Social Media Autoposting System for a Cosmetics E-commerce Brand

Beautyholic came to Pixerry Studio with a request that sounded simple — post to all our social networks at once — and got complicated the moment you looked at the volume:

~10 posts per day, released at 10-minute intervals, each carrying a variable number of media files that never share the same dimensions

The goals were shaped around four priorities — one control point instead of five, zero manual media preparation, automated short-video production, and full visibility over what actually went out.

• Build a single control panel where the whole publishing schedule lives, using an interface the client already works in every day
• Publish to Telegram, VK, Dzen, Instagram and YouTube from one row, with post text automatically adapted to each platform’s format
• Handle up to 10 media files per post — photos and videos, mixed ratios — and reformat them automatically for Instagram instead of cropping by hand
• Automatically assemble Reels and YouTube short videos: normalise all media to 1080×1920, stitch them into one clip, and lay a soundtrack over the top
• Keep a human checkpoint for video: nothing goes out until the owner has watched the finished clip
• Avoid a paid third-party scheduler by building on Google Sheets, which already hosts the client’s CRM
• Connect a familiar cloud media library so files can be dropped in once and picked up automatically

Making one spreadsheet row produce five correctly formatted publications, without the client ever touching an editor, was the core engineering challenge.

• Variable media, fixed requirements: each post carries a different number of files, in different sizes and aspect ratios. Instagram needed automatic reformatting; the pipeline had to cope with any combination of photo and video without breaking.
• Five platforms, five sets of rules: text formatting, media limits and delivery methods differ across Telegram, VK, Dzen, Instagram and YouTube. The system adapts the same source text per destination rather than forcing one lowest-common-denominator version.
• Publishing cadence and reliability: ~10 posts a day at 10-minute intervals means the scheduler has to be precise and resilient — no duplicates, no silent failures, no drift between the planned time and the actual publication.
• Video assembly inside a no-code stack: normalising mixed media to 1080×1920, concatenating it into one clip and trimming a music track to the exact clip length required a rendering step engineered into the Make.com scenario.
• Scope discipline on AI features: we prototyped a richer variant — AI-written video script, synthesised voiceover, automatic subtitles. It worked, but it wasn’t right for the brand’s tone, so the production version stayed with clean music-only edits.

The delivered system reduced a full day of publishing labour to filling in a spreadsheet. The client plans content; the automation formats, renders, schedules, publishes and reports.

• Sheet 1 — Input. Date and time, post text, media links, and a set of toggles for choosing destination platforms. Status set to “Publish” and the row is live in the queue.
• Automatic per-platform adaptation. At the scheduled moment the system pulls the row, rewrites the text into each selected platform’s format, and delivers it together with the media.
• Automatic short-video production. Tick the video option and the system collects the media, normalises everything to 1080×1920, stitches it into a single clip and overlays the soundtrack — trimmed to the exact length of the video.
• Sheet 2 — Video review. Filled in by the bot: the caption and a preview link to the finished clip. The client watches it, selects platforms, hits publish. Nothing leaves without approval.
• Sheet 3 — Reporting. Every publication is logged automatically — what went out, where, and when.
• Telegram notifications. The client is pinged when a video is ready and again after each publication, so the system is auditable without opening the spreadsheet.
• Dropbox media library. Files are dropped into the client’s usual cloud storage and pulled into posts automatically.
• A reusable content base. Repeat posts are duplicated with a new date and time — the archive becomes a scheduling asset rather than dead weight.

Beautyholic now plans weeks of content in one sitting, on the days when there’s time for it, and the system carries it out on the days when there isn’t. The immediate win was hours returned; the durable one is a growing library of approved posts that can be switched back on with a copied row.

This project connected content planning, media processing, video rendering, multichannel delivery and reporting into a single workflow — with the client’s existing tools as the interface.

• A control panel built inside Google Sheets, alongside the CRM the client already runs there — no new subscription, no new interface to learn
• Media preparation removed from the human workflow entirely, including automatic reformatting for Instagram
• Short-form video produced end to end by the system, with a mandatory human review before anything goes public
• Full publication logging in the spreadsheet and in Telegram, so nothing is lost or duplicated blind

The result is a publishing operation that scales with content volume instead of with headcount.

Key numbers

• 1 month from brief to launch
• 5 platforms published from a single spreadsheet
• ~10 posts per day, at 10-minute intervals
• Up to 10 media files per post, auto-formatted
• Up to 10 media files per post, auto-formatted
• 1080×1920 Reels assembled and scored automatically
• Dozens of hours of manual work saved every month