Strategy
Why Amazon Brands Fail at Scale
Key Takeaways
- More choice does not automatically reduce conversion. A well-known jam experiment found fewer purchases from a 24-option display than a 6-option display, but a later meta-analysis found no average choice-overload effect across 63 experimental conditions. Context and choice difficulty matter.
- Miller’s “seven, plus or minus two” is not an Amazon SKU limit. The 1956 paper examined immediate memory and information-processing tasks. It supports reducing unnecessary complexity, not a rule that shoppers can evaluate only a fixed number of products.
- Shoppers often use shortcuts rather than exhaustively comparing every option. Herbert Simon’s work on bounded rationality and satisficing is a useful lens for making product roles and tradeoffs easier to understand.
- Catalog structure is an operating decision. Product overlap, unclear variations, weak margins, and competing ad targets should be diagnosed with account data rather than attributed to one universal psychology rule.
What Is Catalog Architecture (And Why Should You Care)?
Catalog architecture is how your products are structured, positioned, and connected to each other on Amazon. It’s the organizational layer that determines:
- Which products shoppers find when they search
- How product relationships and variations are presented
- Whether campaigns support distinct product roles or compete for the same demand
- Which products carry growth and which put pressure on margins
Most Amazon brands treat their product catalog as a list. Add a new product, create a listing, run some ads. But a growing catalog creates dependencies: variation structure affects navigation, positioning affects conversion, and product economics affect how much traffic a SKU can profitably acquire.
This is the second pillar of the AMDC Growth System because the other work depends on it. Listing priorities require clear product roles. PPC decisions require SKU-level economics. Margin decisions require enough visibility to separate volume from profitable contribution.
What Does the Research Say About Choice and Conversion?
The Jam Study: A Specific Result, Not a Universal Rule
Sheena Iyengar and Mark Lepper published “When Choice Is Demotivating” in the Journal of Personality and Social Psychology in 2000. In the first of three experiments, researchers used a tasting booth at an upscale grocery store. The booth displayed either 24 jams or 6 jams.
The observed result in that setting was:
- 24 displayed jams: 60% of passersby stopped; 3% of those who stopped later purchased
- 6 displayed jams: 40% stopped; 30% of those who stopped later purchased
That experiment shows that a larger display can attract attention while shoppers who stop are less likely to buy under some conditions. Everyone who bought jam later encountered the full assortment on the store shelf, a further reason not to treat the booth result as an Amazon conversion forecast. The study does not establish that reducing options produces the same effect in every category or interface.
A later meta-analysis of 63 conditions from 50 published and unpublished experiments found a mean effect size near zero and substantial variation between studies. Some experiments found choice overload, while others found no effect or an advantage from more choice. The useful operating question is therefore not “How many options are too many?” It is “Can a shopper understand the differences and choose with confidence?”
How Does This Apply to Amazon?
On Amazon, the research suggests hypotheses to test rather than fixed SKU rules:
Product variations: A large set of size, color, material, or style options may be useful when the distinctions match how shoppers buy. It becomes a problem when labels, images, or grouping make those distinctions hard to parse.
Overlapping SKUs: Similar products at similar prices can compete for the same searches without giving shoppers a clear reason to choose one. The remedy may be sharper positioning, a clearer upgrade path, a different variation structure, or consolidation. Account and customer data should determine which.
Parent-child structures: Variations should represent meaningful versions of the same core product and help shoppers locate the right option. A parent that combines products with unclear differences can make comparison harder and complicate performance analysis.
For the allergy bedding brand I worked with for more than five years, catalog review was part of broader account management. That experience informs the diagnostic approach here; it is not a controlled test of choice overload.
What Miller’s “Seven” Does and Does Not Mean
George Miller’s 1956 paper, “The Magical Number Seven, Plus or Minus Two,” reviewed experiments on absolute judgment and immediate memory. Miller discussed limits that varied with the task and the way information was grouped or “chunked.” He did not test e-commerce catalogs, Amazon search results, or an ideal number of SKUs.
The practical lesson is narrower: organize information so shoppers do not have to decode unnecessary distinctions. Use consistent variation names, make product roles explicit, and group related options in a way that reflects the buying decision. Treat any proposed catalog change as a hypothesis and check the result in traffic, conversion, returns, and contribution margin.
Satisficing: Designing for a Workable Decision
Herbert Simon received the 1978 economics prize for research into decision-making in economic organizations. His work on bounded rationality challenged models that assume decision makers have unlimited information and computational capacity. His prize lecture describes satisficing models as ways to reach good-enough decisions at reasonable computational cost.
For Amazon operators, this is a useful design lens. A shopper may narrow the field with a few requirements rather than calculate the best product across every attribute. A listing and catalog should therefore make the relevant criteria easy to see:
- Which use case does this product serve?
- How does it differ from the adjacent option?
- What tradeoff explains the price difference?
- Which size, material, or configuration fits the shopper’s need?
Those questions do not prove how any individual shopper will behave. They give the team concrete points to test in search-term data, listing behavior, customer questions, reviews, and returns.
How Do You Know If Your Catalog Architecture Needs Work?
Five Warning Signs
1. Cannibalizing ad spend: Multiple similar products target the same searches, but the account lacks a clear reason for directing traffic to each one.
2. Unclear product differentiation: The team cannot state who each product is for, what problem it solves, and why it exists beside the next SKU.
3. Thin margins on high-volume products: A best seller creates revenue but contributes little after advertising, Amazon fees, returns, and product costs.
4. Search and conversion mismatch: A product receives traffic for terms that do not match its offer, or closely related SKUs show materially different conversion patterns without a clear explanation.
5. Variation confusion: Customer questions, reviews, returns, or support contacts indicate that shoppers selected a different version than they expected.
The Structural Assessment
When I assess a brand’s catalog, the first question isn’t “Which products sell the most?” It’s “What role should each product play, and does the data support that role?”
The catalog assessment maps:
- Search demand: What are shoppers looking for in the category?
- Competitive positioning: Where does each product sit in the market?
- Margin viability: Can each product sustain profitable growth at its current price and cost structure?
- Internal overlap: Are products targeting the same demand without a useful distinction?
- Growth leverage: Which products merit more investment based on evidence rather than volume alone?
This framework was part of the Milliard Brands engagement. AMDC’s case study reports $8.45M in total sales and a 57% organic order share for reoptimized categories from January through April 2025. Catalog work, listing improvements, PPC management, and broader account strategy happened together, so those reported outcomes should not be assigned to catalog architecture alone.
How Does Catalog Architecture Connect to Everything Else?
This is what separates the growth system approach from isolated tactics:
Catalog Architecture → Consumer Psychology: Clear product roles make it possible to write copy for a specific customer and use case. A product positioned as premium allergy protection needs a different argument from one positioned as an everyday bed cover.
Catalog Architecture → PPC Systems: Defined product roles help teams decide which searches belong to which SKUs and identify when products are competing for the same traffic. In the Surf-fur engagement, PPC and profitability work contributed to a reported 1.69% ACoS result; the case study provides the engagement context.
Catalog Architecture → Margin Protection: SKU-level contribution data helps distinguish productive volume from sales that lose money. The $10M+ allergy bedding brand transitioned from Vendor Central to Seller Central after broader operational and profitability analysis. Catalog visibility was one input to that decision.
Catalog Architecture → Organic Performance: Relevance, conversion, inventory, price, competition, and other factors can all affect organic placement. Clearer product roles can support relevance and conversion, but catalog structure alone does not establish the cause of a ranking change.
Frequently Asked Questions
How many SKUs should an Amazon brand have?
There is no research-backed universal limit. The right number depends on category demand, operational capacity, economics, and whether each product serves a distinct need. Audit overlap and profitability rather than managing toward a generic SKU count.
Should I cut products that are selling but not profitable?
First verify the full contribution economics and the product’s strategic role. Some products may support a broader assortment, acquisition strategy, or repeat purchase path. If a SKU has no credible path to positive contribution and no documented strategic role, consolidation, repositioning, repricing, or removal may be appropriate.
How do I restructure my catalog without losing existing sales momentum?
Start with diagnosis. Map traffic, conversion, contribution margin, inventory, variation relationships, and search overlap. Prioritize reversible changes, define the metric each change should affect, and review the result over a period that reflects the category’s traffic and buying cycle. Structural changes involving parent-child relationships also require careful attention to Amazon’s current policies and implementation risks.
Does catalog architecture matter for brands with fewer than 10 products?
Yes, because product roles, variation clarity, search relevance, and profitability still matter. A small catalog may be easier to inspect, but the impact of any change still has to be measured in that account rather than inferred from catalog size.