Frames and Treatments
The core concepts behind how Coframe models and applies changes
Copy this guide and your current settings as a prompt for your AI agent.
Preview agent promptSee the guide, code and settings included when you copy.
On this page
01The experimentation model
A surface is where Coframe operates: a single page (a url selector) or a set of pages (a url_regex pattern). An element is a specific piece of the DOM on a surface, addressed by a css or xpath selector — the hero headline, a call-to-action button.
A frame — for example, homepage-hero — is a container you create on top of them. It groups the elements you want to change, the goal you want to move, and the strategy for testing, and it holds the experiments. A frame can contain multiple experiments, each targeting a different audience (desktop, mobile, global); for any given visitor, the first matching audience wins and only that experiment activates.
A treatment is the set of changes Coframe applies to a frame when a visitor is assigned to a non-baseline outcome. Treatments are the mechanism; variants are the specific versions they deliver. A single experiment may include several variants (v0, v1, v2), each representing a different treatment of the same frame.
The baseline is the control group: visitors in the baseline see the original, unmodified content of a frame but are still tracked so their behavior can be compared against treatment visitors. A holdout is a group of visitors excluded from an experiment entirely — either at the global level (excluded from all experiments project-wide) or at the per-experiment level (excluded from a specific experiment and treated as the comparison baseline for that experiment).
For the full assignment flow and the IDs each step produces, see Experiment Assignment & IDs.
# Frames and Treatments The core concepts behind how Coframe models and applies changes ## The experimentation model A surface is where Coframe operates: a single page (a `url` selector) or a set of pages (a `url_regex` pattern). An element is a specific piece of the DOM on a surface, addressed by a `css` or `xpath` selector — the hero headline, a call-to-action button. A frame — for example, `homepage-hero` — is a container you create on top of them. It groups the elements you want to change, the goal you want to move, and the strategy for testing, and it holds the experiments. A frame can contain multiple experiments, each targeting a different audience (desktop, mobile, global); for any given visitor, the first matching audience wins and only that experiment activates. A treatment is the set of changes Coframe applies to a frame when a visitor is assigned to a non-baseline outcome. Treatments are the mechanism; variants are the specific versions they deliver. A single experiment may include several variants (v0, v1, v2), each representing a different treatment of the same frame. The baseline is the control group: visitors in the baseline see the original, unmodified content of a frame but are still tracked so their behavior can be compared against treatment visitors. A holdout is a group of visitors excluded from an experiment entirely — either at the global level (excluded from all experiments project-wide) or at the per-experiment level (excluded from a specific experiment and treated as the comparison baseline for that experiment). [Video: Video: surface → frame → experiments, how treatments deliver variants, and the three-step assignment gate (78s, narrated)](assets/videos/fig06-frames-and-treatments.mp4) Watch: 78s — surface → frame → experiments, how treatments deliver variants, and the three-step assignment gate For the full assignment flow and the IDs each step produces, see [Experiment Assignment & IDs](../docs/experiment-assignment.html).