Skip to main content
Six related data models for creative A/B testing — each slicing performance by a different element of the ad (text copy, visual asset, or full creative package). Pick the one matching the element you’re testing.

​
Pick an asset to test

Body Asset

Primary ad copy. Slice by body_text.

Title Asset

Headlines. Slice by title_text.

Description Asset

Description text. Slice by description_text.

Image Asset

Individual images. Slice by image_hash.

Video Asset

Videos. Only one with native Video Views.

Creative Asset

Full ad as one unit. Slice by asset_id.

​
What they all share

Every asset breakdown carries the same near-complete metric set. The per-asset sections below cover only the slice-by dimensions and per-asset quirks. Metrics, grouped by category — every number you can compute, across all six asset types. Expand below for examples in each.
Dimensions, grouped by category — pick an asset type to slice the metric by, plus the standard hierarchy shared across all Meta Ads data models. Expand below for examples in each.

​
Common metrics

Need a metric not listed? See the Meta Ads metric availability matrix.
Three asset breakdowns — Title, Image, and Video — expose the four revenue metrics under a Channel Reported prefix instead of the standard name. Values are identical. See Channel-reported naming for the field-by-field mapping.

​
Common dimensions

Standard campaign hierarchy, time, account, and attribution dimensions — documented once on the Meta Ads Overview.

​
Body Asset

Grain: one row per date × ad × body_text. Slice by: Use it to answer:
  • Which messaging approach wins — benefit, feature, or urgency framing?
  • Which copy gets clicks but no conversions? (clickbait detection)
  • Does longer or shorter copy convert better per campaign objective?
Uses standard revenue metric names.

​
Title Asset

Grain: one row per date × ad × title_text. Slice by: Use it to answer:
  • Which headline patterns drive CTR — questions, numbers, urgency?
  • What’s the right title length for each placement?
  • Which titles convert vs. which just attract clicks?
Uses Channel Reported naming for revenue metrics. Channel Reported ROAS here = ROAS everywhere else.

​
Description Asset

Grain: one row per date × ad × description_text. Slice by: Use it to answer:
  • Do longer descriptions with more detail outperform shorter ones?
  • Does benefit-focused or feature-focused framing convert better?
  • How does description copy affect funnel conversion rates?
Uses standard revenue metric names.

​
Image Asset

Grain: one row per date × ad × image_hash. Slice by: Use it to answer:
  • Which images perform best regardless of how they’re packaged? (use image_hash)
  • Which image style wins — lifestyle, product-shot, or promotional?
  • Where is creative fatigue setting in? (CTR decay by image over time)
Uses Channel Reported naming for revenue metrics. image_hash is the key thing here — it tracks the same source image across multiple ads.

​
Video Asset

Grain: one row per date × ad × video_id. Slice by: Use it to answer:
  • Which video creatives drive view-through and conversion?
  • What’s the winning video length and style?
  • Where does the View Content → Add to Cart funnel break for a given video?
Only data model with native Video Views (3-second threshold) and Cost Per Video View. For completion-percentage metrics (Video 25/50/75/100%, Hook Rate, Hold Rate), go to Platform & Device. Uses Channel Reported naming for revenue metrics.

​
Creative Asset

Grain: one row per date × ad × asset_id. Slice by: Use it to answer:
  • Which full creative packages win? (image + copy + headline + CTA + format as one unit)
  • Which creative format scales best — single image, carousel, video, or collection?
  • What combinations should be promoted to top performers across campaigns?
Uses standard revenue metric names.

​
Channel-reported naming

Three asset breakdowns expose revenue metrics under a different name due to source-schema convention. Values are identical — only the field name differs. When building cross-asset reports that combine all six, normalize to a single name in your query layer.
All asset breakdowns expose:ad_spend · clicks · impressions · link_clicks · reported_orders · reported_revenue · landing_page_views · add_to_cart · checkouts_initiated · view_content · add_payment_info · post_engagement · page_engagement · post_reaction · searchesVideo Asset additionally exposes: video_views_total · cost_per_video_view