# How do Ninetailed experiences and variants map onto Croct's model?

Asked by Priya Nair on 2026-09-17. Tags: ninetailed, experiences, migration.

I am rebuilding what our team had in Ninetailed on Contentful, and before I
touch anything I want a clean one-to-one mapping of the concepts so I can plan
the work instead of discovering the differences mid-rebuild.

Specifically: how do Ninetailed's experiences and variants translate into
Croct's experiences and slots? If someone can lay out which concept becomes
which, I can size the rebuild properly.

## 2 answers

### Accepted answer from steffi_r (2026-09-19)

The mapping is close enough that planning is straightforward. In Croct an
experience defines three things: who (the audience), where (the slot), and
what (the content for that audience). That covers the targeting-plus-content
part of a Ninetailed experience directly.

The variant/experiment part is separate but attaches to the experience: an
experiment has a primary goal and 2 to 5 variants. So a Ninetailed
experience that was really an A/B test becomes a Croct experience with an
experiment on top; one that was pure targeting is just the experience with
no experiment.

The audience piece is where it stops being a straight rename. Croct
audiences are CQL conditions evaluated in real time per interaction, not
precomputed segments you build up in advance. There is a fuller breakdown of
that in [how Ninetailed audiences map to Croct](/answers/ninetailed-audiences-map-to-croct).

### Answer from kwabena_o (2026-09-19)

One detail to add for planning, since Ninetailed let experiences overlap on
the same component. When several experiences match the same slot, Croct
resolves by experience priority: a unique numeric rank per slot, and the
higher one wins. So if you had stacked/ordered targeting before, that
becomes explicit priority numbers rather than implicit ordering.

Also the statistics are Bayesian and unsampled with a built-in engine, so
your experiment results do not depend on wiring up an external analytics
tool the way some Ninetailed setups did.

#### Reply from Priya Nair (2026-09-20)

This is the model I needed. Experience = who/where/what, experiment
bolts on for the A/B cases, priority handles the overlaps. Much clearer
now, thank you.