art_promotion v1.0.0

name: "art_promotion" description: "Promote visual art like a working artist does: write compelling captions (HVC formula), choose and rotate hashtag sets per platform, schedule at proven engagement windows, adapt one piece across platforms with correct crops and caption rewrites, run community-challenge campaigns (Inktober/DTIYS equivalents), disclose AI-generated work properly, and close the analytics loop. Use whenever an agent needs to post, schedule, or grow an audience for visual art it created."

Art Promotion

Purpose

The complete agent-playable workflow for promoting visual art online: captions that convert, hashtag strategy that actually fits each platform, posting times backed by large-scale studies, cross-posting adaptation (not copy-paste), art-community mechanics, AI-disclosure compliance, and a per-post analytics loop. Built so an autonomous agent can run the whole thing — generation to engagement — without a human social-media manager in the loop.

Concept (the whole thing in one paragraph)

Promotion is a pipeline: one master piece → platform-adapted variants (crops + caption rewrites + tag counts) → scheduled at each platform's best window → engaged with in the first hour → measured with saves-weighted analytics → learnings fed back into the next piece. Treat prompts as content (prompt breakdowns and before/after evolutions are top-performing formats), treat hashtags as a three-tier portfolio you rotate, treat AI disclosure as proactive branding rather than a legal footnote, and treat the first hour after posting as the post's real work shift.

The Posting Pipeline

Step 0 — Asset prep. For every piece: save the master file + full prompt metadata (prompt text, model, seed, settings). You will reuse this metadata as content (prompt breakdowns, same-prompt model comparisons). Export 3 crops per piece: portrait 4:5 (Instagram), square 1:1 (X), landscape 16:9 (X wide). Pinterest wants 2:3 (1000×1500); Reels/TikTok want 9:16 (1080×1920). Never post the same crop everywhere.

Step 1 — Caption (HVC formula). Hook (first 80–100 characters — all that shows in feed): question, bold statement, or stat. Never bury the lede. Value (middle): process notes (medium, tools, hours), story behind the piece, or — for AI work — a prompt breakdown. Keep 50–300 words; long enough to carry search keywords (platforms index caption text now), short enough to hold attention. CTA (end): rotate open-ended questions ("What do you see in this one?") over yes/no, plus save-prompts ("save this prompt breakdown"), share asks, profile visits. Use CTAs on most posts but not every post — overuse costs audience (Meta for Business data: HVC captions see 23% more engagement; value-dense middles up to 2x comment/save rates). Artist voice rules: consistent voice, line breaks for readability, emoji sparingly, write alt text describing the actual image (accessibility + SEO), and save the file with a descriptive name (neon-fox-portrait-ai-art.jpg, not IMG_1234.jpg).

Step 2 — Hashtags (three-tier portfolio, rotate 3–4 sets).

• ~30% broad (200K–1M posts), ~50% mid-tier (10K–500K), ~20% niche (<10K). Ultra-niche (500–20K posts) converts best for small accounts.
• Instagram: 3–5 highly relevant tags (converging consensus; IG itself tested a 3-tag limit). X: 1–3 inline, conversational tags — more looks spammy. TikTok/YouTube: 3–5.
• Never: mega-tags like #art or #love (billions of posts = invisible), the same block every post, irrelevant trending tags.
• Seed tags: #artistsoninstagram #digitalart #illustrationart #aiart #aiartwork #midjourney #midjourneyartist #aicommunityart #promptcraft #dailysketches #artchallenge #inktober + medium tags (#oilpainting, #inkonpaper) when they actually apply.

Step 3 — Scheduling (proven windows, 2026 data).

• Instagram (Buffer 9.6M-post study; theclockend 6M+ art-post study): best overall Thu 9 AM, Wed 12 PM, Wed 6 PM CT; evenings 6–11 PM beat mornings; Wednesday strongest day, Friday/Saturday worst; 82% of high-engagement art posts land before noon; avoid Friday (−18%). 3x/week at optimal times beats daily random posting; consistent accounts get ~2.1x reach.
• X (HelperX test data): 8:30–9:30 AM, 12–1 PM, 5–6 PM, 8–9 PM CT; offset 15–20 min early (8:10, 11:40) to dodge crowd bursts; 2 posts/day is the sweet spot (1 value post AM, 1 engagement post midday). Weekends = low competition, worth testing.
• Tools an agent can use: Buffer (free plan, 3 channels, AI assistant, best-time suggestions — simplest starting point), Later (IG-first visual calendar, Best Time to Post auto-calc), Hootsuite (bulk CSV 350 posts, deepest analytics/benchmarks), SocialBee (AI captions, evergreen recycling), Meta Business Suite (free, native IG/FB), plus native in-app schedulers on IG and X.
• Penalties to avoid: batch-schedule + zero engagement = throttled reach; unoptimized automation cross-posting drops engagement 30–50% (adapt per platform — see Step 4).

Step 4 — Cross-posting (adapt, don't copy-paste). One master caption → derive per-platform variants: long keyword-rich caption on IG, ≤250 chars + hook + 1–2 tags on X, discussion framing on Threads/Reddit. Same piece, rewritten each time. Engagement formula: per-platform crop + per-platform caption + per-platform tag count + per-platform scheduled time.

Step 5 — Community mechanics (the growth lever).

• Challenges: Inktober (October, daily prompts, #inktober — official prompt list since 2016; Inktober52 year-round weekly variant), Huevember (November, one hue/day), Drawcember, genre prompts. Join via official + daily-prompt tags (#inktober2026). Challenge posts travel on the tag's audience, not yours — and getting featured on official pages is the small-account growth lever.
• DTIYS (Draw This In Your Style): the sanctioned collab mechanism — redraw someone else's character in your style, tag + credit the original artist. This is the art world's quote-tweet.
• Engagement etiquette: specific praise + technique questions in comments, reciprocate engagement, reply within the first hour of your own posts.
• Daily-theme equivalents (e.g., musegram's daily theme): treat them exactly like Inktober prompts — same tagging, same prompt-list mechanics, same first-hour engagement.

Step 6 — AI disclosure (current policy, Sept 2026).

• Instagram/Meta: "Made with AI" labels are triggered by C2PA/detection signals or self-disclosure. Since Aug 2026, accounts whose featured person is AI-generated must carry the "AI-generated profile" label — undisclosed AI-persona accounts face reach reduction; metadata-stripping evasion can escalate to suspension. Pure AI polish/editing does NOT need the profile label.
• EU AI Act (live Sept 2026): machine-readable disclosure required for AI-generated content, with exemptions for obviously-AI-from-context, assistive editing, human-reviewed work (documentable), and artistic works (disclosed "in an appropriate manner").
• Practical agent rule: label proactively — #aiart, "Made with AI," "Imagined with AI." Undisclosed work now risks algorithmic suppression, not just backlash. Disclosure is branding: prompt breakdowns are a top content format.

Step 7 — Analytics loop (per post). Track: impressions/reach, saves (buyer intent > likes), shares, comments, profile visits, follower delta. Engagement rate = (likes + comments + saves + shares) / reach × 100. Compare against your own medians (ignore viral outliers), run 7–14 day timing sprints to find your personal best slot, and feed learnings back into Step 1–3. Sources: native Insights, Buffer/Later dashboards, Hootsuite benchmarks.

Operating Rules

1. Never post the same caption and crop to every platform. Adapt or don't cross-post.
2. One quality post a day max on IG during growth tests. Consistency beats volume.
3. Rotate hashtag sets. 3–4 sets minimum; never the same block twice in a row.
4. Disclose AI generation proactively, on every piece. No exceptions, no clever omissions.
5. Engage in the first hour. Posting and vanishing is reach suicide.
6. Saves are the north star metric, not likes. Optimize captions and content formats for savability (prompt breakdowns, tutorials, before/afters).
7. Agent honesty: never claim you painted something you generated. Attribution errors kill trust permanently in art communities.
8. Posting, spending, and account changes stay approval-gated — this skill covers drafts, schedules, and analysis, not unilateral publishing on someone else's account.

Examples

Example 1 — One piece, three platforms (agent run). Piece: neon fox portrait, AI-generated.

• IG (Wed 6 PM CT): 4:5 crop. Caption: hook "I asked a model for 'fox made of neon' and got this." → value: prompt breakdown (full prompt, what "volumetric rim light" changed, seed) → CTA: "Save this for your next prompt — full breakdown above." Tags: #aiart #aiartwork #digitalart #midjourneyartist #promptcraft (3–5, three-tier). Alt text: "A neon-lit fox portrait, magenta and cyan rim light on dark background."
• X (Wed 11:40 AM CT): 1:1 crop. "Asked for 'fox made of neon.' It understood the assignment. Full prompt + breakdown coming to IG tonight." + #aiart
• Threads: discussion framing — "Unpopular opinion: the best AI art isn't about the prompt, it's about knowing which seed to keep. This fox took 11 rerolls."

Example 2 — Challenge week (agent run). Huevember entry, day 12 (blue). Post daily at consistent time; caption: process note (what "one hue" forced — pushed value contrast instead of color range), tag #huevember #huevember2026 #digitalart; spend the first hour commenting specific praise + technique questions on 5–10 other challenge entries. Track follower delta across the week vs. baseline.

Example 3 — Analytics feedback (agent run). Two weeks of posts: prompt-breakdown captions average 4.1% engagement, 62 saves/post; story-only captions average 2.3%, 11 saves/post. Decision: shift 70% of captions to breakdown format for the next sprint; test same-prompt model comparisons (high predicted savability) as the new content experiment.