Amazon advertising has changed so much since 2017 that many campaign playbooks now feel like they were written for a different sport. In 2017, a marketplace advertiser could still treat Amazon mostly as a high-intent PPC channel: find the keyword, bid carefully, harvest search terms, protect branded traffic, and keep ACOS somewhere finance could tolerate.
That world has not disappeared. Sponsored Products are still the workhorse. But the operating model around them has changed completely. Amazon Ads moved from a seller tool into a retail media ecosystem with Sponsored Brands, Sponsored Display, DSP, video, streaming TV, new-to-brand reporting, AI campaign tooling and enough acronyms to make a sensible operator quietly open a spreadsheet and breathe into it.
The named mistake I see is timeline blindness: brands study the history of Amazon advertising as a feature timeline, then manage the account as if every new feature deserves budget. It does not. A new ad format is not a strategy. A new dashboard is not a margin model. And a bigger retail media market does not make an unprofitable SKU more deserving of clicks.
My stance is simple: the real lesson from Amazon advertising history is not that brands should use every new format. The lesson is that each wave of sophistication created one extra decision layer. First keyword control. Then brand defence. Then audience reach. Then incrementality. Now profit permission across Amazon, bol and MediaMarkt. If your operating model did not evolve with the channel, your campaigns may look modern while your P&L is still stuck in 2017.
What the existing history pieces explain well
The best competitor content covers the product timeline clearly. Podean’s history article highlights the 2017 and 2018 moment when Amazon advertising became hard to ignore: personalized display, video ads, the 2018 rebrand into one Amazon Advertising umbrella, and the move from names like AMG, AMS and AAP toward Sponsored Brands, Advertising Console and Amazon DSP. It also points to Amazon’s reported $10.1 billion advertising revenue for 2018.
Omnitail is useful because it untangles the naming mess. It explains how Amazon Ads became the umbrella, with Amazon Ads Console for PPC and Amazon DSP for programmatic reach. It also notes the March 2020 shift where ad management for sellers moved into the Amazon Ads Console, creating a more unified workflow for vendors and sellers.
BidX focuses on the revenue curve. Its 2022 analysis notes that Amazon disclosed advertising services revenue separately in Q4 2021, covering sponsored ads, display and video advertising, and that Amazon’s 2021 advertising revenue reached about $31.16 billion, ahead of YouTube’s reported $28.85 billion that year. That is the moment many operators stopped seeing Amazon Ads as “extra visibility” and started seeing it as a market in its own right.
DEPT adds the full-funnel lens: Sponsored Products, Sponsored Brands and Sponsored Display behave more like search and retail media, while DSP uses Amazon’s audience data to reach people on and off Amazon. Hightouch’s bol.com case study widens the frame for Benelux teams: bol sells sponsored products, brand advertising, performance marketing and off-platform advertising, serving more than 13 million active customers and over 52,000 partners, with first-party data improving audience reach by 109% and brand-ad CTR by 33%.
Reddit adds the part polished guides usually miss: seller pain. Threads are full of operators saying PPC keywords became expensive, TACoS climbed, agencies improved reports without improving profit, and products that looked profitable before launch stopped working once advertising costs were added. One seller described roughly $15,000 monthly spend with ACOS and TACoS creeping up after outsourcing. Another warned that items can look profitable until PPC costs are included in the bottom line.
The gap is obvious. Most content explains how Amazon Ads became bigger and more complex. Less content explains how operators should change decision-making when an ad platform becomes a retail media tax on marketplace growth.
The missed angle: every era changed the permission to spend
Here is the operator version of the history.
2017-2018 was the control era. Amazon advertising became more structured, more visible and more manageable. Sponsored Products recommendations, rebranding, consolidated metrics and clearer product names made it easier for sellers to operate campaigns.
2019-2021 was the dependency era. Advertising became less optional. More brands entered auctions, CPCs increased in many categories, and sellers learned that organic rank often needed paid fuel. Sponsored Products were no longer a launch lever only; they became part of ongoing marketplace rent.
2022-2024 was the retail media era. Amazon Ads became a full media business, while bol, Walmart, MediaMarkt and other retailers matured their own propositions. Brands started comparing budgets across retailer media networks, not just campaigns inside one platform.
2025 onward is the profit-permission era. AI tooling, unified campaign workflows, DSP access and better audience products make it easier to spend. That is exactly why brands need stricter rules on where spend is allowed.
The uncomfortable truth: ad platforms usually improve faster than finance workflows. Amazon can add new placements, bol can mature first-party targeting and MediaMarkt can package electronics audiences, but your team may still be deciding budgets from platform ROAS screenshots. That mismatch is where margin leaks.
What changed commercially after 2017
The biggest change is not the number of ad formats. It is the number of questions a good advertiser must answer before raising budget.
In a simple Sponsored Products world, the key questions were: does the search term convert, is ACOS acceptable, and can we scale without breaking the campaign? Today, those questions are not enough. You also need to know whether the SKU has enough contribution margin, whether stock cover can handle extra demand, whether the sale is incremental, whether branded spend is defending or cannibalising, whether DSP is feeding future demand, and whether another marketplace could use the same euro better.
That is why FiveX connects advertising data with marketplace analytics, profitability dashboards, inventory insights and AI recommendations. The point is not to make an operator stare at more numbers. The point is to stop treating campaign performance as separate from unit economics.
Scenario 1: the 2017-style Amazon account that looked efficient and lost profit
A Dutch home brand spends €18,000 per month on Amazon Ads. The account is tidy: Sponsored Products split by exact, phrase and auto campaigns; Sponsored Brands for hero keywords; branded defence separated. The blended ACOS is 21% on €85,700 attributed sales. A 2017-style audit would call this healthy.
Then the SKU economics are mapped. The top-selling kitchen organiser sells for €29,95. Referral and fulfilment fees are €9,10, landed cost is €11,40, average coupon impact is €1,20 and return reserve is €0,85. Contribution before ads is €7,40, or 24,7%. At 21% ACOS, only 3,7% remains before overhead. Worse, non-brand campaigns on that SKU run at 29% ACOS because competitors keep pushing CPCs.
The decision is not “Amazon Ads is bad”. The decision is to change permission. Branded defence stays live with a tight budget. Exact non-brand terms with proven basket value stay. Broad discovery stops until the SKU price or bundle economics improve. FiveX would flag this as a margin-permission issue: the campaign is not inefficient by platform standards, but the SKU cannot afford the traffic mix.
Scenario 2: the full-funnel budget that needed incrementality rules
A Belgian beauty brand adds Sponsored Brands video and DSP retargeting after seeing Amazon’s full-funnel options mature. Monthly spend rises from €9,000 to €14,500. Platform ROAS falls from 5,8 to 4,1, which makes the PPC team nervous. But total Amazon sales rise from €62,000 to €81,000 and branded search volume grows. So is the new budget good or bad? Annoyingly, both answers are possible.
The brand runs a simple operating test. It caps DSP at €2,500 for four weeks, keeps Sponsored Brands video at €1,800, and measures total sales, new-to-brand orders, organic rank and contribution margin after ads. The hero serum has 38% contribution before ads and 46 days of stock. The travel-size product has only 16% margin and 19 days of stock.
The result: Sponsored Brands video remains funded for the serum because it lifts new-to-brand orders and keeps post-ad contribution above 12%. DSP retargeting is excluded from the travel-size SKU because the extra demand moves a low-margin product with fragile stock. This is the modern lesson: full-funnel advertising is useful only when SKU-level permission travels with the funnel.
Scenario 3: Benelux budget moved from Amazon to bol and MediaMarkt
A consumer electronics seller spends €22,000 per month across Amazon.nl, bol and MediaMarkt. Amazon gets €13,000 because it has the most mature campaign tooling. bol gets €6,000. MediaMarkt gets €3,000 for selected retail media placements.
The dashboard says Amazon ROAS is 6,2, bol ROAS is 5,4 and MediaMarkt ROAS is 4,8. If the team ranks channels by ROAS, Amazon keeps the money. But FiveX-style profit analysis changes the view. The Amazon hero accessory has 18% contribution before ads and high price pressure. The bol bundle has 31% contribution, lower return cost and 52 days stock. The MediaMarkt laptop add-on has 28% contribution and converts better when promoted next to relevant electronics categories.
The operator move: shift €2,500 from Amazon generic keywords to bol Sponsored Products for the bundle and €1,200 to MediaMarkt category placements. Amazon still matters, but it stops being the default winner. After four weeks, blended ROAS falls slightly from 5,8 to 5,5, yet total contribution after ads rises from €11,900 to €13,450. That is the trade-off many ad teams must learn: lower platform ROAS can be the price of higher marketplace profit.
The five rules for managing Amazon Ads after the history lesson
1. Separate legacy metrics from operating metrics
ACOS, ROAS and CPC are useful, but they are not final decision metrics. The operating layer should include contribution margin after ads, break-even ACOS, TACoS, stock cover, return rate, price position and channel role. If those numbers are not visible beside campaigns, your team will optimise what the platform shows instead of what the business needs.
2. Give every SKU a permission label
Use simple labels: Scale, Harvest, Protect, Fix or Stop. A high-margin SKU with 60 days of stock can scale. A low-margin product with ranking value may harvest exact terms only. A product with 12 days of stock should protect visibility, not accelerate demand. FiveX can help make these labels dynamic by combining ads, stock, marketplace performance and margin data.
3. Treat full-funnel formats as jobs, not trophies
Sponsored Brands, Sponsored Display, DSP and video are not automatically “advanced”. They are useful when the job is clear: defend brand demand, launch a product, fill retargeting pools, test new audiences or support a retail moment. If a format cannot explain its job in commercial language, it probably does not deserve budget yet.
4. Compare marketplaces by profit capacity
Amazon may have richer advertising options than bol or MediaMarkt, but richer tooling does not guarantee richer profit. For NL/BE sellers, budget should move toward the marketplace-SKU combination with the best available contribution, stock cover and strategic role. This is where cross-marketplace analytics beats platform-native reporting.
5. Automate the guardrails before the bids
AI recommendations and bidding automation are powerful, but only after the business rules are clear. Automating bids without SKU margin, inventory and channel priorities is just a faster way to make the wrong decision. Automate budget caps, negative-margin flags, stock-cover rules and break-even thresholds first. Then let bid automation work inside the rails.
Conclusion: Amazon Ads history is really a warning about operating models
Since 2017, Amazon advertising has become bigger, smarter and more full-funnel. That growth created enormous opportunity for brands. It also created a quiet trap: the easier it becomes to buy visibility, the easier it becomes to buy unprofitable growth.
The winning teams will not be the ones that simply adopt every new Amazon Ads feature first. They will be the teams that translate every new feature into a sharper operating question: which SKU, which marketplace, which margin, which stock position and which customer journey role justifies this spend?
That is the Advertentie Service standard we believe in at FiveX. Manage Amazon, bol and MediaMarkt advertising as part of the full marketplace P&L. Let platform metrics inform the decision, but let profit, inventory and channel strategy make the decision. History is useful. Margin is kinder.