Test Two Designs Against Each Other
Run a fair head-to-head between two of your designs on the same catalog — build the branches in Meta Ads Manager, pin each one to a different design with the Renable Options panel, and read the result on Meta's side.
Updated September 21, 2026
You have two creative concepts and you want to know which one sells. Because every Renable design
runs off the same catalog, you don't need a second feed, a second catalog, or a second set of
products — you need two ad sets pointed at the same catalog, each pinned to a different design.
Renable does not build the test for you. You build the branches in Meta Ads Manager the way you
would build any A/B test, and Renable decides what each branch shows. This article is the practical
version of that split.
Before you start#
- A Meta channel that's published and live, with an ad account connected — see
The Renable Options browser extension.
- The two designs you're comparing, already in that channel's feed.
- On a channel with adaptive designs on, that means two design
groups — see Set up design groups.
- On a channel that still picks designs by position, it means two designs in the feed. Both work;
the difference is only what you click in the panel.
- A campaign you're willing to change. Switching a design has a cost on Meta's side (below), so this
is not something to do casually on your best-performing ad set.
1. Compare like with like#
A test only answers a question if the two branches differ in one thing. Two mistakes make the
result meaningless before you start:
- Different shape coverage. If one design group carries a tall design and the other doesn't, the
second one can show letterboxed on Stories and Reels-style surfaces, and you're measuring shape
coverage rather than creative. Give both groups the same shapes, or accept that you're comparing
two different things.
- Different media. A group carrying video and a group carrying only images aren't the same kind
of thing either — Meta chooses image or video per impression on the first one. That can be a
perfectly good test, as long as it's the test you meant to run.
Instagram, Stories and Reels are not part of the test, and can't be. Those marks live on the
Default designs group and, in our own testing, apply to every ad, so both arms share them: an ad
pinned to either group shows the same default Story design on Stories. Meta documents none of this,
and we've confirmed it in ad previews rather than in live delivery. What you're comparing is the feed and 4:5
surfaces, where each group's own designs run. See
Instagram, Stories and Reels. If an older group still carries one of those
marks, its row in the dialog says so — clear it before you start, or it's deciding a surface for
both arms at once.
The Adaptive designs dialog nudges you when a group is missing a shape the base group has, and warns
on the Default designs row when your defaults have a gap. Both are worth clearing before you
spend budget.
2. Check both designs are actually ready#
Open one of the ads you're about to change. Before you commit budget, the Renable Options panel
already tells you the three things that go wrong:
-
Coverage — "Renders for 92% of products in this feed". This is the number that quietly ruins
tests: where a design is missing for a product, that product falls back to your default
designs in that branch, so a branch with 60% coverage is 40% a test of your everyday creative. A
very low or 0% figure means the design isn't meaningfully in the feed yet.
Where you can read that number matters. On a channel with adaptive designs on, the panel
shows these percentages only under the group the ad is currently running — the groups listed
under Other design groups show their shapes, not their coverage. So check a group by opening
an ad that already runs it, or pin the first branch and read its figures there before you start
the second. On a channel that still picks designs by position, every listed design carries its own
figure, so you can compare both before switching either.
-
Freshness — when Renable last updated the feed and when Meta last fetched it. A design change
only reaches your ads after both have happened, and the panel names whichever step is still
pending.
-
Whether Meta has the design yet — a design Meta hasn't ingested is marked on its row and again
in the switch confirmation. That's a warning, not a block: in our own testing the ad kept running
normally on the default designs and started showing the pinned design as soon as Meta fetched the
feed. But that is exactly the problem for a test — an ad spending its first days on your default
designs is not the branch you think you're measuring, so wait for the fetch before you start
counting.
3. Build the branches in Ads Manager#
This part is entirely Meta's. At a high level:
- Take the ad set you want to test and duplicate it, once per design you're comparing.
Duplicating keeps the ad set's setup — same catalog, same audience, same budget, same
placements — so the design is the thing you changed.
- Name each ad set after the design group it will run (
black-friday, evergreen). You'll be
reading these names in a report later, and Meta's report will not know anything about your
designs unless the names carry it.
- If you want Meta to run the comparison formally rather than eyeballing two ad sets, use Meta's
A/B test feature. Meta splits your audience into random, non-overlapping groups, one per
branch, which is what keeps the branches from competing with each other — two ad sets run side
by side without it are not split evenly. You can start one by ticking an ad set in Ads Manager
and choosing A/B test from the toolbar, or from Meta's Experiments tool. Meta moves these
entry points from time to time, so follow its own guides for the current steps and for how it
reports: About A/B testing and
Create an A/B test by duplicating an ad set or ad.
4. Pin each branch to its design#
Open the first branch's ad in the Ads Manager ad editor, with the
Renable Options extension installed.
- With adaptive designs on, the panel lists your design groups. Click Use on the group
this branch should run, review the confirmation, and confirm.
- Without adaptive designs, the panel lists individual designs and you pick one the same way.
Repeat on the other branch's ad, picking the other design.
Ads Manager's own preview doesn't refresh by itself, so the switch dialog offers a one-click
Reload page afterwards.
5. What a switch costs#
The panel says this on every switch, and it's the reason not to do this on a live winner without
thinking:
- Ad creatives can't be edited on Meta, so switching creates a new version of the ad's creative
and points the ad at it.
- Meta may review the ad again before it runs.
- A live ad's delivery learning may restart. Switching on a paused ad avoids that cost — so
duplicate, pin while paused, then start the branches together. That also makes the test cleaner,
because both branches begin from the same point.
6. Read the result on Meta's side#
Results live in Ads Manager, not in Renable. Renable knows which design each ad runs; it does not
receive performance data back from Meta, so nothing in Renable will tell you which branch won. Read
the branches in Meta's reporting the way you'd read any other ad-set comparison — which is exactly
why step 3's naming matters.
How many branches, and how much budget#
Fewer branches beat more. Every branch you add splits the same budget and the same conversions, so
two designs compared properly tell you more than five compared badly.
Meta needs enough delivery in each branch to tell a real difference from noise, and how much that is
depends on your budget, your price point, and your conversion rate. Nobody — including Renable — can
give you a product-wide number for that up front, so don't take one from anyone who offers it. Run
the branches long enough that the numbers stop moving around, and be honest with yourself about
whether the gap you're looking at is a gap.
One thing worth doing regardless: change one design at a time. Adding a design group never
changes which image an unpinned ad shows, so you can leave your evergreen ads running untouched while
the test runs beside them.
Video is the exception. A group's videos are the product's videos, and any ad with Dynamic Media
on can draw on them whether or not it's pinned — so adding video to one branch's group can change
what your untested ads show too. Keep video out of a test you want cleanly isolated, or add it to
both branches deliberately. Run different designs for different campaigns or
seasons covers the scoping in full.
After the test#
The winner is already a design group in your channel, so "rolling it out" is just pinning more ads to
it — see
Run different designs for different campaigns or seasons.
The loser can stay in the feed doing nothing: an unpinned group costs nothing and changes nothing.
Delete it only if you actively want it gone.
Troubleshooting#
Both branches look identical in Ads Manager's preview.
The preview doesn't refresh on its own — reload the page. If it still looks the same, check the
panel's freshness line: Meta may not have fetched the feed carrying one of the designs yet.
One branch doesn't look like the design I pinned.
Check that design's coverage figure in the panel — with that branch's own ad open, since the panel
reports coverage for the group the ad is currently running rather than for the ones it offers. Where
the design is missing for a product, that product falls back to your default designs in that branch,
so a partly-covered group shows a mix of the concept you're testing and your everyday creative.
The panel won't let me switch a design.
The panel explains why in that feed's own terms. The two common cases are a feed whose design
positions vary per product and a feed sending images two different ways — different problems with
different fixes, both described in
The Renable Options browser extension.
Where to go next#