The Search Term Funnel: How to Turn Auto Campaigns Into a Keyword Discovery Engine

The Problem With "Set and Forget" Auto Campaigns

Most Amazon sellers launch an auto campaign, glance at ACOS once a week, and call it a day. The auto campaign quietly spends, Amazon decides which queries trigger your ads, and you never build a system to capture what's working and cut what isn't.

The result is a campaign structure that leaks budget on irrelevant terms while failing to capitalize on the high-converting queries buried in your search term reports.

The search term funnel is not a set of tricks. It is a repeatable framework for turning raw search data into a compounding advantage — one that gets more efficient the longer you run it.

Match Types: What Each One Actually Does

Before building the funnel, you need to understand what each match type is doing under the hood. This is where most guides oversimplify.

The critical point: these match types are not alternatives. They are stages in a pipeline. Each one feeds the next.

The Auto-to-Exact Funnel

The funnel has three stages, and the discipline is in moving keywords through them — not letting them stagnate in discovery forever.

Stage 1: Discovery (Auto and Broad)

Auto campaigns and broad match campaigns are your prospecting layer. Their job is to surface queries you didn't know about. You deliberately accept higher ACOS here because the goal is data, not immediate profitability.

• Auto campaigns let Amazon match your listing against queries based on its own relevance models
• Broad match campaigns target seed keywords and let Amazon expand them to related terms
• Budget is controlled but not restrictive — starving discovery starves your entire funnel

Stage 2: Expansion (Broad and Phrase)

Once a query shows potential in Stage 1 (clicks, some conversions, reasonable CTR), you move it to a phrase match campaign. This narrows the query set while still capturing useful variations. Broad match continues to run for the original seed terms, discovering new queries you haven't found yet.

Stage 3: Control (Exact)

Queries that prove themselves — consistent conversions, ACOS at or below your target — graduate to exact match campaigns. Here you have maximum bid control and the cleanest performance data. This is where profitability is built.

How to Read the Search Term Report

The Search Term Report (STR) is the raw material for your entire funnel. Every decision — harvest, negate, or watch — flows from this data. Here is how to categorize what you find.

What to Harvest (Move to Exact)

A search term is ready for exact match when it meets these thresholds:

• 30+ clicks — enough data to be statistically meaningful
• 12%+ conversion rate — the query is clearly purchase-intent aligned with your product
• ACOS below your target — the term is profitable or close to it at current bids

When all three criteria are met, create an exact match keyword in your control campaign. This is harvesting.

What to Negative (Block)

The thresholds are not arbitrary. At 15 clicks with zero conversions, you have enough data to know the term does not match your product's purchase intent. The $10–15 spend threshold catches low-CTR terms that accumulate cost slowly but add up over time.

What to Watch

Not every term falls neatly into harvest or negate. These require patience:

• High CTR, low CVR — customers are clicking but not buying. This could be a listing quality issue, a price problem, or a relevance mismatch. Investigate before negating.
• Low impressions, good CVR — a niche term that converts well but doesn't get much traffic. Consider increasing bids in phrase or exact to capture more volume.
• Under 15 clicks — not enough data to decide. Let it run longer before taking action.

N-Gram Frequency Analysis

Individual search term review misses patterns. N-gram analysis reveals the underlying keyword components that drive conversions across multiple queries.

How It Works

Break every search term in your STR into fragments: 1-word (unigrams), 2-word (bigrams), and 3-word (trigrams). Then count how frequently each fragment appears across converting terms versus non-converting terms.

This analysis often surfaces patterns invisible at the individual query level:

• Modifier patterns — words like "large," "organic," or "professional" that consistently appear in converting terms
• Negative patterns — fragments that appear almost exclusively in non-converting terms, signaling queries to block in bulk
• Category language — how customers naturally describe your product, which may differ from how you describe it in your listing

Negative Keyword Strategy

Negation is half the funnel. Discovery without negation is just spending. Here is how to think about it systematically.

Negative Exact vs. Negative Phrase

Thresholds for Negation

• $10–15+ spend with zero conversions — negate immediately
• 15+ clicks with zero conversions — negate immediately
• If n-gram analysis reveals a fragment that appears consistently in non-converting terms, consider a negative phrase to block the entire cluster

Campaign-Level vs. Ad Group-Level

Campaign-level negatives apply across all ad groups in that campaign. Ad group-level negatives apply only to the specific ad group. Use ad group-level negatives when a term is irrelevant for one product but may be relevant for another product in the same campaign. Use campaign-level negatives when the term is categorically irrelevant.

The Step Most Sellers Miss: Negative in the Source

This is the single most common structural error in Amazon PPC. The harvest without negation creates an internal bidding war where you pay more for the same click and split your conversion data across two campaigns, making neither one's performance data reliable.

The workflow is always two actions, never one:

1. Add the keyword to the exact match campaign
2. Negate the keyword in the source campaign (auto, broad, or phrase)

Bid Logic: Exact > Phrase > Broad > Auto

Your bid structure should reflect your confidence in each match type. Exact match keywords have proven conversion data — they deserve the highest bids. As you move down the funnel toward discovery, bids decrease because uncertainty increases.

Running the same bid across all match types is a common mistake. It overpays for unproven discovery traffic and underpays for high-converting exact matches — the worst of both worlds.

One Product Per Ad Group (OPAGA)

The cleanest campaign architecture assigns exactly one product per ad group. This is not a minor organizational preference — it directly affects the quality of your data and your ability to optimize.

OPAGA requires more campaigns and more management overhead. But the data clarity it provides makes every other optimization decision in this guide more reliable.

Placement Analysis: Top of Search, Product Pages, Rest of Search

Amazon reports performance by placement: Top of Search (first page, above the fold), Product Pages (ads on competitor listings), and Rest of Search (everything else). Each placement has fundamentally different economics.

• Top of Search — highest visibility, highest CTR, typically highest conversion rate. Also the most expensive. Worth paying a premium for proven exact match keywords.
• Product Pages — lower intent (the customer was looking at a specific product, not searching). Can convert well for complementary or competitive products, but often at lower rates.
• Rest of Search — mixed bag. Includes lower positions on search results pages. Often the cheapest clicks but with variable quality.

Use Amazon's placement bid adjustments to allocate budget where your data shows the best return. Do not assume Top of Search is always best — some products convert better on product pages, especially if your product is a clear upgrade to competitors.

Brand vs. Non-Brand: Two Different Games

Brand terms (queries containing your brand name) and non-brand terms (generic queries) behave so differently that analyzing them together produces misleading numbers.

Brand traffic is defense: customers already know you and are searching for your product by name. You bid on these to prevent competitors from capturing your traffic. Non-brand traffic is offense: this is where growth happens, where you acquire new customers who don't know your brand yet.

Segment your reporting. Set different ACOS targets. Manage them as separate strategies.

Common Mistakes That Silently Drain Budget

These errors don't cause dramatic failures. They create slow, invisible waste that compounds over weeks and months.

1. Cutting terms too early — Negating a keyword with only 5–8 clicks is premature. You don't have enough data to know if it converts. Wait for 15+ clicks or $10+ in spend before making a decision.
2. Over-harvesting without negation — Moving keywords to exact without adding negatives in the source campaign. Both campaigns compete for the same query, inflating costs.
3. Not checking listing quality — A keyword with high CTR and low CVR might not be a bad keyword. It might be a bad listing. Check your images, pricing, and copy before negating terms that show purchase intent but fail to convert.
4. Ignoring n-gram patterns — Reviewing search terms one at a time misses the forest for the trees. Aggregate analysis reveals which keyword components actually drive conversions.
5. Mixing branded and non-branded analysis — Averaging brand and non-brand performance hides the true cost of customer acquisition and makes non-brand campaigns look better than they are.
6. Running the same bid across all match types — This overpays for discovery and underpays for proven converters. Bid should reflect confidence: exact gets the highest bid, auto the lowest.

Putting It All Together: The Weekly Workflow

The funnel only works if you work it consistently. Here is the cadence that keeps it running.

Weekly Tasks

1. Pull the Search Term Report — Filter for terms with 15+ clicks or $10+ in spend. These are the terms with enough data to act on.
2. Harvest and negate — Move qualifying terms to exact match. Add negative exact in the source campaign. Negate non-converting terms that hit your thresholds.

Biweekly Tasks

1. N-gram analysis — Break search terms into fragments. Compare frequency in converting vs. non-converting terms. Identify new keyword components to target and negative patterns to block.

Monthly Tasks

1. Placement review — Analyze performance by Top of Search, Product Pages, and Rest of Search. Adjust placement bid modifiers based on where your ads convert best.
2. Brand vs. non-brand segmentation — Separate brand and non-brand performance. Verify that non-brand ACOS is trending in the right direction independently of brand traffic.