The Results page provides detailed insight into how respondents interact with image and video content within an Ad Test, offering engagement behaviors, playback actions, and viewing patterns that reflect how the ads were experienced.

Analyzing Ad Test Metrics

On the Results page, Ad Test simulations display a variety of metrics depending on the type of stimulus: Image or Video. All metric definitions remain consistent regardless of the simulated social media platform (Instagram, Facebook, or TikTok).

1. Click the Metric types dropdown and select or deselect the checkboxes to control which metrics are displayed on the chart.

Some metrics report the number of respondents, while others report the total count of actions:

• Respondent — number of unique respondents.
• Count — total number of occurrences; a respondent may contribute multiple times.

Image Stimuli Metrics

Metrics available for image-based ads:

• Engagement — a respondent metric representing the number of respondents who interacted with the ad at least once through a Like, Comment, Share, or Save. Each respondent is counted once, even if they perform multiple actions.
• Likes, Comments, Shares, and Saves — represent respondents who performed that specific action. If a respondent clicks a button multiple times, only the final state is recorded (e.g., "liked" or not liked).
• Views — indicates whether the respondent scrolled to the post and it appeared on their screen.
• Ad banner clicks — records whether the respondent clicked the ad banner button included in the post.
• Average view duration — the arithmetic mean of the total time the post was visible on respondents' screens.

⚠️ Note: Because Engagement counts unique respondents across all interaction types, the totals for Likes, Comments, Shares, and Saves will not sum to Engagement.

Video Stimuli Metrics

Video stimuli include all metrics listed above, plus additional video-specific measures:

• Total plays — counts how many times respondents watched the entire video.
• Play rate — indicates whether the respondent watched at least 2 seconds of the video. For videos shorter than 3 seconds, scrolling past is counted as a view.
• Video completion rate — indicates whether the respondent watched the video in full at least once.
• Video Watch Duration — a respondent-level metric showing how much of a video each viewer watched. Respondents are grouped into ranges based on the percentage of the video they completed (each respondent appears in only one group):
  • 25% ≤ Watch Duration < 50%
  • 50% ≤ Watch Duration < 75%
  • 75% ≤ Watch Duration < 100%
• Sound metrics — indicate the audio state of the ad when viewed and whether the respondent changed it during playback:
  • Sound was on, stays on — the ad played with sound enabled and the respondent did not mute it.
  • Sound was on, turned off — the ad started with sound enabled and the respondent chose to mute it.
  • Sound was off, stays off — the ad played muted and the respondent did not enable sound.
  • Sound was off, turned on — the ad started muted and the respondent chose to turn the sound on.
• Bounce rate — indicates if the respondent started watching but did not complete the video at least once.
• Autoplay — indicates whether the video began via autoplay or manual play before the respondent manually paused it.
• Pause — captures whether respondents manually paused the video and later resumed it. Auto-pausing due to scrolling out of view does not count.

⚠️ Note: Respondents who watched the entire video at least once (100%) are not included in the Watch Duration groups — they are counted in the Video Completion Rate. For video stimuli, Average view duration counts all time the video was visible, whether playing or paused.

Exporting the Results

To export your Ad Test data:

1. Click the Export icon on the Results page.
2. Choose one of the available formats:
   • RAW Exports — provide a structured file with one column per metric, illustrating whether a respondent was shown the question.
   • Summary Exports — provide detailed ad view data including the metrics as shown on the Stats page, with one column per metric.