Marketplace Listing Automation:
Reducing Resale Listing Prep from 25 Minutes to 5 Minutes Per Item

For this project, I built a semi-automated marketplace listing workflow for a resale business that needed to increase listing output without making the process harder to manage, train, or delegate.

The original listing process worked, but it was slow and overly dependent on one person manually moving information between photos, notes, research tools, AI tools, spreadsheets, and a cross-listing platform. Each item required product photos, storage tracking, item research, SEO title creation, description writing, thumbnail editing, category mapping, marketplace-specific subform details, and final form entry.

Even with helpful tools like Canva templates, AI-assisted writing, and cross-listing software, the process still took approximately 18 to 25 minutes per item. Hiring a VA who specializes in cross-listing research and form creation would have cost around $1 per listing. The automation I built reduced the direct per-item automation cost to roughly $0.03 to $0.05 per item, excluding fixed software subscription costs, while also reducing hands-on listing preparation time to approximately 2 to 5 minutes per item.

The goal was not to remove human judgment entirely. The goal was to create a system where humans were spending their time on the parts that actually needed human attention, while the repetitive administrative, formatting, folder creation, transcription, data structuring, and form-entry steps were handled automatically.

Marketplace listing sounds simple from the outside: take photos, write a description, add a price, and publish. In practice, especially for resale inventory, the process has many small steps that depend on accuracy.

Before the automation, the workflow looked something like this:

1. Source the item.
2. Photograph the item.
3. Record or remember where the item was stored.
4. Transfer photos to the computer.
5. Create a placeholder listing in the cross-listing software.
6. Upload the item photos.
7. Research the item.
8. Create an SEO-friendly title.
9. Write the product description.
10. Create or edit the thumbnail image.
11. Fill out the main listing form.
12. Fill out marketplace-specific subforms, such as category paths, tags, shipping size, style, or item details.
13. Review the listing and save or publish it.

The process was functional, but it created several operational issues.

First, it was time-consuming. Each item took around 18 to 25 minutes to prepare properly, and that time added up quickly across batches of 20, 30, or more items.

Second, the process was difficult to hand off. Important information lived in different places: photo folders, phone notes, human memory, AI chats, spreadsheet rows, storage labels, and cross-listing form fields.

Third, the business needed a workflow that could grow. If the only person who understood the full process was the owner or the person sourcing inventory, the business would stay limited by that person's available time.

The question became:

How can this process be rebuilt so that the work is faster, easier to delegate, and still accurate enough for marketplace listings?

I started by mapping the actual human process before building the automation.

That was important because the goal was not just to automate random steps. I needed to understand what order the work happened in, where the bottlenecks were, what information was being repeated, and which parts of the process actually required human judgment.

Once I looked at the workflow from beginning to end, I separated it into four categories:

This became the core philosophy of the project: automate the repetitive structure, preserve human review where quality matters, and document the system so another person could operate it without needing to understand every technical layer underneath.

The final workflow became a semi-automated listing system built around audio capture, Google Drive organization, spreadsheet data, AI-assisted listing generation, Telegram bot triggers, and browser automation.

Instead of manually writing notes after photographing each item, the person photographing inventory wears a wireless microphone and verbally describes each item while they are already handling it. They use a consistent spoken structure and code words to separate one item from the next.

For each item, they state information such as:

• Item number
• Brand
• Size
• Condition
• Defects
• Desired listing price
• Storage location or label
• Any internal notes for the person reviewing the final listing

This means the item documentation and the photo process happen at the same time instead of being treated as separate tasks.

After the inventory batch is photographed, the audio is uploaded to a specific Google Drive folder. The user then sends a Telegram command to trigger the automation.

From there, the automation transcribes the audio, separates the transcript into individual items, parses the spoken information into structured fields, creates spreadsheet rows, and generates Google Drive folders for each item. The person then uploads each item's photos into the matching folder.

A separate photo-editing tool handles the thumbnail process, including background removal and creating edited images. Those edited photos are added to the item folders as well.

Once the photos are organized, another automation step creates public image links and file paths. This allows AI agents to reference the item photos for listing generation and allows the browser automation tool to later retrieve the correct photos from the synced Google Drive desktop folder.

The structured item data and photo links are then used to generate the listing content. This includes SEO-friendly titles, product descriptions, keywords, category suggestions, tags, internal notes, storage labels, and default values like pre-owned condition and standard disclaimer text.

Finally, Axiom.ai inputs the generated information into the cross-listing software. The human reviewer then checks the final output, handles marketplace-specific subform details, cleans up any occasional formatting issues, and publishes or saves the listing.

The system can be understood in five main phases.

One part of the automation that shows the logic of the system especially clearly is the audio-to-folder workflow.

This is the step where a spoken inventory recording becomes a structured spreadsheet and a set of item-specific Google Drive folders.

The logic works like this:

1. The user sends a Telegram command to start the automation.
2. Make.com searches a reusable Google Drive parent folder for the uploaded inventory audio.
3. The automation downloads the audio.
4. The audio is sent to AssemblyAI for transcription.
5. A Google Doc transcript is created and saved in the same folder as the original audio.
6. The transcript is scanned for the code words used during the inventory recording.
7. The automation separates the transcript into individual item sections.
8. Each item section is parsed into structured fields, including item number, description, price, storage label, and internal notes.
9. Each item becomes one row in a spreadsheet.
10. The automation retrieves the item numbers from the spreadsheet.
11. Matching Google Drive folders are created for each item.
12. A Telegram update is sent to the user confirming that the inventory folders are ready for photo upload.

This step is a good example of how I approached the larger project. I was not just connecting tools together. I was creating a repeatable operating logic: a human performs the part that needs physical interaction, then the automation converts that human input into structured data that the rest of the system can use.

Click to expand

The final system significantly reduced both time and cost.

This project used a combination of automation, AI, storage, communication, and browser tools.

• Make.com Main automation builder and workflow hub.
• Telegram Bot API Used to trigger automation steps remotely and send progress or error updates to the user.
• Google Drive Audio storage, item folders, photo organization, and archiving.
• Google Sheets Used to structure item data and pass information between automation steps.
• AssemblyAI Used to transcribe the spoken inventory recordings.
• AI AGENTS (MAKE.COM & RESPONSE API) Generate SEO-friendly titles, descriptions, tags, category suggestions, internal notes, and marketplace-specific listing details.
• REMOVE.BG API + LOVEABLE.DEV Used to automate the listing-ready thumbnail image edits.
• Axiom.ai Automates the form-entry process inside the cross-listing software.

The tools mattered, but the more important part was the system design: each tool had a specific role, and the workflow was built so that the human user did not have to manually manage every transition between them.

This project reflects how I approach operations and automation work.

I do not start by asking, "What can I automate?" I start by asking, "What is the actual process, where is the friction, and what parts of this work should a human still be responsible for?"

In this case, the answer was not to fully remove the person from the workflow. The better answer was to redesign the workflow so that the person's time was spent on sourcing, photographing, reviewing, and making judgment calls, while the automation handled transcription, structuring, organizing, generating, formatting, and form entry.

The result was a marketplace listing automation system that reduced time, lowered per-item processing cost, improved delegation, and created a more scalable operations workflow for the business.