Amazon’s Sponsored Products Algorithm Shift Is Upending PPC Economics for Mid-Market Sellers

Sometime in late May 2026, Amazon’s Sponsored Products algorithm quietly began weighting search term relevance signals differently — deprioritizing broad-match keyword bids in favor of tighter phrase and exact-match clusters that correlate more closely with recent purchase intent data. The shift wasn’t announced in any Seller Central bulletin. Most sellers noticed it first in their ACoS dashboards.

By June, the pattern was unmistakable. Across dozens of seller accounts, broad-match campaigns that had been delivering ACoS figures in the 18–24% range for months were suddenly running at 31–38%. Impression share on top-of-search placements was collapsing for SKUs that hadn’t changed a single listing element. Meanwhile, competitors running tightly structured exact-match campaigns were gaining Buy Box share and organic rank simultaneously.

“We saw it hit our home goods catalog first,” said Kiri Masters, founder of Bobsled Marketing and a longtime Amazon advertising analyst. “Broad-match terms that had been profitable for 18 months started burning cash almost overnight. The clients who had leaned heavily on auto campaigns and broad match to scale were the ones most exposed.”

“The algorithm is essentially telling sellers: stop spray-and-pray. If your keyword architecture wasn’t built on purchase-intent specificity, you’re paying a tax now.” — Kiri Masters, Bobsled Marketing

What exactly changed in Amazon’s Sponsored Products ranking logic?

According to multiple agency operators and tool vendors who track Amazon’s ad auction behavior, the shift appears to involve how Amazon’s relevance model weights recency of purchase signal against keyword-bid combinations. Previously, a broad-match bid on “stainless steel water bottle” could capture a wide range of adjacent queries profitably. Now, Amazon’s system is increasingly penalizing broad-match bids where the historical purchase-to-click ratio on long-tail variants falls below a threshold — effectively raising the effective CPCs on those terms by reducing Quality Score-equivalent signals.

Mike Frekey, VP of Marketplace Strategy at Envision Horizons, described it as Amazon moving closer to a Google-style Quality Score model, but without disclosing the mechanic. “We’ve been reverse-engineering it through bid simulator data and impression-share reports across about 200 active accounts,” Frekey said. “The accounts that maintained strong click-to-purchase rates on exact and phrase match — those accounts are actually seeing CPCs drop slightly. The broad-match-heavy accounts are getting squeezed on both sides: higher CPCs and lower impression share.”

• Broad-match ACoS up 8–14 percentage points on average across surveyed accounts, per Envision Horizons internal data
• Exact-match campaigns showing 6–9% CPC decreases in categories including kitchen, pet supplies, and sports
• Auto campaigns underperforming primarily on long-tail discovery terms — a historic strength
• Top-of-search placement costs rising faster than product page placement, suggesting demand-side pressure from better-structured advertisers

Which seller profiles are getting hit hardest?

The sellers taking the biggest margin hits tend to share a common profile: mid-market operators doing $2M–$15M annually on Amazon, running catalogs of 50–300 ASINs, who scaled their ad infrastructure during the 2022–2024 period when broad-match efficiency was at its peak. Many of them relied on tools like Perpetua or Teikametrics’ automated bidding to manage campaign structures that were wide but not particularly deep.

“We had a client in the baby gear category — about $6M annual Amazon revenue — who was running 14 broad-match campaigns covering his entire catalog,” said Elizabeth Greene, founder of Junglr, an Amazon PPC agency. “His ACoS went from 22% to 34% between May 15 and June 10. We rebuilt everything into tightly themed exact-match ad groups, harvested his best-converting search terms from two years of search term reports, and within five weeks he was back to 24% ACoS. But that transition cost him probably $40,000 in wasted ad spend during the diagnostic period.”

“The sellers who have precise search term data going back 18+ months are at a massive structural advantage right now. That historical purchase-intent data is the raw material for rebuilding correctly.” — Elizabeth Greene, Junglr

Aggregators are also feeling the pain. Multiple sources at firms managing post-acquisition catalog portfolios — including operators familiar with Thrasio’s restructured ad teams and Perch’s current agency relationships — report that portfolio-level ACoS figures deteriorated noticeably in Q2 2026. One aggregator source, who asked not to be named, said their blended Amazon ad spend efficiency dropped from a 4.1x return on ad spend (ROAS) in April to 3.4x in June, representing millions in annualized margin erosion.

How are top agencies restructuring campaigns in response?

The tactical response among leading Amazon PPC practitioners has converged around a few core moves, executed with varying degrees of speed and sophistication depending on account size and available data.

• Search term harvesting at scale: Agencies are pulling 24-month search term reports and filtering by orders-per-click, not just clicks. Only terms with documented purchase conversion are being promoted to exact-match campaigns.
• Tight single-keyword ad groups (SKAGs): A technique borrowed from Google Ads but adapted for Amazon’s structure, SKAGs allow granular bid control and prevent keyword cannibalization across a catalog.
• Negative keyword architecture: Several agencies report adding 200–400 negative exact-match terms per campaign as part of initial rebuilds — aggressive pruning to prevent broad and phrase campaigns from capturing low-intent traffic.
• Dayparting and placement bid adjustments: With top-of-search CPCs rising, some operators are shifting budget to product-page placements where competition hasn’t spiked as sharply, accepting lower conversion rates in exchange for lower CPCs.
• AI-assisted keyword clustering: Tools like Helium 10’s Adtomic, Pacvue, and Scale Insights are being used to cluster semantically related search terms into tightly themed campaigns, reducing the manual labor of rebuilding large catalogs.

What does this mean for Amazon’s ad revenue and seller economics longer-term?

There’s an uncomfortable tension at the center of this story. Amazon’s advertising business — which generated $56.2 billion in revenue in 2025 and is tracking above that pace in 2026 — is structurally dependent on sellers continuing to spend. An algorithmic shift that punishes inefficient spend could, in theory, reduce total ad dollars in the auction.

But most analysts believe the net effect runs the other way. By rewarding precise, purchase-intent-aligned campaigns with lower CPCs, Amazon is incentivizing sophisticated sellers to reinvest efficiency gains into higher bid ceilings. Meanwhile, less sophisticated sellers either exit the auction (reducing clutter) or upgrade their infrastructure, ultimately spending more. “Amazon is essentially forcing professionalization of the seller ad stack,” said Frekey. “The amateur hour broad-match campaigns are being taxed out of the most competitive placements. That’s good for Amazon’s auction quality and, honestly, good for the serious sellers who invest in doing it right.”

“This is the Amazon ads maturation moment. Google went through it between 2012 and 2016. Amazon is compressing that same evolution into 18 months.” — Mike Frekey, Envision Horizons

Which tools and platforms are best positioned to help sellers adapt?

The rebuild cycle is creating a clear winner tier among Amazon advertising platforms. Pacvue and Scale Insights are drawing comparisons for their granular bid-management capabilities at the keyword level. Helium 10’s Adtomic has seen a reported surge in agency inquiries specifically around its search term isolation and SKAG-building workflows. Perpetua, which has historically favored automation and goal-based bidding, is facing more pointed questions from agencies about whether its abstraction layer is appropriate for the current environment.

“We moved three of our largest accounts off Perpetua and onto Scale Insights in June,” said Greene. “Not because Perpetua is bad — it’s a great tool for growth-phase accounts — but because the current environment demands more surgical bid control than a goal-based system naturally provides.”

For sellers who can’t afford agency management or enterprise tooling, the playbook is more manual but achievable: export 12–24 months of search term data, sort by orders and order rate, manually promote the top 50–100 converting search terms per ASIN into exact-match campaigns, set aggressive negatives in existing broad campaigns, and audit weekly. It’s time-intensive — plan for 10–15 hours of setup per 20-ASIN catalog — but multiple practitioners confirmed this approach is recovering ACoS within 4–6 weeks.

What should sellers do right now to protect their margins?

The consensus from operators and agency leads who have already navigated the transition is clear: the sellers who act in July will spend significantly less on the diagnosis than those who wait until Q4, when holiday auction pressure compounds the existing dysfunction.

• Audit your broad-match and auto campaign ACoS against your Q1 2026 baseline — if variance exceeds 7 percentage points, treat it as a structural problem, not noise
• Pull your full search term report going back to January 2025 and identify your top 100 converting queries per ASIN; these are your exact-match campaign seeds
• Set product-targeting negatives in auto campaigns for any ASIN where auto ACoS exceeds your target by more than 10 points
• Reduce broad-match budgets by 30–40% immediately and reallocate to exact-match campaigns built from harvested data
• Book a campaign audit with your tool vendor — Pacvue, Scale Insights, and Helium 10 Adtomic all offer diagnostic reports that can baseline your account health in under 48 hours

The sellers who built precision into their PPC architecture before May are watching their competitors’ chaos from a comfortable margin position. For everyone else, the restructuring work is urgent — and the cost of delay is measurable in real dollars, not hypothetical risk.