A weekly PPC harvesting process to isolate converting terms, reduce waste, and scale profitable spend.

Amazon PPC Keyword Harvesting Routine (Weekly)

Keyword harvesting is the process of moving search terms that are already converting in your auto or broad match campaigns into dedicated exact match campaigns where you control the bid precisely. It sounds simple. Most sellers understand the concept. Almost none of them do it consistently enough to matter.

The sellers who run this routine every week — without exception — end up with PPC accounts that get sharper over time. Their auto campaigns stop overpaying for terms they could own for less. Their exact match campaigns accumulate a growing list of proven keywords with tightly managed bids. Their ACoS drops while order volume climbs. That compounding improvement only happens through consistent weekly execution.

This guide covers the complete process: what to download, how to filter it, where each term goes, how to structure your tracking sheet, and what a real example looks like from start to finish.

What Keyword Harvesting Actually Is

When Amazon runs your auto campaign, it matches your ads to search terms based on your listing content and its own algorithm. You don't choose these terms — Amazon does. Some convert. Many don't. All of them cost money.

Harvesting is the weekly discipline of reviewing those search terms and making two decisions:

1. This term converts — move it into an exact match campaign so I control the bid and stop overpaying for it in auto.
2. This term doesn't convert — add it as a negative exact match to the source campaign so it stops spending.

Every week you skip this process, your auto campaign continues paying whatever Amazon decides for terms you already have data on. Proven converters keep running at uncontrolled auto bids. Known non-converters keep burning budget. The wasted spend is not random — it's structured and predictable, which means it's entirely avoidable.

The harvesting loop is how PPC accounts mature. Without it, you're funding Amazon's algorithm indefinitely with no compounding return on your data.

Why This Is the Most Important Weekly PPC Habit

Most sellers treat auto campaigns as passive discovery engines. They set a budget, let it run, and check ACoS at month's end. This approach has a structural flaw: auto campaigns are optimized for coverage and impressions, not your margin. Amazon will keep spending on broadly relevant terms regardless of whether those specific terms have ever produced an order for your product.

The math compounds against you fast. If your auto campaign spends $30/day and 40% of that is on terms you already know don't convert, that's $12/day in waste — $84/week, $360/month. Meanwhile, terms that do convert are running at whatever bid Amazon assigned, which is often 20-40% higher than what you'd pay in a controlled exact match campaign.

Starting with the right seed keywords changes the timeline dramatically. Tools like Launch Fast identify high-intent terms before you launch, which means your auto campaigns surface genuinely relevant search traffic from the beginning. When your auto campaign starts from a strong seed list, the harvesting loop finds golden terms in week 2 instead of month 2.

Weekly cadence is the right frequency. Daily harvesting produces too little data per term to make reliable decisions. Monthly harvesting lets wasted spend compound for weeks. Seven to fourteen days gives you enough clicks per term to read the signal clearly without letting money drain unnecessarily long.

The 6-Step Weekly Harvesting Process

Step 1: Download the Search Term Report

Go to Amazon Advertising → Reports → Sponsored Products → Search Term Report. Set the date range to the last 7 days. If a term has been running longer and you want more data, use 14 days — but never go beyond 14 or you lose the ability to act on fresh signals before conditions shift.

Download the report as a CSV. Open it in Excel or Google Sheets.

Columns you need: Search Term, Campaign Name, Ad Group Name, Impressions, Clicks, Spend, Orders, Sales, ACoS.

Step 2: Filter for Clicks Greater Than or Equal to 8

Delete or hide every row with fewer than 8 clicks. Terms with less than 8 clicks don't have enough data to make a reliable decision. Acting on them leads to over-negating terms that just needed more time, or harvesting terms that got lucky on a single click.

Eight clicks is the minimum threshold for a readable signal. For high-ticket products (over $80 ASP), you can lower this to 5 clicks because the revenue per order is high enough that even 5 clicks with 1 order is meaningful.

Step 3: Split Into Two Lists

After filtering, you have two groups:

• Converting terms: Clicks ≥ 8 AND Orders ≥ 1. Candidates for harvesting into exact match.
• Non-converting terms: Clicks ≥ 8 AND Orders = 0 AND Spend ≥ $3. Candidates for negative exact match.

The $3 spend floor on non-converting terms ensures you're not negating terms that barely ran. A term with 8 clicks and $1.20 spend might just need more impressions. A term with 8 clicks and $4.50 spend and zero orders has had a real chance and did not convert.

Step 4: Handle Converting Terms

For each converting term:

First — add it as a negative exact match to the source campaign (the auto or broad campaign where it was discovered). This prevents the source campaign from continuing to bid on a term you're now controlling in exact match. Without this step, you'll have both campaigns bidding against each other and inflating your own CPC. This is the step most sellers forget.

Second — add it as an exact match keyword to your exact match campaign. Set the opening bid at: reported CPC × 0.9. If the auto campaign was paying $1.20 CPC for this term, start your exact match bid at $1.08. You now control it at a slightly lower entry point, with room to raise if performance warrants it.

If this term already exists in your exact match campaign, do not add it again. Check the current bid against this week's reported CPC and adjust accordingly.

Step 5: Handle Non-Converting Terms

For each non-converting term (clicks ≥ 8, spend ≥ $3, orders = 0):

Add it as a negative exact match to the source campaign only. Do not add it at the account level unless you're certain this term is irrelevant across every product you sell.

Campaign-level negatives are safer and more reversible. You can remove a campaign-level negative if you later see evidence a term converts for a different product in a different campaign. Account-level negatives are permanent and broad — use them rarely.

Step 6: Check for Duplicates Before Finalizing

Before completing your harvest additions, verify: does this search term already exist as a keyword in any other campaign? If yes, you are not harvesting — you are duplicating. Duplication creates internal competition and pushes up your CPCs.

The correct action when a term already exists in exact match: skip re-adding it. Look at the bid on the existing keyword, compare it to this week's CPC from the STR, and adjust if there's a material gap. Then add the term as a negative to the source campaign to stop double-bidding. That completes the loop without creating a duplicate.

Setting Up the Harvesting Spreadsheet

A tracking spreadsheet prevents the harvesting process from becoming chaotic as your keyword list grows. You want six months of data visible in one place so you can spot patterns over time.

Use these five columns:

Add a date column header for each week's pull. Keep all weeks in the same sheet, stacked chronologically. Color-code by action: green for harvested, red for negated, yellow for watching.

Over six months this sheet becomes a reference that shows you when you harvested each keyword, what it was doing when you took action, and whether that decision held up. Terms you negated that later reappear as winning terms in competitor research get flagged here — a useful reminder that you may have negated too early on thin data.

Keep the spreadsheet under 500 rows by archiving terms older than six months to a separate tab. Anything older than that rarely needs active reference.

When to Harvest More vs. Less Aggressively

New launch (weeks 1-8): Harvest every 7 days without exception. Your auto campaign is discovering the search landscape and generating data fast. Leaving waste in place for more than a week compounds losses during the phase when you're most budget-sensitive. Every misallocated dollar during the launch window slows velocity and hurts ranking momentum.

Growth phase (months 2-6): Weekly harvesting stays the standard. As your exact match campaign grows, each week's harvest adds fewer new terms — you've already captured the obvious ones — but the negating work stays consistent. New seasonal terms, competitor launches, and algorithm shifts keep introducing new non-converting traffic.

Mature account (6+ months, stable ACoS): Bi-weekly is acceptable once your exact match campaign contains 40+ proven terms and your auto campaign's spend is largely well-targeted. Monthly review is the absolute outer limit — never go longer. Even stable accounts drift. Seasonal terms appear, competitor activity shifts search behavior, and Amazon occasionally starts matching your ads to new term clusters without warning.

The cost of skipping one week is modest. The cost of skipping four consecutive weeks is a month of compounding waste that takes another month to reverse.

Ad Group Structure for Harvested Terms

When you harvest a new keyword into your exact match campaign, you need to decide where it lives structurally.

Two options:

Themed grouping: Multiple related keywords in one ad group. For example, an ad group called "Silicone Spatula — Kitchen" contains "silicone spatula set," "silicone cooking spatula," and "heat resistant silicone spatula." All three terms share the same listing and ad. Bids are set at the ad group level, meaning all three keywords share one bid.

The limitation: you cannot bid differently on "silicone spatula set" (high volume, high competition) versus "heat resistant silicone spatula" (lower volume, lower CPC). Themed grouping averages your bids across terms with different competitive dynamics.

SKAG (Single Keyword Ad Group): Each harvested keyword gets its own ad group with its own independent bid. "Silicone spatula set" is one ad group. "Heat resistant silicone spatula" is another. You control each bid separately and read performance data cleanly per term.

The limitation: SKAGs create significant account complexity fast. Managing 80 individual ad groups is a substantial time commitment.

When to use each:

The goal is matching your management overhead to your data volume. SKAGs only pay off when you have enough clicks per term to justify individual bid optimization.

Mistakes That Break the Harvesting Loop

Adding negatives to the wrong campaign level. Account-level negatives block a term across every campaign you run. If you accidentally block a converting term at the account level because it was irrelevant in one campaign, you can kill performance across multiple campaigns simultaneously. Always default to campaign-level negatives.

Duplicating terms without checking for existing entries. Adding a keyword that already exists in your exact match campaign creates internal competition. Both entries bid against each other and you pay yourself up in the auction. Always verify the exact match campaign before adding a harvested term.

Harvesting from brand campaigns. If you run a branded defense campaign, search terms in that campaign are people searching your brand name specifically. Harvesting branded terms into your main exact match campaign muddles data and wastes budget on traffic you were already capturing cheaply. Keep brand campaigns separate and do not harvest from them.

Forgetting to negative the source campaign. This is the most common mistake. You harvest a term into exact match but leave it running uncontrolled in auto. Both campaigns now bid on the same term. The auto campaign often wins the auction because its bids are not capped, and you lose the cost savings you set out to capture. Every harvest session must end with: did I add the harvested term as a negative exact to the source campaign?

Harvesting on too little data. Adding a term to exact match after 3 clicks and 1 order is not harvesting — it's gambling. One conversion on 3 clicks is a 33% CVR, which almost certainly will not hold at scale. Use the 8-click minimum filter. The patience to wait for real data is what separates systematic harvesting from impulsive keyword management.

Real Example Walkthrough

Your auto campaign has been running for 3 weeks on a silicone spatula product. You pull this week's STR and sort by spend descending.

One row: "silicone spatula set" — 15 clicks, 2 orders, $18.50 spend, $42.00 in revenue. ACoS: 44%. Reported CPC: $1.23.

Here is the exact sequence:

1. Check the threshold. 15 clicks (≥ 8), 2 orders (≥ 1). This is a harvesting candidate.

2. Calculate the opening exact match bid. $1.23 × 0.9 = $1.11.

3. Check if "silicone spatula set" already exists in your exact match campaign. It does not — this is a new harvest.

4. Add "silicone spatula set" as an exact match keyword to your exact match campaign, ad group "Silicone Spatula — Core," at a bid of $1.11.

5. Add "silicone spatula set" as a negative exact match to the auto campaign. The auto campaign will no longer bid on this term. Your exact match campaign now owns it.

6. Log it. Search Term = "silicone spatula set," Source Campaign = "Auto — Silicone Spatula," Weekly Clicks = 15, Weekly Orders = 2, Action = "Harvested to Exact @ $1.11."

Three weeks later, you check the exact match campaign STR. "Silicone spatula set" has 28 clicks, 5 orders, $34.40 spend, $105 revenue. ACoS: 32.8%. That's under a 35% target. The bid holds. At week eight, ACoS is 28%. You have room to raise the bid 10% to $1.22 and capture more top-of-search placements.

That single keyword, properly harvested and managed, went from 44% ACoS in an uncontrolled auto campaign to 28% ACoS in a controlled exact match campaign — at lower cost per click.

Monthly Rollup Review

Once a month, open your harvesting spreadsheet and look at all terms harvested in the prior 30 days that have at least 30 days of data in your exact match campaign.

Ask two questions:

Which harvested terms have sustained ACoS under target for 30+ days? These are proven performers. Increase bids 10-15% to capture more impression share and take volume from competitors bidding on the same terms. If a keyword is converting at 22% ACoS against a 35% target, you have margin to spend more aggressively.

Which harvested terms have deteriorated? A term at 30% ACoS in week one that has drifted to 58% over 30 days tells you something changed — competitor bids increased, seasonality shifted demand, or your listing conversion rate dropped. Reduce the bid 15-20% and monitor for two more weeks before making a final decision to pause or negate.

The monthly rollup keeps your exact match campaign clean. Without it, you accumulate stale keywords that drain budget on terms that stopped performing months ago. It's the maintenance layer that keeps the weekly loop from becoming a graveyard of old decisions.

Putting It All Together

The weekly harvesting routine is not glamorous. It takes 30-45 minutes per week. But it is the single process that separates PPC accounts that compound gains from accounts that plateau.

Run the loop every week. Download the STR, filter to 8+ clicks, split into converting and non-converting groups, action each group, verify negatives are in the source campaign, log everything. That's the complete process.

Do it consistently for twelve weeks and your exact match campaign will contain 30-60 proven terms at controlled bids. Your auto campaign will be spending on genuinely new territory instead of recycling terms you already know the answer on. Your ACoS will be measurably lower than where you started — not because you got lucky, but because the loop compounded week over week.

Related: Amazon PPC for Beginners: First Campaign Setup | How to Lower ACoS Without Killing Sales | PPC Tools