How a Canned-Water Founder Tested Demand Before Manufacturing Anything

Liquid Death tested the most uncertain part of a canned-water business before it could manufacture a can. Mike Cessario spent $1,500 on a deliberately extreme brand video and about $600 promoting it. The video approached three million views, while distributors and 7-Eleven operators asked how to stock a product that did not yet exist. That did not prove profitable demand, manufacturing feasibility, or repeat purchase. It did prove that the positioning could make ordinary water noticeable. A physical-product founder can copy the sequence without pretending to sell a nonexistent item: test the promise, capture qualified retailer and customer actions, then spend on samples and production only if those actions cross a written threshold.

The first prototype was a piece of media

Cessario came from advertising and observed that musicians often poured water into energy-drink cans at shows. His concept was not a new molecule. It was water packaged and marketed with the codes of beer and heavy music. In a 2019 interview with The New Consumer, he said the company launched on social media before it had product, shot a $1,500 video, and put roughly $600 behind it.

The nearly three million views were not the strongest signal. Direct messages from distributors and convenience-store operators were closer to the future transaction. A view says the creative interrupted attention. A stocking inquiry says a buyer can imagine shelf demand. Neither is a purchase order, but they sit at different distances from revenue.

The experiment isolated brand risk from manufacturing risk. It asked whether people would react to the name, can, and tone before the founders solved filling, shelf life, freight, retail terms, and inventory. That sequencing saved the larger production spend from carrying an untested message.

Build an evidence ladder, not a viral target

Rank test actions by commitment:

1. A video view or like is attention.
2. An email signup is permission to follow up.
3. A completed price survey is stated interest.
4. A sample request with address is effort.
5. A retailer meeting or written stocking inquiry is channel intent.
6. A refundable deposit or real purchase order is financial commitment.

A founder should decide in advance which rung justifies the next expense. For example, a $2,500 creative test might need 300 qualified email signups at $8.33 each, 30 completed sample requests, and five retailer calls. Failing the view target but passing the retailer target can still justify samples. Passing views without any higher-commitment action should not justify a factory order.

This is where many "fake door" tests fail. They celebrate clicks and ignore the distance to purchase. Use a transparent page that says the product is in development, shows the expected price and delivery region, and invites a waitlist or refundable reservation. Do not imply stock exists when it does not.

Manufacturing changed the question

Cessario also told The New Consumer that non-carbonated water in aluminum cans without preservatives was difficult to source in North America, so the company initially found a producer in Austria. The creative pilot could not reveal that constraint. After positioning passed, the next experiment had to test production and logistics.

This separation is useful. A concept test should be cheap and narrow. A production pilot should then measure minimum run, filled-can cost, freight per case, damage, shelf life, lead time, and working capital. Combining both into a large launch makes it hard to know whether weak results came from the product promise, price, supply, or distribution.

Illustrative arithmetic: assume a 12-can case sells to a retailer for $14, filled product and packaging cost $6.60, inbound freight costs $1.80, and allowances and damage reserve cost $1.20. Contribution before overhead and marketing is $4.40 per case. A $22,000 pilot therefore needs 5,000 cases just to recover that expense:

$22,000 / $4.40 = 5,000 cases.

If the concept test generated only 200 likely customers, the gap cannot be solved by enthusiasm. The business needs retailer volume, a higher contribution, a smaller pilot, or more proof.

Attention became useful because shelves followed

Fast Company reported that the brand started with its Facebook commercial in 2017 and began selling cans in 2019. It later entered Target, Kroger, and Walmart. The company's reported $263 million in retail scanned sales as of March 2024 is not the same as manufacturer net revenue, and updated sales were not disclosed in that article.

A Harris Poll analysis recorded increases in familiarity, trial, and usage among younger adults from the first to second quarter of 2024. The methodology measured brand perceptions, not causal return on a particular campaign. It supports a narrower point: attention can move further down the funnel when a product is widely available.

That last condition matters. A shareable ad for a product buyers cannot find wastes intent. Physical brands should schedule distribution and media together. The approach differs from Ridge's repeated creator buying, covered in the creator-portfolio case. Liquid Death first tested whether one sharp premise could earn retailer interest before scaling paid reach.

A 60-minute pre-production test brief

Write four lines:

• Risk: the single customer belief that must be true, such as "premium commuters will choose a reusable steel lunch container at $85."
• Artifact: the cheapest honest representation, such as a working prototype video and price-specific landing page.
• Qualified action: the behavior closest to purchase that can be collected now.
• Gate: the minimum count and maximum acquisition cost that unlocks samples, tooling, or a test order.

Then add a disconfirming result. If the test produces cheap clicks but no price-qualified actions, revise the offer rather than declaring victory. If retailers inquire but consumers do not, ask retailers what comparable sell-through makes them interested and design a local trial.

The general channel framework in Where First Customers Actually Come From can identify test audiences. The manufacturing discipline in MNT Reform's supplier map belongs after the message clears its gate.

Liquid Death's pilot was unusually theatrical, but the transferable mechanism is sober. Test one expensive uncertainty with a cheap artifact. Count actions by commitment, not applause. Only then pay to learn the next thing. That sequence does not guarantee a beverage brand, but it keeps a founder from asking inventory to prove a message that a landing page could have disproved.

Prevent a concept test from lying

Use the expected selling price in the test. A striking video with no price may prove curiosity about the idea while saying nothing about willingness to buy. Ask for delivery country and intended use, because interest outside the first shipping region or from people who would never consume the product cannot support the initial order.

Deduplicate signups and separate paid from organic traffic. Record creative, audience, spend, qualified actions, and retailer inquiries in one sheet. Run a neutral version beside the dramatic concept when possible. The difference estimates whether the positioning changes behavior rather than merely whether the category has demand.

Set a stop rule as carefully as a success gate. For example: after $2,500 or 50,000 targeted impressions, stop if fewer than 100 people accept the stated price and fewer than three relevant buyers request samples. A stop rule protects a founder from extending the test until random variation produces a flattering result. If it fails, interview ten people who watched but did not act. Revise the promise or segment before changing the factory plan. The cheap test earns its value when the founder is willing to believe a negative answer.

A cheap demand test earns its value when the founder is willing to believe a disappointing answer.