How to Keyword Stock Video Clips Without Watching Every Single One (With ShotMeta)

Keywording stock video clips takes about three times longer than keywording photos. Most contributors still do it the slow way: watch every clip, type keywords manually, check the agency character limits, then repeat across a folder of 100 files. ShotMeta changes that. Drop in your video files, and Claude AI analyses each one to generate submission-ready titles, descriptions, and up to 40 keywords per clip. It then exports a CSV you can drag straight into Deepmeta (Getty and iStockphoto), Shutterstock, or copy paste into Adobe Stock, or Pond5. Here is the full step-by-step process.

In this article

• Why video keywording is harder than photo keywording
• What you need before you start
• Step 1: Add your clips to ShotMeta
• Step 2: Run the AI analysis
• Step 3: Review and refine the output
• Step 4: Export your CSV and upload to agencies
• Manual vs AI keywording: how the time stacks up
• Frequently asked questions

Why is keywording stock video so much harder than photos?

A photo is a single frame. A video clip has motion, camera movement, subjects entering and leaving frame, and lighting shifts across its duration. Adobe Stock gives you 49 keyword slots per clip; Shutterstock gives you 50. A well-shot 15-second clip genuinely needs 30 to 40 of them once you account for all of the following:

• Subjects and setting
• Motion descriptors (pan, tilt, handheld, timelapse, zoom)
• Shot type (establishing shot, close-up, b-roll, cutaway)
• Mood and concept keywords
  

Typing all of that out manually for 50 to 100 clips is a full work day. Most contributors either under-keyword their video (hurting discovery) or copy-paste photo keywords that do not actually apply to motion footage. Neither works.

What do you need before you start?

ShotMeta is a standalone desktop tool for Mac and PC. Your video files stay on your machine; only sampled frames are sent to the Claude AI API for analysis. You need two things before running your first batch:

1. A ShotMeta licence. One-time purchase at $29. No subscription.
2. An Anthropic API key. This powers the Claude AI analysis and is billed separately. For most batches of video clips, expect a few cents to around 1 dollar of API spend per 100 clips. You set up billing directly with Anthropic at anthropic.com.

Have your video clips ready in a folder before you start. ShotMeta processes batches, so there is no need to feed clips in one at a time.

Step 1: Add your video clips to ShotMeta

Open ShotMeta and browse to your video folder (or individual clips) into the interface. ShotMeta accepts the standard formats contributors use: MP4, MOV. You can mix video and photo files in the same batch. ShotMeta detects each file type and handles it correctly, exporting video metadata as a CSV and photo metadata as embedded XMP/IPTC.

If you are running a mixed batch, there is no need to separate your files beforehand. One folder, one run.

Step 2: Run the AI analysis

Give ShotMeta some context, city, country, concept. Hit Run. ShotMeta samples frames from each video at timed intervals, then sends those frames to Claude AI for analysis. The AI reads subjects, motion type, camera movement, setting, lighting, and shot type from the frames. For a batch of 100 clips, expect roughly 10 to 15 minutes depending on your connection speed.

You do not need to watch any clips while this runs. ShotMeta does not just analyse the thumbnail frame; it samples across the clip's full duration to catch motion, reveals, and subjects that appear mid-shot. That is the key difference between AI video analysis and dragging your photo keywords across to a video file.

Step 3: Review and refine the output

Once the batch completes, ShotMeta exports a csv with each clip's generated title, description, and keyword list. The AI is accurate on visual content but you can review the csv file and alays check. I presonally do not check it anymore. Check for the following:

• Motion keywords. Did it catch the camera movement (pan, tilt, pull-back)?
• Specifics. If the clip is of a named city or landmark, add that. The AI describes what it sees, not what you know.

Spending 10 to 20 seconds on this review beats spending 5 minutes per clip writing from scratch. Edit any field directly in the csv.

Step 4: Upload to agencies

Here is how to use that CSV at the main agencies:

• Getty Images and iStockphoto: Upload your video files through Deepmeta, drag the video files, and csv together. Deepmeta will do the rest.
• Shutterstock: Upload your video files through the Contributor portal, then attach the CSV as the metadata file. Shutterstock matches clips by filename.
• Adobe Stock: Use the bulk upload spreadsheet import in the Adobe Stock Contributor portal. The CSV column structure aligns with Adobe's template.
• Pond5: accepts CSV-style bulk metadata. Check each platform's current bulk upload template and map ShotMeta's column headers if they differ.

If you also shot stills on the same job, ShotMeta writes XMP/IPTC metadata directly into those image files, ready for drag-and-drop upload without a separate CSV. One tool handles both.

Manual vs AI keywording: how the time stacks up

Method	Time per 100 clips	Keywords per clip	Accuracy
Manual (watching each clip)	6 to 8 hours	20 to 30 (fatigue sets in)	High if done carefully
Copy-pasting photo keywords	1 to 2 hours	30 to 40	Low (misses nuances)
ShotMeta	15 min	Up to 40	High for visual

ShotMeta does not replace your eye for what makes a great clip. It replaces the typing.

If you also need a faster photo keywording workflow, the same tool covers that too. See our guide: How to Keyword 100 Stock Photos in Under an Hour.

Frequently asked questions

Does ShotMeta work with MOV and MP4 video formats?

Yes. ShotMeta accepts common video formats including MP4 and MOV. Check the ShotMeta product page for the current full list of supported formats before purchasing.

Does ShotMeta upload my video files to the cloud?

No. Your video files stay on your computer. Only sampled frames (not the full video) are sent to the Claude AI API for analysis. The resulting metadata is written locally.

How much does the Anthropic API cost for video batches?

API costs vary by batch size. For most batches under 100 clips, expect a few cents to a dollar or two of API spend. ShotMeta does not include API credits. You need your own Anthropic API key, billed directly by Anthropic at your actual usage.

Can ShotMeta handle video and photo files in the same batch?

Yes. Drop a mixed folder in and ShotMeta processes both. Photos get XMP/IPTC metadata embedded in the file. Video clips get a CSV export ready for agency bulk upload.

How many keywords does ShotMeta generate per video clip?

Up to 40 keywords per clip. Adobe Stock allows 49 and Shutterstock allows 50, so 40 keywords covers the practical working range while leaving room for your manual additions.

Is ShotMeta a subscription or a one-time purchase?

One-time purchase at $29. No monthly fees. The only ongoing cost is your Anthropic API usage, which you control.