Skip to content
← All articles
Catch the Bad Crop Before Your Client Does

Catch the Bad Crop Before Your Client Does

Most agency revision rounds are not arguments about strategy. They are the client seeing the rendered post for the first time and reacting to a crop, a truncated sentence, or a preview image they did not expect. The idea was approved days earlier in a document that could not show any of it.

Review the render, not the doc.

In short

  • Clients reject renders, not ideas. If the first time they see the render is after it is scheduled, you have bought a revision round.
  • Every network silently reframes media and folds text: video gets 9:16, Instagram square or story tall, LinkedIn and Facebook wide, and microblogs almost 2:1.
  • Preview-first review costs one extra minute per post and removes the round trips that cost an hour, but it cannot catch feed variation, link cards or emoji rendering. Say so before the client does.

The two classics

The truncated caption is the first. Someone writes a caption where the payoff lands in the fourth line, the network folds after roughly two, and the post now reads as a fragment that ends mid-clause. On Instagram the visible slice is short enough that anything below about 125 characters is behind the more link. On LinkedIn it is longer, around 210, and still shorter than most people’s instinct.

The beheaded image is the second. A designer exports a 16:9 hero with the subject on the right, the post targets a vertical video slot, and the crop takes the subject’s head off. Nobody sees it because the upload succeeded and the thumbnail in the queue looks fine at 80 pixels wide.

Both failures share a shape: the post is technically valid, so nothing in the pipeline objects. Only a human looking at the render objects, and if your process does not put a render in front of that human before publishing, the objection arrives after.

What a skipped preflight actually costs

Put a number on the habit or nobody keeps it. Say a three-person agency produces forty posts a week across six clients, and one revision round lands on every fifth post. Each round is a message to the client, a small rewrite, a re-crop, a second approval and a rescheduling click. Twenty minutes of touching, plus the context switch, so call it half an hour of billable time that nobody bills.

Eight rounds a week at half an hour is four hours. At a blended rate of $60 an hour that is roughly $240 a week of unbilled rework, or about $1,000 a month, before counting the credibility cost of a post that published with the wrong crop. That is the budget a preview habit is competing against, and the habit costs about forty minutes a week on the same volume.

The same arithmetic explains why the first round is the expensive one. A client who sees the render before approval has nothing left to object to when it publishes, so the second round, the one that usually involves the account manager apologising, stops happening.

What each network actually changes

Here is the framing rule set the composer preview in PostSider uses today, which is worth comparing against whatever you review on:

Preview familyMedia framingText fold simulated
Video (TikTok, YouTube)9:16 verticalnone
Instagram (feed, or story)1:1 feed, 9:16 story125 characters
LinkedIn, Facebook, article targets16:9 wide210 characters for LinkedIn, none for the rest
Microblog, community, email1.91:1 widenone

Two things to notice. The framing column is not a suggestion, it is what the feed reserves: a 16:9 asset placed in a 9:16 slot does not shrink, it gets cropped. And the fold column is honest about being partial. The composer simulates the shortest folds and leaves the rest to your eye, because a fake fold for a network that does not truncate the same way in every surface teaches the wrong lesson.

Text limits are the other per-network trap, and they do not match the fold. A platform can accept 3,000 characters and still show 210 of them, so a post that passes the length check can still fail the review. The maintained table lives in character limits on every network, and the asset dimensions in social media image sizes.

Preview-first approval is the process fix

The fix is not a longer checklist. It is a change to the order of operations: nothing gets approved before it is seen as it will publish, and approval happens against that rendered version.

Practically, that means three edits to how the work runs.

  1. Review per channel, not per draft. One draft usually becomes five posts. Approve the five renders, with their own crops and folds, rather than the caption text once and hoping. PostSider keeps a per-channel body override so a variant can carry a different first line for the network that cuts it worst.
  2. Record the decision against the version. Approval needs to be a state on a specific version of a specific post, with a name and a timestamp, not a checkbox that survives an edit. In the product this is a pending, approved or rejected state per post, and the decision is separate from the scheduler.
  3. Give the approver a way in that is not your password. A link that opens the pending items for one post, expires on its own and can be revoked is enough for most clients. PostSider issues guest approval links that default to 48 hours and can be set from 1 to 168, stored hashed, and revoked by round.

The economics are simple. One extra minute per post against one revision round per client per week, where a round costs you the rewrite, the re-crop, the re-approval and the small credibility hit of “we will fix that before it posts”.

What a preview cannot catch

Being straight about the limits is what makes the habit stick, so here is the list I give clients.

  • Feed variation. The same asset can be displayed differently across surfaces of the same platform. A preview shows one layout confidently.
  • Link cards. When the platform regenerates a preview card from your URL, the image and title in your post are not what you approved. Test the card once per domain.
  • Emoji and font rendering. The same code point renders differently on iOS, Android and web. Anything where an emoji carries meaning is a risk.
  • Video first frames and autoplay muting. The player picks the frame and the sound state, not you. Check the first second, and assume captions are mandatory.
  • Alt text and accessibility are invisible in the render and get skipped precisely because the render looks fine.
  • Anything the client reads on their phone at 40 percent brightness. A crop that survives your monitor can still fail there.

A five-minute preflight

The routine I would keep if I could keep only one:

  1. Open the preview for every channel in the batch, not just the first.
  2. Read the first two lines aloud and check the payoff is above the fold.
  3. Look at the media in the narrowest slot in the batch, which is the vertical one.
  4. Check the link card once if a URL is in the post.
  5. Approve on the render, and leave the note in the approval rather than in a chat.

That is the whole habit, and it replaces a class of corrections that never needed to happen. The broader version of this checklist is in social media proofing, and the mechanics of sign-off in the client approval workflow.

One caution about routines: they rot. Put the preflight in the same place as the scheduling step, so a post cannot reach the queue without being opened in preview first. A checklist that lives in a document is a suggestion, and suggestions lose to a Friday afternoon.

If you want to see the per-platform preview and the approval queue running together, the agency setup walks through it on a real client roster.

Frequently asked questions

Why do clients reject posts that looked fine in the document?

Because they were approving a document, not a post. A caption in a Google Doc has no fold, and an image in a slide has no crop. Most revision rounds are the client seeing the rendered version for the first time and objecting to framing or truncation, not to the idea.

How much of a caption does each network show before the more link?

It varies, and the safe answer is to assume the first two lines carry the whole post. In PostSider's composer preview the fold sits at 125 characters for Instagram and 210 for LinkedIn, with no truncation simulated for the microblog and community families, so on those networks you should check the fold by eye.

Can a preview catch everything that will go wrong?

No. A static preview shows the crop your upload implies and the fold the network applies to text. It cannot predict a feed layout variation, a link card the platform regenerates, emoji rendering differences across operating systems, or the frame a video player picks as a thumbnail.

Should the person approving be the person who scheduled the post?

No, and that is the point of a separate approval step. The value is a decision recorded against a named person on a specific version of the post. If approval and scheduling are the same click, you have a workflow that proves nothing when a claim goes out wrong.

Let your AI post
to social media for you.

Connect your accounts, add PostSider to your AI and publish your first post in a few minutes.

7-day free trial · 30+ networks · No credit card · Cancel anytime