This guide walks you through building a contextual ad package from start to finish. This walks a user through each step in the process from confirming your Dynamic Tags via VLA (Video Level Aggregation), to configuring targeting logic for a package, and finally exporting a bundle to your ad server.
Context Studio reads from your existing tag library in VLA. Tags must be in Validated status before they’ll appear in Context Studio and here’s how to get them there.
Before this step, the tag group must be in Published status. See the Dynamic Tags guide for details on publishing.
Once your tags are validated, return to Context Studio to build the package.
There are two ways to build your targeting criteria, feel free to use one or both:
After entering a description, click Generate targeting criteria. Context Studio will match your plain-language description against your validated tag library and recommend inclusion and exclusion tags automatically. Review the suggestions, accept what fits, and remove anything that doesn’t.
The more specific your description, the better the suggestions.
Example description: Promoting a premium sports drink brand targeting competitive athletes and fitness-focused viewers. Content should feature athletic training, championship moments, and scenes of physical achievement and celebration. Ideal visual settings include gyms, stadiums, outdoor sports fields, and team competitions.
If you already know which tags you want — or prefer to skip the description — click Add Tag in the Targeting Criteria section to select from your validated tag library directly.
Tags added here (via either method) are applied at the video level, meaning the entire video episode is evaluated against the tag criteria. To target at the scene level instead (evaluating individual ad slots between ad breaks), see the Scene Segmentation Guide.
After generating or manually adding tags, review what’s been added under Include content → Video. Tags at the Video level apply to the full episode which is any video where a matched tag fires will be pulled into the package.
To remove a tag that isn’t a good fit, click the three vertical buttons on the tag chip and remove it. To add additional tags, click + Add tag.
Click on any inclusion tag to configure how it’s evaluated at the video level:
Aggregation Method: choose how shot-level scores are combined to determine whether a video passes for this tag:
Minimum Confidence Score: use the slider to set how confident the AI must be before a shot counts (higher = more precise, lower = broader reach)
Save to confirm the settings
Exclusion tags disqualify content that would otherwise match. If a video or scene triggers any exclusion tag, it’s removed from the package — regardless of how many inclusion tags it satisfied.
brand_safety_violence, NSFW)Exclusion tags use “Any Hit” by default, which is usually appropriate for brand safety use cases. If you want to exclude only content that is substantially about a topic (not just briefly), switch to Coverage Threshold.
In addition to tag-based targeting, you can filter your package by video-level metadata. These filters are applied on top of your tag logic. A video must satisfy both the tag criteria and the metadata filters to be included.
Available metadata filters:
To add a metadata filter:
The Match Percentage threshold controls how many of your inclusion tags a video must satisfy to be included in the package.
Once your targeting criteria are set, click Generate Package. The right panel updates to show the videos your package contains.
Review these key metrics:
Below the overview, you’ll see a list of matching videos or scenes. Each row shows:
Click any row to open the Timeline View for that piece of content.
Click into a video to see the timeline view. This gives you a visual breakdown of exactly where your targeted moments appear within a specific video.
If gray vertical lines appear on the timeline, those are ad break cue points provided by the publisher. When Scene Segmentation is enabled, the content between those lines is evaluated as independent scenes. See the Scene Segmentation Guide for more.
Before sharing with an advertiser or account team, you can generate a PDF summary of the package.
When you’re ready to activate the package, click Export bundle.
Exporting a package locks it, preventing any further changes. This ensures your ad server receives is exactly as you intended. Review your targeting criteria and results carefully before exporting.
The Export bundle modal will show you the Full export path which is the S3 location where your JSON file will be saved. This path is determined by the cloud storage connection configured for your organization.
Click Export bundle to confirm. Your JSON file will be saved directly to your S3 bucket at the path shown.
The export path requires a cloud storage connection to be configured before you can export. There are two ways to set this up: