Should You Test Ad Copy or Creative? What Liquid Death’s 229 Meta Ads Show

Liquid Death runs 158 creatives on 42 headlines. Here is why it keeps the words and changes the pictures, and how to give your image and copy separate jobs.

Share This Post

Watch the episode

Watch the episode: Liquid Death's Meta Ads Break Every Rule (On Purpose) (8:28) · Watch on YouTube · More breakdowns on the channel

Liquid Death had 229 live Meta ads on September 23, 2026. I read all of them in the Meta Ad Library, collapsed the duplicates, and went through the actual creative rather than the summary table. On paper the account breaks almost every rule you have been taught: no discounts, no lead forms, no urgency, two button types, and not one tagged link. It is not broken. It is playing a different game, and once you can see which game, every strange choice turns into something you can copy on purpose. This is the written version of episode one of Steal the Strat, my series where I take a brand’s public ad account apart and pull out what a smaller brand can use. The full episode is above.

Key takeaways

  • Liquid Death had 229 live Meta ads on September 23, 2026, built from 158 distinct creatives but only 42 headlines, about five and a half ads for every headline.
  • In Liquid Death’s product ads the image carries the joke and the primary text carries a plain spec line; the most repeated line, “Low calorie soda flavored sparkling water with no artificial sweeteners,” ran on 24 ads for 147 days.
  • 65 Liquid Death Meta ads launched in the 15 days before September 23, 2026, 32 of them on September 22, and almost none of them sold a drink.
  • Only 4 of Liquid Death’s 229 live Meta ads carried an offer, all free shipping, and none of the 229 links carried tracking parameters.
  • A brand that sells mainly on its own website should copy Liquid Death’s creative mechanics, the two voices and the settled copy, but not its untagged links.

The short answer

Should you test ad copy or creative on Facebook?

Test the creative and settle the copy. Once a line of copy has survived a few months, stop rewriting it and start rebuilding the picture around it. Most advertisers do the opposite: they lock one image and write twenty versions of the text underneath it. That is why so many accounts feel busy without moving. The picture is what stops the scroll, so the picture is where the variation belongs.

Liquid Death is the cleanest public example I have found. It locked the text and kept making pictures. Here is what that looks like in the account, and the four parts a smaller brand can take.

The numbers

How many Meta ads does Liquid Death run?

229

live Meta ads on September 23, 2026

158

distinct creatives once the duplicates are collapsed

42

distinct headlines, each reused about five and a half times

2

button types: Shop now on 219 ads, See details on 10

229 live ads on September 23, 2026. Collapse the duplicates and there are 158 distinct creatives. The most duplicated creative in the whole account runs as only 7 ads, and the top three creatives together are 6.1 percent of everything live. Nobody is leaning on a single winner.

Now the number that tells you how the account is run: 158 distinct creatives against 42 distinct headlines and 117 distinct primary texts. Every headline is reused about five and a half times. The copy is settled. The art is the variable.

A few other things in the shape are worth knowing. There are two button types across all 229 ads: Shop now on 219 and See details on 10. All 229 ads run on Instagram and 108 on Facebook, so this is an Instagram account that also runs on Facebook. It is moving to vertical: the 9:16 creatives have a median age of about 41 days against about 100 days for the older 4:5 ones. And there is not one human face on any static ad, no founder, no customer, no testimonial. For a brand this full of personality, the personality is entirely in the writing and the artwork.

Steal the strat · four plays

Steal one: give the image and the copy different jobs

Every Liquid Death product ad is two pieces doing two different jobs. On the image: “Don’t be scared, it’s just iced tea.” In the copy underneath: low calorie iced tea with no artificial sweeteners, B vitamins and natural caffeine. On another image: “You won’t believe it’s not soda.” In the copy: ten calories, no artificial sweeteners, and 90 percent less sugar than the top sodas.

The picture does the joke and the copy does the spec sheet. The image stops the scroll; the copy hands you a reason you can repeat to yourself later, standing in front of a cooler. Most brands make one piece of text do both jobs, and it does neither, because a joke that has to explain itself is not funny and a spec that has to be charming is not clear.

Look where the plain line lives. The most repeated primary text in the account, “Low calorie soda flavored sparkling water with no artificial sweeteners,” runs on 24 ads and had been running for 147 days, the oldest thing in the account. From the brand that built itself on murdering your thirst. Two caveats so you do not over-read it: days running means days since an ad last started delivering, not the day it was created, so treat it as a floor; and an ad that has run that long has survived something, but what it survived is not visible from outside.

What to do

Pull your longest-running ad and check whether the image and the copy are doing different work or saying the same thing twice. If they are saying it twice, rewrite the copy as the plainest possible description of what the product is, and run that as a variant.

Steal two: settle your copy and vary your art

42 headlines across 158 creatives. Once a line has survived a few months, Liquid Death stops rewriting it and keeps making new pictures for it. The words are the slow, risky part of an ad; the pictures are where a fresh attempt costs least and teaches most.

What to do

Find the two or three lines of copy in your account that have run longest. Freeze them. For the next month, every new ad is a new picture on one of those lines, not a new line.

At your size

Three settled lines and a handful of new pictures a week is enough. If producing the pictures is the bottleneck, that is what static ad creation is for.

Steal three: aim the joke at the reader, not the product

65 of Liquid Death’s ads launched in the 15 days before I pulled the account, 32 of them on a single day, September 22. Almost none of them sell a drink. A zip hoodie. A doormat. Pint glasses. A golf polo. A tackle box. Read the copy on those: “Your thirst hates you and everyone you ever loved.” “Do you have arms? Do you occasionally go into public?” “Finally, a doormat that helps your visitors question why they came over in the first place.”

Not one of those is a joke about water. They are all jokes about the reader. Brand humor usually points at the product, which just makes it an ad. Liquid Death points it at the person reading, which makes it a bit you are in on. So the account runs two engines: the drinks are sold on substitution, in flat spec-sheet copy that has not moved in months, and the merch is sold on identity, in the voice the brand is famous for. The boring copy sits on the thing that pays the bills.

What to do

If your brand has a voice, spend it on the reader, not on describing the product. Keep the product copy plain.

Steal four: put the comparison in the name

Say Liquid Death’s product names out loud: Killer Cola, Cherry Obituary, Rest in Peach, Sweet Reaper. Every one is a joke about the aisle the drink is taking share from. Because the product is named after the fight, the ad almost never has to run a comparison. The comparison is already on the can.

When the claim is made out loud, it is kept sharp: 90 percent less sugar than top sodas, 75 percent less sugar than top iced teas, 2 grams against 39, 10 calories against 140. The qualification goes into an asterisk and small type instead of softening the line until it is safe.

What to do

You probably cannot rename your product, but you can rename the thing in the ad. Name the category you are taking share from wherever the eye lands first, and keep any comparison sharp, true and qualified in small type. If you want help with the wording, that is conversion copywriting.

The one that decides

Which of these should you copy for your brand?

Look at the bottom corner of Liquid Death’s product ads and you will find the line that explains the whole account: available at Amazon. Then put the strange numbers back on the table. Only 4 of the 229 ads carry an offer, and all 4 are free shipping. None of the ads running 90 days or longer carries a discount. The ads that send people to the home page are the old ones, at a median of about 103 days, while the newer ads point at product pages. And not one of the 229 links carries tracking parameters.

Read as a direct response account, that looks careless. Read correctly, it is coherent. The ads are not trying to close a sale inside Meta. They are trying to make you recognize the can in a cooler, at a gas station or in an Amazon search you were going to run anyway. That is my reading of public data, not a fact about their results; nobody outside the company can see their spend. The next two teardowns show each game played all the way: Native sends nine in ten clicks to a retailer, and Quince sends almost every click to its own site.

So work out which game you are in before you copy anything. If you can name the retail shelf your product sits on, you are partly in Liquid Death’s game, and recognition matters more than tagging. If your only checkout is your own website, you are in the other game: copy the two voices, the aimed joke and the sharp claim, and do not copy the untagged links.

If you want to see how your own account compares with a competitor’s, my free Facebook ad audit reads both pages from public data and emails you the write-up in about a minute. For the full service, see my ecommerce Facebook ads page.

FAQ

Frequently asked questions

Should you test ad copy or creative first on Facebook?

Creative. Settle copy that has survived a few months and put the new variation into the pictures. Liquid Death's 229 live Meta ads ran on 158 distinct creatives but only 42 headlines.

How many headlines should a Facebook ad account have?

Fewer than most accounts use. Liquid Death reused each of its 42 headlines about five and a half times across 229 live Meta ads, and its most repeated primary text ran on 24 ads for 147 days.

Why does Liquid Death not use discounts in its Meta ads?

Only 4 of its 229 live Meta ads carried an offer, all free shipping. Its product ads print available at Amazon, which suggests the ads are built for recognition at retail rather than closing a sale on its own site.

Does Liquid Death use tracking parameters on its Facebook ads?

Not on any of the 229 live Meta ads I read on September 23, 2026. A brand that sells mainly on its own website should not copy that.

Where does this data come from?

The public Meta Ad Library, read on September 23, 2026, covering all 229 live Liquid Death ads. The library shows ads, copy and delivery start dates, not spend or results.

Method

How was this measured?

Everything here comes from the Meta Ad Library, read on September 23, 2026 through my own teardown tool: 229 of 229 live ads, with distinct creatives counted after removing duplicates. Days running means days since an ad last started delivering. Meta publishes no spend, impressions, click-through rate, cost per acquisition, return on ad spend or revenue for US commercial ads, so none of those is claimed here. Why Liquid Death leaves its links untagged and what the Amazon line means are my reading of public data, not measurements. I have no relationship with Liquid Death.

Want this built for your brand?

If you want your own account read this way and rebuilt around the game you are actually in, that is the work I do with DTC brands.

← Back

Thank you for your response. ✨

Subscribe To Our Newsletter

Get updates and learn from the best

More To Explore

Do You Want To Boost Your Business?

drop us a line and keep in touch