Insights

experiments

A new brand instead of a discount. How to find out in eight weeks whether a different customer will buy the same product

Sep 15, 2026 · 9 min read · Human

[post image]

Royal Vision had great dog beds and a price war on its hands. Instead of rebuilding the brand, we launched LivaMio alongside it. Here’s the experiment model behind that decision.

When margins fall, the first reflex is to cut the price. The second is to spend more on ads. Both work for a short while, and both damage the numbers for much longer. There’s a third way, rarely chosen because it seems more expensive and riskier: test whether a different customer will buy the same product at a different price, because they’re buying something different.

Royal Vision makes beds for dogs and cats. Good ones, made in Poland, from certified fabrics, with orthopedic inserts. Its Royal Pet brand sells them to customers who are looking for a pet bed: they compare sizes, read about washing, look at the price. But we noticed a second group of people. People furnishing their living room who want the dog’s bed to match the sofa. They aren’t looking for a pet bed. They’re looking for furniture. They speak a different language, shop in different places, and are willing to pay more.

Why not just change Royal Pet

Because Royal Pet works and has its own customers. If we had rebuilt its messaging for interior lovers, we might have gained new ones, but we would certainly have confused the existing ones. Rebuilding a brand that works is the most expensive experiment imaginable, because it tests a hypothesis on the entire business at once. Instead, we set up a second brand alongside it: LivaMio, with its own voice, photos, a Shopify store and campaigns aimed at people furnishing their homes.

The experiment model

  1. Hypothesis. One sentence that can be disproved: “Customers furnishing their interiors will buy the same bed at a higher price if they see it as part of their home décor.”
  2. MVP. Only what’s needed to test the hypothesis: a brand, a few dozen products, a store, content and campaigns. No full catalog, no loyalty program, no marketplaces.
  3. Traffic and data. Meta and TikTok campaigns aimed at real customers. We don’t measure likes, we measure acquisition cost, conversion, average order value and margin after costs.
  4. Decision. After six to eight weeks, one of three: scale, refine and repeat, or stop. Each is a good outcome, because each is a decision based on data, not on gut feeling.

Launching an experiment like this takes four to six weeks, and validation another six to eight. The cost is fixed and known upfront, because the scope is limited by design. That matters: an experiment that’s allowed to grow along the way stops being an experiment and becomes a never-ending project.

What was hardest

Not the technology. A Shopify store and AI content come together quickly. The hardest part was staying disciplined: not adding more products, not widening the audience, not fixing everything at once when the first data was inconclusive. The temptation to “just add one more thing” is enormous, and every addition blurs the answer to the question we asked at the start.

The second challenge was honest accounting. A new brand has costs you don’t see in the ad dashboard: team time, content, a separate data setup, customer service. We counted them, because without them the ROAS would have looked beautiful and the bottom line wouldn’t.

What we want to achieve and how we’ll know

An experiment only makes sense if the goals are written down before launch. For LivaMio, we set them as follows:

  • Average order value at least 15% higher than at Royal Pet. This tests the main hypothesis: an interior-minded customer pays more for the same product.
  • Positive contribution margin from campaigns by week eight at the latest, after deducting goods, shipping, returns, advertising and management.
  • Customer acquisition cost no higher than PLN 110, with store conversion of at least 1.5%.
  • New customers making up more than 80% of LivaMio sales, to make sure the new brand isn’t cannibalizing Royal Pet.
  • No drop in Royal Pet sales during the test. If they fall, the hypothesis about a different customer is false.

Result after 8 weeks: iterate. Values achieved: AOV PLN 449, conversion 1.6%, CAC PLN 118, contribution margin after advertising PLN 22 per order.

When this model makes sense

  • You have a good product and a price war in your category.
  • You suspect a different group of customers would pay more, but you have no proof.
  • You don’t want to put a brand that’s making money today at risk.
  • You have suppliers, know-how and products a competitor can’t build in eight weeks.

If those four sentences describe your company, you don’t need another discount. You need a hypothesis, eight weeks, and the courage to make the decision the data points to at the end.

Running an online store?Let’s start with a 30-minute conversation about your catalog, campaigns and margin.Explore AI eCommerce

Contact

Want to earn your audience’s trust?