# Testing a whole landing page layout vs a single Storyblok block

Asked by Verity Ashworth on 2024-10-30. Tags: storyblok, ab-testing, methodology.

CRO consultant here, recently moved a client from Google Optimize to a
Storyblok stack with Croct. On Optimize my default for big layout changes was
a redirect test: build a second page, split traffic between URLs, reconcile
the numbers afterwards.

On the new stack I am unsure where to draw the line. Block-level tests are
obviously straightforward, but what about a hypothesis that touches the whole
landing page, hero, social proof, and CTA together? Do I test at page level or
compose something from blocks, and what does the platform actually support in
terms of scope? Keen to get the methodology right before we start randomising.

## 2 answers

### Answer from Daniel Okafor (2024-10-30)

The scope options are wider than the block-or-page framing suggests. The
platform supports A/B, A/B/n, and multivariate tests, and they can run on
individual elements or on full journeys spanning multiple pages, so a
whole-funnel hypothesis is testable as one experiment rather than three.

For your landing page case: block-level tests map naturally to the Croct
field on a single block. For the bigger hypothesis, connect the hero, social
proof, and CTA blocks to their own slots and compose all of them in one
experience. Each variant then defines coherent content across the three
slots, so a visitor in variant B sees the B hero, B proof, and B CTA
together. That gives you the "whole layout" test without a second page.

Methodologically that is cleaner than what redirect tests gave you, because
the unit of randomisation is the visitor, not the URL they landed on.

### Answer from jenny_cro (2024-10-31)

Coming from Optimize myself, the part worth internalizing is that there is
no redirect step at all here. Content is decided server-side per visitor, so
there is no flash of the original page before the variant loads, and no
split URLs to reconcile in analytics afterwards. Your GA4 landing page
reports stay on one URL and the variant lives as a dimension, not a page.

One habit to keep from your Optimize days though: composed multi-slot tests
have more moving parts, so preview each variant end to end before launch.
The bigger the scope, the easier it is to ship a variant where the hero and
CTA tell different stories.