penalised by Meta (engagement bait), perceived as artificial | An open question linked to the content |
| Outbound link in the main text on LinkedIn and Facebook | Reduced distribution for posts that send people off the platform | Link in the first comment or in a story, self-contained text |
| Posting the same image on every network at the same time | Unsuitable format, shared audience saturated | A version per network, staggered posts |
| Engagement pods and bought interactions | Off-target audience, degraded distribution in the medium term | Genuine outbound engagement |
| Five calls to action in one post | None is followed | One only, consistent with the objective |
| Three months of offer posts | Engagement falling, audience unfollowing | 60 to 70% useful content; see evergreen or trending content |
The testing protocol
General recommendations give a starting point; only your audience decides. The protocol: change a single variable (hook, format, time, length), over at least five posts, and compare engagement by reach with the median of the account's last 90 days. A gap of less than 20% is not conclusive. Record every test and its conclusion; after a quarter, you have a list of rules specific to your account that is worth more than any guide. The steps for reading them are detailed in analysing your social media performance.
How GreenRed helps
Rather than juggling several tools, GreenRed's Social media module brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.
Frequently asked questions
Which lever is the most effective for increasing engagement?
The hook: the first line of the text or the first two seconds of a video decides whether scrolling stops, and therefore everything else. A precise figure, a named mistake or a real customer question does better than any graphic optimisation. Then comes the choice of format, carousel or short video.
Does asking for likes or shares still work?
No. Meta classes these explicit requests as “engagement bait” and reduces their distribution; the audience perceives them as artificial. An open question linked to the content, or an invitation to save a useful resource, gets more interactions without any penalty.
Should you reply to every comment?
Yes, including the shortest ones, and if possible within the hour following the post. Each reply counts as an interaction and restarts distribution, particularly on LinkedIn and Facebook. Follow up with a question rather than a closed thank-you to keep the conversation going.
How long does it take to see engagement improve?
A better-chosen hook or format has an effect from the very first posts. A lasting improvement in the median takes two to three months of regularity, daily conversation and recorded tests. Always compare with the median of the last 90 days rather than with an isolated peak.