If you’ve noticed your go-to “bid high on big keywords, fight for top 3 organic positions” Pay-Per-Click (PPC) strategy isn’t delivering the same results it did 2–3 years ago, you’re not alone. I’ve spent the past 7 months testing ad performance across 12 of my brand’s Amazon Standard Identification Numbers (ASINs) under the COSMO algorithm, and the core shift is clear: we’re no longer buying keywords — we’re buying user intent. This post breaks down exactly how to adjust your strategy to fit the new system, no fluff, just actionable steps I’ve used to cut my overall Advertising Cost of Sale (ACOS) by 21% and stabilize rank across US, CA, and DE sites.

All performance benchmarks referenced are consistent across North American, European, and UK Amazon sites; adjust bid levels and budget thresholds to match your specific site’s average Cost-Per-Click (CPC) norms.

  1. Relevance splits into two layers: static (your Listing setup, the entry ticket to visibility) and dynamic (performance metrics, what earns you top ranks).

  2. The intent-based tagging strategy I’ve tested cuts ACOS by an average of 21% by targeting 30–50 unique purchase triggers instead of 5–10 high-volume keywords.

  3. Use bounded Broad Match (core high-intent terms + pre-emptive negative keywords) to scale intent signals without wasting ad spend.

A9 vs. COSMO: The Core Algorithm Shift Killing Your Old PPC Strategy

To understand why your old tactics no longer work, think of the two algorithms as completely different store employees:

  • The old A9 algorithm acted like a directory administrator. Its only job was to confirm you qualified to be in the store: your Listing matched keyword terms, your category node was correct, and you met basic eligibility requirements. It cared only about what you said your product was, not how well it served customer needs.

  • The new COSMO algorithm acts like a personal in-store shopper. For the exact same search term, it will show different products to different users based on their past behavior, purchase context, and specific needs for that search. It cares far more about whether your product is the right fit for that specific user’s intent than whether you have the exact keyword in your title.

This is why so many sellers see unexpected fluctuations in performance even when they don’t change their Listing or ad setup: the algorithm is no longer just matching keywords — it’s matching intent.

Relevance Has 2 Distinct Layers — Ignore One, and You’ll Never Rank

Most sellers only track one type of relevance (either page copy or ad performance), but COSMO evaluates two separate layers. Clarifying these will simplify your entire strategy:

Static Relevance: Your Entry Ticket to Visibility

Static relevance is the basic eligibility check the algorithm runs before it will ever show your product to users. It includes:

  • Alignment across your title, bullet points, A+ Content, and backend Search Terms

  • Correct category node and attribute field setup

  • Consistent variant structure and parameter data

  • Clear, accurate product imagery that matches your written copy

Static relevance only determines if you are eligible to be shown — it does not guarantee top positions. You can have a perfectly optimized Listing and still never rank if your dynamic relevance is poor.

Dynamic Relevance: What Gets You Pushed to the Front of the Line

Dynamic relevance is the signal COSMO prioritizes above all else: how well your product performs for a specific user intent. It is measured by:

  • Click-Through Rate (CTR) for that intent

  • Conversion Rate (CVR) for that intent

  • Add-to-cart rate, repeat purchase rate, and other positive engagement signals

  • Negative signals including return rates and 1-star reviews tied to that intent

Dynamic relevance is earned through consistent performance for specific user intents, not just keyword matches. A product with solid but not perfect static relevance and strong dynamic performance will almost always outrank a perfectly optimized Listing with low intent-aligned performance.

In the COSMO Era, PPC Buys You Algorithm Trust, Not Just Ranks

Most sellers still view PPC as a tool to buy exposure, orders, or temporary rank boosts. In the COSMO system, that view will waste your ad budget.

PPC now functions as a way to buy the algorithm’s trust for specific intents. Every click, conversion, and positive engagement signal from your ads is real user behavior data that COSMO uses to judge whether your product is a good fit for that intent.

This is why increasing your budget no longer guarantees immediate performance spikes: you’re buying the opportunity to show your product to users, but you only earn long-term, stable traffic when you consistently perform well for that intent. The more positive signals you send for a specific intent, the more likely COSMO is to push your product to more users with that same intent, both in ad results and organic rankings.

The Intent-Based Tagging Strategy: How to Capture COSMO’s Intent Preference

Instead of focusing on 5–10 high-volume “big keywords”, this strategy breaks your product down into 30–50 unique purchase triggers: the specific, real reasons a user would choose to buy your product (not just the generic words they search). This lets you build positive intent signals across dozens of high-intent use cases, rather than gambling performance on a handful of competitive terms.

Purchase triggers fall into 5 actionable categories:

  • Use case scenarios: Home essentials, travel must-haves, gifting, emergency use, outdoor applications

  • Target audiences: Parents, camping enthusiasts, pet owners, senior users, small business owners

  • Core needs: Durability, ease of use, safety, portability, fast results

  • Timely events: Back-to-school, holiday gifting, severe weather preparation, peak travel seasons

  • Comparative value: Upgrades from older, less effective product solutions (ensure all claims are compliant with Amazon Ads policies)

Every small win for a specific intent builds more algorithm trust than one big win for a generic high-volume keyword. You are not just expanding your keyword list — you are expanding the number of reasons the algorithm has to show your product to users.

Step-by-Step SOP for the Intent-Based Tagging Strategy

This is the exact process I use for all my brand’s ASINs, adjusted to minimize wasted spend and maximize intent signal clarity:

  1. Build your intent tag library first

    Don’t overcomplicate this — pull all purchase triggers from three free, high-intent sources:

    • Customer reviews and Q&As on your Listing and top competitor Listings (look for exact phrases users use to describe why they bought the product)

    • Competitor bullet points and A+ Content (note the core value props they prioritize)

    • Your Search Term Report from past PPC campaigns

    For every tag you identify, map it to: 3 representative Keywords, 10 competing ASINs that target the same intent, and 1 core value prop you can highlight in your Listing to align with the intent.

  2. Tier your tags to minimize wasted spend

    Start with high-conversion intents first, then expand to lower-intent tags to avoid blowing your budget on unproven traffic:

    • Tier 1 (High Relevance): Core product functions and primary use cases — these have the highest baseline CVR, so prioritize these first to build initial algorithm trust.

    • Tier 2 (Medium Relevance): Scenario + need combined triggers (e.g., “waterproof dog bed for camping”) — use these to expand your reach once Tier 1 tags are performing consistently.

    • Tier 3 (Low Relevance): Broad audience or event-based triggers — test these with small, tightly controlled budgets and pause immediately if performance is below your threshold.

  3. Identify high-value root terms, then scale with bounded Broad Match (this is the core of the strategy)

    Most sellers either avoid Broad Match entirely because of unqualified traffic, or jump straight to Exact/Phrase Match after mining root terms. This method balances reach and precision:

    1. First, run automatic campaigns and open Broad Match campaigns for 2–3 weeks to gather full search term data. Don’t worry about high ACOS in this stage — your goal is to uncover which root terms actually drive conversions, not hit performance targets.

    2. Pull your Search Term Report and split all root terms into two groups: high-value root terms (drive CVR ≥ your Listing’s baseline CVR) and invalid root terms (drive clicks but 0 conversions, or CVR 50% below your baseline).

    3. Instead of moving all high-value terms to Exact/Phrase Match, create new Broad Match campaigns using [high-value root term + core product term] as your target Keywords. Add all invalid root terms as Phrase Match negatives before launching the campaign to block unqualified traffic from the start.

    This method lets you leverage Broad Match’s ability to capture all variations of a specific user intent, while eliminating the unqualified traffic that usually makes Broad Match inefficient. Monitor search terms daily during the first 7 days, and add new negatives immediately if unrelated terms appear.

  4. Optimize based on performance to avoid sending bad intent signals

    After 2–4 weeks of running the bounded Broad Match campaigns, segment all tags/keywords into three groups and take action:

    • High-performing (CVR ≥ baseline, ACOS ≤ target): Increase budget by 10–15% every 3–5 days to scale reach for that intent.

    • Potential (CTR ≥ baseline but CVR below target): Update your Listing’s imagery, bullet points, or value props to better align with that specific intent to improve conversion.

    • Low-performing (CTR < 80% of baseline, or ACOS > 2x your target): Pause the keyword and add it as a negative if it drives unqualified traffic. Never continue running low-performing, unaligned terms — these send bad intent signals to the COSMO algorithm that can hurt your organic and ad performance long-term.

Key Best Practices to Avoid Common Mistakes

  1. Target specific intent phrases, not generic terms

    Don’t waste budget on vague terms like “gift” or “travel” — target high-intent phrases that align with actual purchase behavior, such as “gift for hikers”, “travel size skin care for carry-on”, or “emergency weather kit for cars”. These terms have far higher conversion intent and send clearer signals to the algorithm.

  2. Restructure campaigns by intent, not just keyword volume

    Create separate campaign budgets for each core intent group to avoid overspending on low-priority intents, and make it easier to track performance per intent type.

  3. Align your Listing content with your target intents

    Add scenario-specific imagery, bullet points, and Q&As to your Listing that speak directly to each high-priority intent. This is especially important as Amazon’s Rufus AI shopping assistant prioritizes clear, scenario-aligned content for user queries.

  4. Don’t “stuff” scenario terms

    A common mistake I see is sellers adding dozens of irrelevant scenario terms to their Listing to try to capture more intent matches, which is just the modern version of keyword stuffing. You need to balance algorithm visibility and user experience:

    • Too many irrelevant scenario terms will make your Listing feel generic and lower CVR for your core audiences.

    • Too few intent-aligned terms will limit the algorithm’s ability to match you to relevant user intents.

    Focus on adding only intent terms that actually drive conversions, not every possible scenario you can think of.

Have you tested intent-based PPC strategies under the COSMO algorithm? Did you see better stability and lower ACOS than traditional big-keyword bidding? Drop your experience in the comments below — I’m curious to hear what’s worked for other sellers!