⏱️ Lectura: 11 min
Amazon makes nearly a billion dollars a week from search ads alone within its marketplace, according to an analysis published by writer Seth Godin on August 18, 2026. The figure is already enormous on its own, but what stands out is something else: a study cited in that same piece claims that an online store with search ads enabled sells fewer units than the same store without them.
📑 En este artículo
- TL;DR
- What happened
- The mechanism: why search ads don’t create new demand
- How the auction works under the hood
- Amazon isn’t the only one: how it compares to other marketplaces
- How to detect and measure the effect in your own searches
- The side effects: worse quality and cheaper products
- Impact and analysis
- What’s next
- Frequently Asked Questions
- References
That makes search ads a zero-sum game. They don’t create new demand, they only decide who captures the demand that already existed. For a developer building or integrating search and ranking systems in ecommerce, understanding this mechanism matters as much as understanding the relevance algorithm itself.
TL;DR
- Amazon brings in nearly US$1 billion a week in profit from search ads alone, according to Seth Godin.
- With that weekly profit, Amazon could give every employee a US$35,000 bonus and still have money left over.
- Sellers spend more than US$50 billion a year on this type of ads on marketplaces like Amazon.
- A study cited by Godin found that an ecommerce site with search ads sells fewer units than the same site without them.
- The most expensive ad Godin’s publisher tested cost close to US$1 per click, just to appear in the exact search for the book’s title.
- The US Federal Trade Commission sued Amazon in 2023, and one of the accusations was degrading its search quality to force more paid advertising.
What happened
The specific case that triggered Godin’s analysis is personal: his publisher started buying search ads on Amazon last week to promote his new book. The best-performing ad turned out to be one bidding on the exact name of the book itself, at close to a dollar per click. In other words: the publisher pays Amazon to show an ad to someone who was already searching for that specific book.
That pattern repeats in almost every category. If you search for an air fryer, Amazon already knows which model is best rated, with fewer returns and a better price. The only job left for the ad is to convince you to pick a different model, or to keep you focused on the one you were already planning to buy. Either way, the search stops being neutral.
The mechanism: why search ads don’t create new demand
Traditional advertising works by pushing the demand curve upward: you see an ad for something you didn’t know about, a new interest awakens, and total sales in the category go up. Search ads work differently. The user already arrived with a defined intent, they didn’t stumble on an ad in passing. The ad doesn’t generate that intent, it only competes to capture it.
When there are few ads, the maker of the best product can ignore the auction and keep selling on its own merit. When there are many ads, that same maker has to start bidding too, even if it doesn’t want to, just to avoid losing the sales it would have gotten organically. Godin sums it up by citing a study according to which an ecommerce site with search ads enabled sells fewer total units than the same site without advertising: the pie doesn’t grow, it just gets split differently, with a middleman taking a cut of every slice.
💭 Key point: when the platform already knows which is the best product for your search, the only job left for the ad is to get you to choose something else.
How the auction works under the hood
A typical search ads system combines two numbers for each bidding seller: how much they’re willing to pay per click and how relevant their product is to that query. The final ad ranking usually comes from multiplying both factors, not just taking the highest bid. That’s what lets the platform claim, with some justification, that it doesn’t simply sell the top spot to the highest bidder.
But relevance is measured using data the platform itself controls, and there’s real room to tilt the scale toward whoever pays more. The following Python simulation shows the core point of Godin’s argument: with fixed demand, turning on ads doesn’t increase total sales, it just changes who ends up with them.
import random
def simular_busqueda(usuarios, productos, con_ads):
ventas = {p: 0 for p in productos}
for _ in range(usuarios):
if con_ads:
# the ad competes with the best-rated result
elegido = random.choices(productos, weights=[2, 3, 1, 1])[0]
else:
# without ads, almost everyone picks the best-rated one
elegido = random.choices(productos, weights=[7, 1, 1, 1])[0]
ventas[elegido] += 1
return ventas
productos = ["freidora_top_resena", "freidora_marca_b", "freidora_marca_c", "freidora_marca_d"]
sin_ads = simular_busqueda(usuarios=10000, productos=productos, con_ads=False)
con_ads = simular_busqueda(usuarios=10000, productos=productos, con_ads=True)
print("Without ads:", sin_ads, "total:", sum(sin_ads.values()))
print("With ads:", con_ads, "total:", sum(con_ads.values()))
When you run it, the total number of units sold (sum(sin_ads.values()) and sum(con_ads.values())) is the same in both scenarios: 10,000. The only thing that changes is the distribution across products. That’s exactly the zero-sum effect described by the study cited by Godin: total demand doesn’t move, it just gets redistributed in favor of whoever paid for the top spot.
flowchart TD
A["User searches for 'air fryer'"] --> B["Amazon already knows the best product"]
B --> C{"Is ad budget available?"}
C -->|"Yes"| D["Real-time auction among sellers"]
C -->|"No"| E["Original organic result is shown"]
D --> F["Winning product is inserted at the top"]
F --> G["User sees ads and organic results mixed"]
E --> G
Amazon isn’t the only one: how it compares to other marketplaces
The same mechanism shows up, with variations, on almost any platform that combines search with its own advertising. What changes is the bidding model and how mixed the ads end up with the organic results.
| Platform | Bidding model | Who pays | Effect on the organic result |
|---|---|---|---|
| Amazon Sponsored Products | CPC in real-time auction | Seller | Takes the top positions, ahead of the organic ranking |
| Google Shopping Ads | CPC | Advertiser | Separate carousel, above the organic results |
| Apple App Store Search Ads | CPT (cost per tap) | Developer | Takes the first result, ahead of any organic app |
| Meta/Instagram Shopping Ads | CPM or CPC | Brand | Inserted into the feed, doesn’t compete with an equivalent organic search |
How to detect and measure the effect in your own searches
You don’t need special access to see this pattern in action. Any developer can inspect an Amazon results page and count how many of the top spots are ads.
import requests
from bs4 import BeautifulSoup
headers = {"User-Agent": "Mozilla/5.0 (compatible; investigacion-personal/1.0)"}
resp = requests.get("https://www.amazon.com/s?k=freidora+de+aire", headers=headers, timeout=10)
soup = BeautifulSoup(resp.text, "html.parser")
resultados = soup.select('[data-component-type="s-search-result"]')
patrocinados = [r for r in resultados if r.select_one('[data-component-type="sp-sponsored-result"]')]
print(f"Total results: {len(resultados)}")
print(f"Sponsored results: {len(patrocinados)}")
if resultados:
print(f"Sponsored percentage: {len(patrocinados) / len(resultados) * 100:.1f}%")
To verify the result without writing code, just open the browser’s developer tools (F12), go to the rendered HTML, and look for the attribute data-component-type="sp-sponsored-result". Any element that has it is an ad, even if it’s visually mixed in among the best-rated products.
⚠️ Heads up: scraping Amazon results at scale can violate its terms of service. For one-off tests, manually inspecting a few searches is enough; for anything sustained over time, the correct alternative is the official Product Advertising API.
The side effects: worse quality and cheaper products
Godin points to two consequences that go beyond the buyer’s wallet. The first: if brand reputation matters less than the budget for bidding on clicks, some manufacturers prefer to lower product quality to have more margin and fund more ads. The second is even more uncomfortable: the platform itself has an economic incentive to make its organic search a little worse, because that pushes more sellers to buy ads to compensate.
Neither consequence is illegal on its own. But together they explain why, for years, Amazon built its reputation by lowering prices and opening the door to small sellers, and why that reputation is starting to crack now that internal advertising has become a revenue source as large as direct sales.
Impact and analysis
In the United States, the Federal Trade Commission sued Amazon in 2023 over alleged monopolistic practices, and part of the claim points precisely to the company degrading its search relevance to push more paid advertising. The case is still working its way through the courts and there’s no final ruling yet, but it confirms that the tension between neutral search and monetized search is no longer just an academic discussion.
For a team building its own marketplace or internal search engine, the technical lesson is concrete: every percentage point of ad inventory added to a results page reduces, by the same proportion, the space available for the organic ranking. That trade-off can be measured and decided explicitly, instead of letting it grow just because it generates short-term revenue.
What’s next
Godin isn’t asking for search ads to be banned, he’s asking for this to be called what it is: not a way to create demand, but a way to redistribute it while taking a cut. As the FTC case moves forward and other platforms (Google, Apple, Meta) face similar regulatory pressure over their own search ad models, the debate is likely to move from marketing blogs to the legal and product teams of every company that runs its own marketplace.
📖 Summary on Telegram: View summary
Try it yourself: open any search on Amazon, turn on the developer tools, and count how many of the first five results carry the sp-sponsored-result attribute.
Frequently Asked Questions
What are zero-sum search ads?
They’re ads that compete for demand that already existed at the moment the user typed their search, instead of generating new demand. The result is that they don’t increase total sales in the category, they only decide which seller gets each sale.
Does Amazon increase its total sales because of these ads?
According to the study cited by Seth Godin, no: an ecommerce site with search ads enabled sells fewer units than the same site without them. Amazon does increase its revenue, though, because it charges for every click regardless of whether total sales go up or down.
How do I know if an Amazon result is an ad?
Look for the visible “Sponsored” label above the product, or inspect the HTML for the attribute data-component-type="sp-sponsored-result", which marks every paid result on the page.
Does this model apply only to Amazon?
No. Google Shopping Ads, Apple App Store Search Ads, and Meta’s shopping tools all run on the same underlying logic, though with different bidding models, as shown in this article’s comparison table.
What can sellers do about this?
Most end up bidding defensively: not because they expect to sell more, but to avoid losing the organic sales they already had before ads showed up in their category.
Are there regulations around this?
In the United States, the FTC’s 2023 lawsuit against Amazon includes accusations related to search quality and internal advertising, although the case still doesn’t have a final resolution.
References
- Seth’s Blog: “The Amazon tax”: the original analysis exposing the weekly ad profit figures and the study on total sales.
- Federal Trade Commission: information about the antitrust case against Amazon filed in 2023.
- Wikipedia: Amazon (company): general context on the marketplace’s business model.
- Google Ads: reference for the CPC auction model used as a comparison in this article.
📱 Do you like this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day.
0 Comments