Social Media Marketing

Social Media Marketing is a social media post generator from video that downloads one accessible source, transcribes its speech locally, and creates reviewable drafts for selected platform, tone, language, and model combinations. Each completed generation task becomes one nine-field Dataset row, with optional source screenshots or styled image variants shared across the run.

• Six required controls. Select the video, image mode, output language, platform profiles, tones, and model routes.
• Nine ordered output fields. Content, labels, images, provider usage, processor, and timestamp use one stable task-row contract.
• Direct-first retrieval. The media route uses an eight-second socket timeout and no more than three public-proxy fallbacks.
• Bundled local ASR. FFmpeg and CPU-int8 Whisper small run inside a 2 GiB image without a cold model download.

At the smallest paid-plan shape, one platform × one tone × one model requests one post call. A call reporting 128 text tokens adds $0.004352 at the $0.000034 rate, plus $0.02 for Actor Start at 2 GiB, before Actor Usage or optional image events.

What does Social Media Marketing do?

It turns spoken video content into structured draft variants without publishing them. The Actor retrieves one media source, extracts audio with FFmpeg, transcribes speech locally, builds tasks from the selected arrays, and sends text prompts through current OpenRouter model routes.

The URL contract is deliberately broad: an absolute HTTP(S) URL handled by the installed yt-dlp extractors. The Actor tries direct retrieval first. If that fails, it can use no more than three shuffled public proxies; it does not spend paid residential-proxy traffic and does not add authenticated source cookies.

Paid task count is platforms × tones × models. Two platforms, three tones, and two models request 12 calls. Duplicate values create repeated tasks. FREE-plan runs use DeepSeek, the first platform, the first tone, and no images, regardless of a larger selection product.

The returned platform is an internal writing profile. It does not connect to a social account, verify current destination limits, publish content, or guarantee engagement.

How do I run Social Media Marketing?

Start with the prefilled 19-second YouTube video and one value in every array. The TrueFetch example intentionally differs from the AgentX family source.

{
  "video_url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
  "image_style": "Original",
  "language": "Spanish",
  "platforms": ["LinkedIn"],
  "tones": ["Professional"],
  "models": ["Gemini"]
}

Open the Actor input page, review the selections, and click Start. On FREE plans, the runtime forces images off and uses DeepSeek plus the first platform and tone. Paid plans attempt the full product and the selected image mode.

After completion, inspect terminal status, charged events, Dataset count, and row content together. Image modes also use the default Key-Value Store. Generated text is a draft: verify every name, number, quotation, claim, language choice, and destination requirement before use.

What data does Social Media Marketing return?

Each saved item represents one successful post-generation task and contains nine fields. A provider response with no text is not written.

Field	Type	Description
content	string	Reviewable generated post text.
platform	string	Requested internal platform prompt profile.
model	string	Uppercase selected model label.
tone	string	Requested tone label.
language	string	Requested output language label.
images	array	Shared KVS URLs; empty when images are disabled or unavailable.
tokens	integer	Provider-reported total tokens for this post call.
processor	URL	TrueFetch Actor URL that created the row.
processed_at	datetime	ISO 8601 UTC timestamp for the row.

{
  "content": "Los elefantes convierten una observación breve en una historia memorable: los detalles concretos hacen que una idea permanezca.",
  "platform": "Linkedin",
  "model": "GEMINI",
  "tone": "Professional",
  "language": "Spanish",
  "images": [],
  "tokens": 128,
  "processor": "https://apify.com/truefetch/social-media-marketing",
  "processed_at": "2026-08-13T17:13:29+00:00"
}

The source URL remains in INPUT and the local transcript remains internal prompt context. Preserve the run ID and INPUT record when downstream systems require complete lineage. tokens does not include local transcription or separate image-discovery and selection calls.

What inputs can I configure?

All six properties are required and validated before media retrieval. Console prefills provide values but do not make fields optional.

Input	Type	Available values	Operational meaning
video_url	string	absolute HTTP(S) URL	One extractor-compatible source; playlists are disabled.
image_style	string	None, Original, 8 styles	None skips images; Original selects frames; styles request generated variants.
language	string	12 languages	Requested generation language, subject to model review.
platforms	array	12 profiles	One or more internal destination-writing profiles.
tones	array	12 tones	One or more prompt tone strategies.
models	array	6 routes	DeepSeek, Claude, Gemini, Grok, Qwen, or GPT.

Unknown keys, invalid URLs, empty values, non-string array members, and invalid enums fail before source requests. Arrays have no maximum length. Deduplicate selections unless repeated tasks are intentional, and estimate the product before starting.

What platforms and markets are supported?

The Actor offers 12 writing profiles and 12 requested languages, not publishing integrations or market guarantees. Platform choices are X-Free, X-Premium, Instagram, LinkedIn, Facebook, Pinterest, Reddit, TikTok, YouTube, Rednote, Threads, and Mastodon.

Languages are English, Chinese, Japanese, Korean, Spanish, French, German, Russian, Portuguese, Italian, Arabic, and Hindi. The model may return incomplete text, the wrong language, or culturally unsuitable phrasing. Reviewers should localize rather than merely translate.

Video-site compatibility follows yt-dlp and the current source. A site can be listed by the extractor project and still fail because of authentication, geography, removed media, rate limits, or upstream changes.

Why use Social Media Marketing?

Use it when repeated, structured draft generation is more valuable than building and operating each pipeline component yourself. It provides media retrieval, local ASR, task multiplication, model routing, optional images, usage capture, and Dataset delivery behind one input contract.

Unique workspaces prevent a prior run's MP4 or screenshot from satisfying a later request. MP4 candidates must pass ffprobe, and FFmpeg audio extraction must complete successfully. The Whisper model is bundled inside the image, avoiding a cold model download during each run.

Empty model output does not become a misleading row. If every task fails to return content, the run fails instead of silently completing with zero drafts. Zero reported tokens are not charged.

Who is Social Media Marketing for?

Editorial, localization, campaign, research, and support teams can use it as the generation stage before human approval. Authorized interviews, demos, webinars, explainers, and short videos can become multiple candidate hooks or summaries without changing the output shape.

Original-image mode can attach selected source frames when reuse rights exist. Styled modes can create visual alternatives from selected frames, but results are best-effort and may contain fewer than six images.

This Actor is not suitable as an autonomous publisher, factual verifier, rights checker, moderation system, or certified transcript service. Those controls belong in the surrounding workflow.

How can I use Social Media Marketing through the API or MCP?

Call the Actor through REST, an Apify client, Schedules, webhooks, or MCP, then route the Dataset to a review queue. Keep tokens secret and set a charge cap sized for the selected product.

{
  "method": "POST",
  "url": "https://api.apify.com/v2/acts/truefetch~social-media-marketing/runs?waitForFinish=300&maxTotalChargeUsd=1",
  "headers": {
    "Authorization": "Bearer YOUR_APIFY_TOKEN",
    "Content-Type": "application/json"
  },
  "body": {
    "video_url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
    "image_style": "None",
    "language": "English",
    "platforms": ["LinkedIn"],
    "tones": ["Professional"],
    "models": ["DeepSeek"]
  }
}

Retrieve defaultDatasetId after success and compare the item count with the expected task product. Webhooks should evaluate terminal status as well as Dataset count because a provider or charge-limit interruption can stop generation early.

How much does Social Media Marketing cost?

Pricing uses Actor Start, Actor Usage, Text Token, and Image Token events. At 2 GiB, Actor Start produces two $0.01 events. Actor Usage is $0.0001 per event. Text Token is $0.000034 and Image Token is $0.000080 per provider-reported token.

Post calls produce text-token events. Paid image discovery and screenshot selection also use text-token events. Styled-image provider usage produces image-token events; Original screenshots do not. Events occur before the final Dataset batch, so a later failure can leave fewer rows than charged calls.

Use one platform, one tone, one model, and None to measure a source first. A larger product increases sequential provider calls, runtime, and spend without guaranteeing more diverse or higher-quality content.

How does Social Media Marketing compare with alternatives?

Use Video To Text when the required output is transcript text, TikTok Video Downloader when the required output is media, and this Actor when the required output is social-post drafts. Instagram Video Transcript provides a platform-specific transcript workflow.

A custom stack offers full control over extraction, ASR, prompts, model providers, storage, and moderation, with corresponding operational work. Manual writing gives maximum context for a few videos. This Actor is designed for repeatable task matrices and structured review queues.

What are the limits and troubleshooting steps?

Classify a failure by the last completed stage: retrieval, transcription, generation, or images. This keeps source-site problems separate from model and storage problems.

• Retrieval can fail for removed media, authentication, geography, extractor drift, or lack of a usable format. Public proxies do not provide cookies.
• Transcription can fail on silent, corrupt, or unsupported media and can misrecognize accents, names, or overlapping speech.
• Generation can fail because a model route changes, a provider returns no content, a timeout occurs, safety controls apply, or a charge limit is reached.
• Images can fail during timestamp choice, frame capture, provider styling, upload, or cleanup. Valid post rows can still contain an empty array.

Long videos consume download, FFmpeg, and CPU time before generation begins. Large selection products run many calls. Start small, observe events and duration, then expand deliberately.

Frequently asked questions

Does Social Media Marketing publish the generated posts?

No. Platform values are prompt profiles. Publish only after review through your own authorized destination workflow.

Which video URLs can I use?

Use an absolute HTTP(S) URL handled by current yt-dlp extractors. The upstream supported-sites list is guidance rather than a current success guarantee.

How many Dataset rows should I expect?

Paid intent is platform count × tone count × model count, including duplicates, until interruption. FREE plans use the first platform, first tone, and DeepSeek.

Why can token charges exceed saved rows?

Provider usage is charged as calls finish, while rows are pushed after generation and optional image work. A later failure or cap can reduce saved rows.

Are the requested language and tone guaranteed?

No. They are prompt instructions. Review for language accuracy, cultural fit, factual support, and destination suitability.

Can Original or styled modes return no images?

Yes. Frame selection, provider output, upload, or limits can return fewer images or none. Text rows remain useful with an empty array.

Is the transcript available in the Dataset?

No. The transcript is internal prompt context. Choose a transcript-focused TrueFetch Actor when transcript delivery is required.

What should I include in a support report?

Include the run ID, UTC time, build, redacted URL, six input selections, expected task product, terminal status, event counts, Dataset count, and the last completed stage. Never share API keys.

Related TrueFetch Actors

• Video To Text — return transcript-oriented text from accessible video.
• Instagram Video Transcript — use a focused Instagram transcription workflow.
• TikTok Video Downloader — retain accessible TikTok media instead of generating drafts.

Support

Use the Store page to run the Actor, the API page for integration details, and the Issues page for reproducible failures. Product questions and updates belong in the TrueFetch community.

Last Updated: August 13, 2026