38 SKUs Tested, 6 Winners: How to Run a Product Test-and-Learn Machine in DTC

Key Takeaways

The standard DTC playbook says: find a product, build a brand, scale it.
Research from CB Insights shows that 35% of startups fail because there is no market need for their product.
Products deployed in waves across 28 months.
Product-market fit is not a single discovery -- it is a portfolio exercise.

The Setup

The standard DTC playbook says: find a product, build a brand, scale it. That playbook produces one of two outcomes -- a hit or a hole in your bank account. There is no middle ground because there is no testing discipline. Every launch is a commitment.

The operators who consistently win in DTC are not better at picking products. They are better at testing products -- running cheap, fast experiments across a portfolio of SKUs, measuring traction within 30-60 days, and killing losers before they burn cash. The difference is not instinct. It is infrastructure. Can your stack support a new product test at near-zero marginal cost? If not, every launch is a bet. If yes, every launch is a data point.

Most DTC operators test 2-5 products before finding something that works. The ones running real test-and-learn infrastructure test at an order of magnitude above that -- and their hit rates tell a different story about what product-market fit actually looks like in practice.

What the Data Shows

Research from CB Insights shows that 35% of startups fail because there is no market need for their product. In DTC specifically, a 2023 Jungle Scout survey found that the average seller tests 3-4 products before finding one that sustains above $10K/month. Shopify's internal data suggests that DTC brands with 5+ SKUs have 2.3x the revenue stability of single-product brands.

Between October 2023 and January 2026, the Stealth Labz portfolio tested 38 distinct products through Konnektive CRM, generating $1,075,946 in gross revenue across 28 months. The segmentation was clear:

Tier 1 (>$25K net): 6 products, 94.5% of net revenue (~$888K)
Tier 2 ($1K-$25K net): 9 products, 5.0% of net revenue (~$47K)
Tier 3 (<$1K, micro-tests): 23 products, 0.5% of net revenue (~$4.3K), averaging $56 each

The 6 winners tell distinct stories. PRD-01 was a flash-scale product -- $509,821 net, peaking at $173K in February 2024, then declining when affiliate traffic dried up. PRD-02 followed the same pattern: $128,794 net, peak month of $93K. PRD-03 was a late-stage breakout -- launched August 2025, hitting $100,909 net in five months with a $59K peak. PRD-04 produced $71,792 net. PRD-06 and PRD-05 delivered steady, consistent monthly revenue at $48,280 and $28,574 respectively.

The 23 micro-tests averaged $56 each. They were killed fast and cost almost nothing because every test ran on shared infrastructure -- same Konnektive CRM, same payment processing, same affiliate tracking, same fulfillment pipeline.

How It Works

Products deployed in waves across 28 months. The pattern: launch 2-4 products in a window, measure results within 30-60 days, scale what shows traction, kill the rest. Then repeat.

The discipline is in what the operator does not do. No extended branding exercises for unproven products. No custom infrastructure per SKU. No emotional attachment to products that are not converting. If a product does not show traction in the measurement window, it gets killed. If it shows traction, it gets fed -- more traffic, more affiliate partners, more budget.

Because all 38 products ran on shared infrastructure, the marginal cost of each test was near zero. Adding a new SKU to Konnektive, setting up a campaign, and routing affiliate traffic does not require rebuilding the stack. It requires configuring the stack that already exists. That configuration cost -- measured in hours, not thousands of dollars -- is what makes a 38-product test velocity possible.

The data also reveals three distinct product lifecycle patterns that inform future launches: flash-scale products (PRD-01, PRD-02) that peak fast on affiliate traffic and decline when the traffic stops; steady growers (PRD-06, PRD-05) with consistent monthly revenue that compounds over time; and late-stage breakouts (PRD-03) that emerge months into the portfolio when the operator's testing instincts and infrastructure are sharpest.

What This Means for DTC Operators

Product-market fit is not a single discovery -- it is a portfolio exercise. The operator who tests 38 products and scales 6 has a fundamentally different risk profile than the operator who builds one product and hopes.

The math is straightforward: 6 of 38 products drove 94.5% of net revenue. That 15.8% hit rate is not a failure -- it is the test-and-learn discipline working exactly as designed. The 23 micro-tests are not wasted CPL spend. They are the cost of finding the 6 winners that generated $888K. And because every test ran on shared infrastructure, that cost was measured in configuration hours, not capital outlay.

If your current stack requires a custom build for every new product test, you are not set up to find winners. You are set up to make bets. Build the infrastructure once, test at velocity, scale what converts, kill what does not. That is how you turn product launches from events into an engine.

References

CB Insights (2023). "Top Reasons Startups Fail." Market need analysis across startup failure modes.
Jungle Scout (2023). "Seller Survey." Average product test velocity for DTC sellers.
Shopify (2024). "Commerce Trends." Revenue stability by SKU count.
Keating, M.G. (2026). "Case Study: The Product Launch Engine." Stealth Labz. Read case study
Keating, M.G. (2026). "The Compounding Execution Method: Complete Technical Documentation." Stealth Labz. Browse papers