Rethinking Amazon PPC from the Ground Up — with Antonio Ziemann

For years, the gold standard in Amazon PPC management was fundamentally reactive. Advertisers pulled weekly bulk sheets, evaluated trailing 7-day or 14-day sales data, and manually tweaked bids in a desperate attempt to herd a chaotic marketplace into a neat target Advertising Cost of Sales (ACoS). It was a game of latency, where human managers played catch-up with algorithms. Today, the infrastructure of Amazon advertising has experienced a paradigm shift. The introduction of the Amazon Marketing Stream—a framework that actively pushes hourly performance data rather than waiting for API pull requests—has split the seller ecosystem into two camps: those still operating on delayed averages, and those leveraging real-time, target-based mathematics. In a recent Amazon Ads Real Talk session, AMALYZE host Chris sat down with Antonio "Tony" Ziemann, an ex-military data purist turned senior PPC strategist, to deconstruct what modern Amazon advertising actually looks like at the edge of this technological frontier.

From Military Precision to Target-Based PPC Automation

Tony Ziemann's approach to Amazon advertising is heavily colored by his background. Bringing an almost military rigor to campaign architecture, Ziemann treats PPC not as a creative dark art, but as an exercise in structural discipline, data reliance, and absolute transparency. Before migrating heavily into the AMALYZE Advertising ecosystem, Ziemann, like a lot of agency-side and freelance operators, witnessed a marketplace drowning in complex naming conventions, overlapping campaigns, and subjective bid adjustments based on "gut feelings."

What initially drew Ziemann to the AMALYZE platform wasn't just the promise of automation, but the synergy of time-saving software paired with an agnostic approach to campaign structure.

"The biggest advantage for all the sellers and operators managing accounts is that their existing structure—aside from a few critical modifier calibrations—could be absorbed and optimized in its current form," Ziemann noted. AMALYZE doesn't force a proprietary naming convention. Instead, it relies on mapping targets via drag-and-drop mechanisms or deep filtering rules. Ziemann views this as an immediate leap forward: it shifts the manager's cognitive load away from administrative campaign building and directly into strategic targeting and bid mathematics.

The Power of Hourly Amazon Marketing Stream Data

If there is a singular technological moat in the current Amazon advertising ecosystem, it is the integration of the Amazon Marketing Stream. Most legacy PPC tools operate by pulling data from the Amazon API once a day, or at best, every few hours. This means bid adjustments are made on a flattened average of the day's performance.

AMALYZE approaches this entirely differently, utilizing the stream to receive real-time, hourly data pushes. As Chris revealed during the session, AMALYZE's utilization of this data is so aggressive and effective that following last year's Prime Day, Amazon formally reached out to inquire about the methodologies driving such immense growth among AMALYZE users.

The secret? Intraday bid elasticity that mirrors consumer demand precisely when it happens.

If search volume and conversion velocity spike at 7:00 PM, a daily-average tool won't adjust bids until the next morning—missing the wave entirely. By running mathematical optimizations on an hourly basis, the tool captures intent as it scales and throttles down when demand wanes. Ziemann is quick to dismiss the notion that this can be replicated manually. "I've seen LinkedIn posts of people claiming they use the marketing stream data manually to make decisions," he laughed. "It goes beyond your capacity as a human to process that kind of data so accurately in that timeframe."

The "Hockey Stick" Problem: Why Average ACoS is a Lie

One of the most persistent illusions in Amazon PPC is the portfolio-level ACoS. Many sellers look at an ad group containing thousands of targets, see an average ACoS of 20%, and declare victory. Chris points out that this aggregated view creates a dangerous "hockey stick" curve of hidden inefficiencies.

In a typical setup managed manually or by basic rules-based tools, an advertiser might hit that 20% average because half of their targets are running at an incredibly profitable 4% ACoS, while the other half are bleeding cash at a 300% ACoS. The blend looks acceptable, but the underlying mechanics are broken.

By hard-coding a flat bid strategy to chase an aggregated average, the advertiser is starving their most profitable keywords of necessary budget, effectively handing market share to competitors. Conversely, they are subsidizing massive, inefficient spend on irrelevant long-tail search terms.

AMALYZE solves this through hyper-granular, target-level optimization. Every single search term, ASIN, or category target is assigned its own mathematical trajectory to hit the goal. If a keyword is converting at 4% ACoS against a goal of 20%, the system will aggressively bid it up, driving massive volume and attacking competitors head-on until the target normalizes at 20%. This fundamentally alters market positioning from defensive to offensive.

The Cart Value Paradox

A sophisticated bid algorithm is useless if the foundational data is misunderstood. The fundamental equation of ACoS is: Bid / (Conversion Rate * Price). Therefore, any fluctuation in the average cart value directly impacts the allowable bid.

Ziemann highlights a critical error most sellers make: grouping product variations with drastically different price points into a single ad group. If you house a €100 premium item and a €10 accessory in the same campaign, the algorithms are blind to the disparity. If a customer clicks an ad intended for the €100 item but purchases the €10 item, the conversion registers, but the ACoS skyrockets.

If the system reacts to this conversion by raising the bid on the keyword, it creates a catastrophic loop. You are now entering the auction with premium-level bids, but converting on low-margin accessories. Ziemann forces a strict segmentation model, advising sellers to ruthlessly parse their products by price point and intent, ensuring that the ACoS targets mathematically align with the actual revenue generated per click.

Time Travel in PPC: Dynamic Lookback Windows

Amazon is a deeply cyclical marketplace. Products live and die by seasonal trends—from summer grilling equipment to Valentine's Day gifts. The historical standard for PPC tools is to look at a trailing 30 or 60 days of data to make bid decisions. But what happens to a Christmas advent calendar that hasn't sold since last December?

Most tools treat it as a new product in November, starting bids low and forcing the seller to pay an "optimization tax" while the algorithm slowly gathers fresh data to relearn what it already knew.

AMALYZE bypasses this by utilizing up to 365 days of historical data. Moreover, it introduces the concept of Dynamic and Static Lookback Windows. If you are launching Easter products, you can explicitly instruct the system to look exclusively at the 2.5 months of data from last year's Easter season.

"It's insane," Chris noted. "The bids instantly revert to the optimal levels from back when the campaigns were highly performant, without us having to invest time and money to 're-learn' the market. The data doesn't lie, and consumer search behavior for these events doesn't change."

Ziemann echoes this, maintaining specific segmentation exactly for seasonal events. He sets extreme targets mapped precisely to the historical Conversion Rate and ACoS of those specific seasonal peaks, allowing the software to lie dormant and then strike with precision the moment the calendar demands it.

The Attribution Gap: Diapers vs. Robot Vacuums

Perhaps the most sophisticated nuance discussed in the Real Talk session is the disparity in purchasing latency, best illustrated by the "Diapers vs. Robot Vacuum" analogy.

If a consumer is searching for diapers, the purchase intention is immediate. They click, they buy within 24 hours. A robot vacuum, however, is a considered purchase. A user might click an ad on Tuesday, watch a review on Thursday, wait for paycheck on Friday, and finally make the purchase on Sunday.

Amazon's API assigns sales attribution on a sliding scale (1, 7, 14, and 30 days). Most standard tools evaluate ACoS based on the immediate or 7-day window. This means the robot vacuum campaign looks like a massive failure on day 3, prompting traditional software to drastically lower bids right as the customer is making their final decision.

AMALYZE treats the attribution window as a third-dimensional control lever. If an advertiser knows their product has a multi-day consideration phase, they can set the software to expect delayed conversions.

For instance, an ad starts with 150 clicks and 5 sales (a 3.3% conversion rate). Based on this, the optimal bid is calculated at €0.17. But, because the system knows to wait for the 14-day attribution window, it monitors the delayed data. Six days later, two more sales attribute to that initial click cohort, raising the true conversion rate to 4.6%. The system now knows it could have bid €0.24 profitably. By adjusting the attribution parameters explicitly per target, AMALYZE prevents advertisers from prematurely killing the bids on high-ticket, long-consideration products.

Tactical Deep Dive: Sponsored Brands and Modifiers

As the marketplace evolves, the real estate on Amazon search engine results pages (SERPs) is increasingly dominated by Sponsored Brands (SB) and Sponsored Display (SD). Ziemann has adapted his strategy accordingly. Because the Cost Per Order (CPO) for Sponsored Brands is typically higher due to the premium placement (Top of Search videos, custom images), precision is paramount.

Ziemann's playbook for SB is surgical: "I set up exact and phrase match campaigns to harvest search terms, cross-reference them with the Search Query Performance report, and then segment the top three converting keywords into Single Keyword Campaigns."

Coupling single-keyword SB campaigns with custom, target-adapted lifestyle imagery yields a highly defensible, high-converting funnel. Furthermore, with AMALYZE now pulling Sponsored Brands data through the Marketing Stream, these premium placements can be subjected to the same ruthless hourly optimization as standard Products campaigns.

But this requires a deep understanding of Amazon's placement modifiers. Modifiers for Top of Search or Product Pages can scale bids by up to 900%. If an advertiser bids €1.00 but has a 400% Top of Search modifier, the actual bid is €5.00. Because modifiers are applied at the campaign level, applying them loosely across ad groups containing dozens of keywords is financial suicide. Ziemann and Chris stress that these multipliers must be mathematically reverse-engineered into your core ACoS goals, requiring highly restrictive, isolated campaign structures to execute safely.

Disproving the 'No ACoS' Launch Myth

In the Amazon seller community, it has become trendy to promote a "No ACoS" phase during a product launch—the idea being that you should spend wildly, regardless of profitability, simply to farm data and secure organic ranking.

Ziemann and Chris fundamentally disagree with this premise, viewing it as a misunderstanding of basic business economics. If you run a launch campaign at a 500% ACoS to force velocity, resulting in €10,000 of overspend, that deficit doesn't magically disappear when the launch ends.

To achieve an equilibrium where the product is truly profitable, the required ACoS post-launch must plunge drastically—often to an unsustainable 10% or less—just to pay back the launch debt. By integrating organic rank tracking (which AMALYZE monitors concurrently across 180+ IP configurations to bypass subjective, cookie-based SERP bias) directly alongside PPC data, Ziemann prefers to launch with high, but mathematically grounded targets. A launch Segment might be dialed to a 100% or 150% ACoS, but it remains a controlled variable, protecting the long-term margin of the SKU.

The Economics of Automation

Ultimately, the barrier to entry for advanced tools often comes down to perceived cost. The AMALYZE model currently charges €300 per month for accounts under €10,000 in ad spend, and a flat 3% for spend above that threshold.

When questioned about the viability of paying €300 for an account spending only €5,000, Chris reframed the equation around the value of an operator's time. If manual management takes four hours a week, switching to an automated, stream-driven system values that time at just €18 an hour. More importantly, it removes the manager from the trenches of bid tweaking, allowing them to focus on high-leverage activities: creative development, Sponsored Display expansion, and portfolio strategy. It changes the mindset from administrative maintenance to active business development.

Whether managing a 50-SKU brand or AMALYZE's largest current client with 1.2 million SKUs, the system remains structurally agnostic. Scale is irrelevant; the math is the same.

The Future of Amazon Placements

As the session wound down, Ziemann offered a final thought on the trajectory of Amazon advertising. The ad space is growing denser. He predicts that in the next two to three years, the real estate taken up by Sponsored Display and Sponsored Brands will evolve, potentially opening up entirely new, off-marketplace or deeply integrated platform placements that we haven't yet seen.

For brands to survive in that future, relying on week-old data and gut-feel bid adjustments won't be enough. The marketplace is moving too fast. It's moving, quite literally, by the hour. And for those equipped with the right data architecture, the shift from defensive spending to mathematically assured, target-level domination is already here.