🎯 Keyword Harvesting: Moving Winners From Auto to Manual Campaigns

Keyword harvesting is the process of identifying high-performing search terms from your Amazon Auto campaigns and promoting them into Manual campaigns where you can control bids, match types, and budgets more precisely. It is one of the most impactful and repeatable PPC optimization habits an Amazon seller can build. In this article, you will learn how to identify winning keywords, when to move them, how to structure your Manual campaigns, and how to avoid the most common mistakes that erode profit.

An Amazon Sponsored Products campaign where Amazon automatically matches your ads to relevant search queries based on your listing content and category. Auto campaigns are ideal for discovery — finding search terms you may not have thought of yourself.

A Sponsored Products campaign where you choose specific keywords or product targets and set individual bids. Manual campaigns give you full control over where your ads appear and how much you pay per click.

A downloadable Amazon Advertising report that shows the actual search queries customers typed before clicking your ad. This is the primary data source for keyword harvesting.

The ratio of ad spend to ad-attributed revenue, expressed as a percentage. A lower ACoS generally indicates more efficient advertising. Formula: Ad Spend ÷ Ad Revenue × 100.

Ad spend divided by total revenue (organic + paid). TACoS reflects the true impact of advertising on your overall business, not just the paid channel.

In Manual campaigns, you control how closely a customer's search query must match your keyword before your ad is shown. The three types are:

Keywords you explicitly exclude from a campaign or ad group so that your ads do not show for irrelevant or unprofitable searches. Negative keywords are essential to preventing Auto and Manual campaigns from cannibalizing each other.

The ACoS at which your advertising spend exactly equals your profit margin — you are not losing money on ads, but you are not gaining either. Knowing your break-even ACoS sets the ceiling for which keywords qualify as "winners."

Before evaluating any keyword data, you need a benchmark. Without knowing your break-even ACoS, you cannot determine whether a keyword is profitable, neutral, or destroying margin.

💡 Pro Tip: If you are in a launch phase and willing to invest in rank and reviews, your target ACoS can approach or slightly exceed break-even. If you are in a profit phase, keep it well below.

Harvesting too early produces unreliable signals. A keyword with one sale and one click looks perfect — but it is statistically meaningless.

💡 Pro Tip: A term with 50+ clicks and zero conversions is a clear negative keyword candidate, even without sales data.

Navigate to Amazon Ads > Reports > Sponsored Products > Search Term Report and download the report for your desired date range. Open it in a spreadsheet tool such as Excel or Google Sheets.

Segment every search term in your report into one of three buckets based on performance against your targets:

💡 Pro Tip: Do not just chase low-ACoS keywords. Also look at absolute spend and sales volume. A keyword with a 5% ACoS but only $2 in sales is less impactful than one with a 20% ACoS and $200 in sales — both may be worth keeping, but for different reasons.

The goal is to move winners into a controlled environment. A clean, scalable structure separates keywords by match type and intent.

A common structure for a single ASIN looks like this:

💡 Pro Tip: Some sellers also run a Manual — Broad Match campaign to test keyword variations before committing to exact or phrase. This is optional but useful for mid-funnel exploration.

In Seller Central, navigate to Campaign Manager > Create Campaign > Sponsored Products > Manual Targeting. Add your harvested keywords and assign initial bids.

This is the step most sellers skip — and it is one of the most important. If you do not negative out your harvested winners from your Auto campaign, both campaigns will compete for the same queries, inflating your overall spend and making it impossible to attribute performance cleanly.

💡 Pro Tip: Use Negative Exact (not Negative Phrase) when negating harvested terms in your Auto campaign. Negative Phrase is too broad and may accidentally block related discovery terms you still want the Auto campaign to find.

Keyword harvesting is not a one-time task. Consumer search behavior changes, competition shifts, and your product may rank organically over time — all of which affect which terms are worth bidding on.

Once your harvested keywords are running in Manual campaigns, manage them like any paid keyword portfolio:

💡 Pro Tip: Avoid changing bids more than once every 7–10 days per keyword. Amazon's algorithm needs time to stabilize after a bid change, and frequent adjustments make it impossible to isolate what is working.

Seller profile: New seller, 3 months on Amazon, selling a private label supplement with a 32% profit margin.

The problem: The seller launched with only an Auto campaign and had no idea which search terms were driving sales. ACoS was 41% — above break-even — and spend was increasing without a clear return.

The action: After downloading the Search Term Report, the seller identified 6 search terms that each had 3+ orders and ACoS below 25%. They moved those 6 terms into a Manual Exact campaign, negated them from the Auto campaign, and reduced the Auto campaign's daily budget by 30% to reallocate toward the Manual campaign.

The result: Within 30 days, overall account ACoS dropped from 41% to 27%, and the Manual campaign was generating 60% of total ad revenue at a more efficient rate. The Auto campaign continued finding new terms at a lower overall budget.

Seller profile: Established seller with 12 ASINs across two categories, monthly ad spend over $15,000.

The problem: The seller had Manual campaigns running but they were built years ago using researched keywords — not harvested data. Many keywords were underperforming, and the campaigns had become bloated and hard to manage.

The action: The seller ran a full 60-day audit across all Auto campaigns, segmented search terms into winners, neutrals, and losers, and rebuilt Manual campaigns from scratch using only harvested, validated keywords. They consolidated from 40+ campaigns down to a cleaner 3-campaign-per-ASIN structure (Auto, Manual Exact, Manual Phrase).

The result: Ad spend dropped by 18% in the first month while ad-attributed revenue stayed flat, resulting in a meaningful improvement in TACoS. Campaign management time was also reduced significantly due to the cleaner architecture.

Seller profile: Intermediate seller, one year on Amazon, selling electronics accessories.

The problem: The seller correctly harvested keywords and moved them to a Manual campaign — but never added negatives to the Auto campaign. Both campaigns were bidding on the same terms, driving up CPCs through self-competition.

The action: After noticing that CPCs on their top keywords were increasing month over month with no external competitive pressure, the seller identified the internal cannibalization. They added all harvested exact-match keywords as Negative Exact terms in the Auto campaign.

The result: Average CPC on the top 10 harvested keywords dropped by 22% within two weeks. The Auto campaign began surfacing new, previously unseen discovery terms because budget was no longer consumed by already-managed keywords.

Why sellers make this mistake: Excitement about early sales leads sellers to immediately promote terms into Manual campaigns after just a handful of clicks or a single conversion.

What to do instead: Wait until a search term has at least 10–15 clicks before drawing conclusions about whether it converts. One sale does not establish a pattern. Premature harvesting leads to Manual campaigns full of terms that waste budget when the data does not hold up at scale.

Why sellers make this mistake: Sellers focus on adding keywords to their new Manual campaign and treat the Auto campaign as a background task that does not need maintenance.

What to do instead: Every keyword you harvest and add to a Manual campaign must be negated (as Negative Exact) from the Auto campaign. This prevents both campaigns from competing against each other for the same traffic, which inflates your spend and muddies your performance data.

Why sellers make this mistake: A keyword performed well in the Auto campaign, so sellers assume they need to bid aggressively in the Manual campaign to "keep" the traffic. They often set bids 2–3x the suggested bid on launch day.

What to do instead: Start Manual campaign bids at or near the actual CPC the keyword was generating in your Auto campaign. Let the Manual campaign gather 1–2 weeks of data, then adjust based on observed ACoS and impression share.

Why sellers make this mistake: After an initial harvesting session shows positive results, sellers stop reviewing Auto campaign data regularly, assuming the work is done.

What to do instead: Schedule recurring harvesting reviews — at minimum monthly, ideally bi-weekly. New search trends emerge, seasonal terms appear, and competitor dynamics shift. Your Auto campaign is constantly generating new data that deserves regular attention.

Why sellers make this mistake: Sellers see any conversion as a win and move every single converting term into a Manual campaign, regardless of volume or business significance.

What to do instead: Prioritize terms that have meaningful volume and realistic potential to scale. A keyword that converts once per month at a great ACoS may not warrant a dedicated Manual campaign slot. Focus your Manual campaign attention on terms that drive consistent, repeatable volume.

When you implement a consistent keyword harvesting workflow, you can expect the following outcomes over 60–90 days:

Progress will not be instant. Give each change 2–3 weeks to reflect in your reports before drawing conclusions, as Amazon's attribution and reporting can lag by several days.

A minimum of 2–4 weeks, but ideally longer if you have a low daily budget that limits data collection. The key threshold is not time — it is click volume. Aim for at least 10–15 clicks per search term before using it to make harvesting decisions.

No. Keep your Auto campaign running at a reduced budget if needed, but do not turn it off. Auto campaigns are your discovery engine — they surface new, relevant search terms that you would not find through manual keyword research alone. Think of Auto as a continuous input feeding your Manual campaigns over time.

Yes. Overall Auto campaign ACoS is not a reliable indicator of individual keyword performance. Auto campaigns often have high blended ACoS because they test many terms simultaneously, including poor performers. Your job is to mine the data for the individual terms that do perform well — even if the campaign average looks bad.

There is no fixed number. Focus on quality over quantity. Even 5–10 high-confidence winners in a Manual Exact campaign will outperform a Manual campaign loaded with 100 speculative keywords. Add more terms as you gather more data over time.

Start with Exact Match for your highest-confidence winners — this gives you maximum bid control and clear attribution. Add Phrase Match for terms where you want to capture relevant variations (for example, if "organic whey protein powder" converts well, a phrase campaign can also capture "best organic whey protein powder"). Use Broad Match sparingly, and treat it as another layer of discovery rather than a primary conversion tool.