Running an image-only campaign

The lowest-lift batch in the tool: a run of single images in your brand look, captioned per platform, dropped into your queues. Here's the full walkthrough, Look check to publish.

·5 min read·Chelle Honiker
Flat illustration of a grid of framed image cards flowing toward a send arrow, one card glowing as if approved

Videos need filming or rendering time. Carousels need panel text and a swipe arc. An image-only campaign needs neither — it's one strong visual and one strong caption per post, and it's the batch most authors should run first.

It's also where the tool's image pipeline shows off: your persona's brand look, a sample image you approve before the batch spends anything, and your actual book cover in real scenes if you want it.

When image-only is the right campaign

  • It's your first batch and you want the feel of the whole loop before committing to bigger formats
  • Your platforms are image-first (Instagram feed, Facebook, Pinterest, Threads)
  • You're maintaining presence during a drafting stretch — steady, low-effort, on-brand
  • You want a visual spine for a launch that video and carousel batches will layer on top of

Setting it up in the wizard

Open the campaign wizard at /dashboard/create:

  • Basics — batch type Images, then how many posts: 1, 3, 7, or 14. That count is per platform — 7 posts across Instagram and Facebook is 7 on each, each with its own caption. The wizard flags any platform that can't take a plain image (YouTube is video-only) the moment you pick it.
  • Platform setup — each network's native options with plain explanations. Instagram can post as Stories instead of feed; Pinterest wants a destination link; and so on.
  • Direction — the shaping step:
    • What is this campaign supposed to do? Launch, prelaunch, sale, backlist, grow your reader list, meet the author, characters & world, or your own. This drives the caption arc and CTA style for the whole batch.
    • Visual style — atmospheric scenes, bold text hooks, trope cards, review spotlights, or book mockups (more below).
    • Variety dial — signature (tight, on-brand scenes), balanced, or wide (maximum scene range).
    • Image format — tickboxes for the crop; leave it on auto and each platform gets a sensible size.
    • Your Brand Look — the persona's image style guide, editable right there. This is what keeps 14 images looking like one author made them.
  • Source material — pick the book. Cover, blurb, reviews, and the book's reference library all feed the plan.

Generate, and the plan writes N themes, N wordless image prompts, and a caption per platform per post.

The Look check: approve one image before you spend on thirty

The Media studio doesn't start by generating the whole batch. It generates one sample image — post 1's real scene, in your Brand Look — and shows it beside the editable style description.

Not your vibe? Edit the look (or tap a register chip like Illustrated or Photoreal), regenerate the sample, repeat. Nothing batch-sized generates until you say "that's my look." A style miss dies on image one, not image thirty. Approving can also save the adjusted look back to your persona as its new default — tick or untick the box.

Book mockups use your real cover

If you picked the book mockups style, the campaign doesn't describe your cover to the AI and hope — it hands the model your actual cover artwork as a reference, so the book on the café table is your book: your title type, your author name, your spine. (Reference-capable models need your own key; on the included tier the mockups fall back to a described cover.)

The studio loop

Batch fills in after the Look check. For each post:

  • The image prompt is right there — edit it and regenerate, and the studio waits for the new image before calling it done.
  • Try a different engine for any single regenerate — the picker only offers providers you hold a key for, and every post shows a "Made with provider · model" line so there's no mystery.
  • Approve & next — approved images are saved to your Media Library automatically for reuse anywhere.
  • Skip any concept that doesn't earn its slot.

Every generated image is normalized to its exact requested dimensions before it's stored, so a platform's aspect-ratio validation never bounces your batch at the finish line.

Captions and publish

The Captions step shows each image beside its per-platform captions — tighten anything that doesn't sound like you (your persona's guides did the heavy lifting; you're doing the last 10%).

Publish runs a preflight: every platform you picked needs a queue that covers it, and the wizard will create one on the spot — scoped to your connected platforms — if you don't have one yet. Then the batch drops into your queues, and the queues own the timing. No calendars, no start dates.

Common image-batch mistakes

1. Skipping the Look check edit. The sample exists so you can argue with it. If it's 80% right, fix the 20% in the style text before the batch runs — don't fix thirty images after.

2. A vague objective. "Promote my book" produces generic captions in any tool. Pick the objective kind AND write two or three concrete sentences of detail — names, tropes, dates, links.

3. Fighting variety with the wrong dial. If every image feels the same, widen the variety dial; if the batch feels scattered, tighten to signature. That dial exists for exactly this.

4. Forgetting the library. Every image you approve lands in the Media Library. Before regenerating something new for a one-off post, check whether last month's batch already made it.

What to do next

Read next