There’s a moment every Amazon seller knows well. You’ve been researching for days, maybe weeks. Search volume, competitors, margins, advertising costs, suppliers, reviews. Everything looks encouraging. You lean back and think: I think I found it.
For most sellers, that’s the finish line. For experienced operators, it’s where the real work begins.
Identifying a promising opportunity is surprisingly easy. Understanding whether it deserves years of your company’s attention is much harder.
That gap explains why experienced businesses pass on products that look extraordinary, and occasionally pursue opportunities everyone else ignores.
Product research has two stages
- Does demand exist?
- Can we manufacture it?
- Can we make money?
- Can we compete?
- What assumptions make this look attractive?
- Which of those assumptions are fragile?
- What happens if they change?
- Can the business still succeed?
Imagine buying a house
You visit a beautiful property. The kitchen is perfect, the location excellent, the price reasonable. Would you immediately sign? Probably not. You’d inspect the foundation, review the roof, check zoning, investigate flood risk, examine the plumbing, study nearby developments.
In other words, once you decide you like the house, you start looking for reasons not to buy it.
Amazon sellers usually do the opposite. Once they like a product, they start collecting evidence that confirms the excitement.
Confirmation bias is the hidden enemy
Once we reach a conclusion, we naturally seek information supporting it. We discount conflicting evidence, reinterpret warning signs and become increasingly confident.
Product research is especially vulnerable. Once you’ve decided to launch, growing search volume feels significant. Positive reviews get more attention. Strong margins appear sustainable. Supplier promises become believable. Negative signals seem temporary.
Nothing about the market. Only your interpretation of it.
Great product research tries to kill the idea
This sounds pessimistic. It isn’t. Imagine a venture firm evaluating an investment. Do they spend weeks convincing themselves to invest? No — their process is designed to eliminate opportunities, and only the strongest survive.
The objective isn’t proving the product is good. It’s discovering whether it deserves to survive rigorous scrutiny. Products that remain attractive after difficult questions are usually far better investments.
Ask the opposite question
“Why will this succeed?”
“Why might this fail?”
That single change transforms the investigation. Instead of searching for supporting evidence, you start searching for weak assumptions:
Those questions rarely appear in beginner research guides, yet they often determine the outcome.
Imagine explaining the failure five years from now
This is one of the most powerful exercises any founder can perform. Pretend the launch failed. Five years have passed. Someone asks what happened. Write the story.
Perhaps demand slowed. Perhaps quality problems emerged. Perhaps competitors entered aggressively. Perhaps inventory became unmanageable. Perhaps regulation changed. Perhaps working capital disappeared. Perhaps the category simply became less attractive.
Which of those risks can we investigate today? The exercise forces your mind to search forward instead of backward.
Every great opportunity depends on assumptions
Suppose your model projects exceptional profits. Now list every assumption holding those projections up: advertising costs stay stable, conversion stays high, suppliers maintain pricing, demand keeps growing, inventory arrives on time, reviews stay positive, platform policies stay unchanged.
The product doesn’t create profitability. Those assumptions do — and understanding them is worth more than refining another spreadsheet.
The strongest products have multiple ways to win
One opportunity depends on precision. The other depends on resilience. Which would you rather own?
Second-order thinking
Most research stops at first-order effects: we launch the product. Experienced founders keep asking.
- We launch the product. Then what?
- Sales increase. Then what?
- Inventory requirements double. Then what?
- Cash flow tightens. Then what?
- We delay future launches. Then what?
- Competitors gain share elsewhere.
Imagine building a bridge
Engineers don’t design bridges for average traffic. They design for unusual events: heavy loads, strong winds, unexpected conditions. A bridge isn’t judged by how it performs on ordinary days, but by how it behaves under stress.
Product research deserves the same philosophy. Average conditions rarely destroy businesses. Unexpected ones do.
The cost of being wrong matters more than being right
Suppose two launches each have a 70% probability of success. One failure costs $50,000. The other costs $700,000. The probability is identical. The decision is not.
Expected outcomes matter; consequences matter more. Experienced operators think continuously about asymmetry — limited downside, meaningful upside. Those are the opportunities worth pursuing.
Imagine the product already exists
Forget launch day and fast-forward three years. The product is successful. Now ask: has it strengthened supplier relationships? Improved the brand? Increased customer lifetime value? Created adjacent opportunities? Become easier to operate?
Or has it simply created another busy department? Long-term questions often reveal weaknesses invisible during launch planning.
Curiosity is more valuable than confidence
Beginners celebrate confidence. Experienced founders become suspicious of it, because confidence often means the investigation stopped too early.
Curiosity keeps asking. What haven’t we considered? Who disagrees with us? What assumptions feel obvious? What data would change our minds?
Confidence closes research. Curiosity extends it. One builds certainty, the other builds understanding.
The dashboard we actually need
Imagine your research platform changing the moment you identify a promising opportunity — instead of ending the analysis, it begins a second phase:
- Challenge every assumption
- Model worst-case scenarios
- Estimate capital sensitivity
- Evaluate supplier dependence
- Test advertising resilience
- Analyse customer behaviour changes
- Measure concentration risk
- Calculate opportunity cost
Software that acts less like a search engine and more like an experienced board of directors.
Great businesses are built on better questions
Technology will keep improving. Search estimates will get more accurate. AI will identify opportunities faster. Competitor analysis will become automated. Finding promising products will become easier every year.
Which means competitive advantage will increasingly come from somewhere else: judgement, discipline, critical thinking, the willingness to investigate beyond the obvious, the courage to reject attractive opportunities. Those are remarkably difficult to automate.
Final thoughts
Finding a promising product is exciting. It feels like progress, and often it is. But excitement can be dangerous.
The moment you become convinced you’ve found a winner is precisely when your thinking becomes most vulnerable. You start seeing confirmation instead of contradiction, possibility instead of uncertainty, optimism instead of risk.
The strongest operators deliberately resist that instinct. They slow down. They challenge themselves. They invite disagreement. They search for reasons the idea might fail — not because they enjoy pessimism, but because extraordinary businesses are rarely built by asking easier questions.
Not “should we launch this?” but “what would have to be true for this product to still be a great decision five years from now?”
That shifts the purpose of product research: from finding products that look promising, to identifying businesses that remain exceptional after reality begins challenging every assumption. And that is where truly great product research begins.
This closes the Product Research set and pairs most closely with product research is really risk research — that piece is what to look for, this one is when to look. See also the wrong data, overestimating TAM and capital allocation.