The auto-targeting problem nobody talks about
Amazon's Sponsored Products auto-targeting is the default for most new campaigns. You set a budget, pick your products, and Amazon's algorithm decides which search terms to show your ads for.
The problem? You're paying for every click — including clicks from search terms that will never convert for your product.
A seller in the phone accessories category might find that auto-targeting is showing their premium silicone case ad for searches like "cheap phone case" or "phone accessories bulk." These clicks cost $1-3 each and convert at near zero. Over a month, that's $200-500 in pure waste.
Step 1: Download your Search Term Report
In Seller Central, go to Reports → Advertising Reports → Search Term Report. Download data for the last 30-60 days. This report shows every search term that triggered your auto-targeting ads, along with spend, clicks, and sales data.
This is the raw data that reveals where your money is actually going.
Step 2: Identify zero-conversion terms
Filter for search terms where:
- Clicks > 5 (enough data to judge)
- Orders = 0 (no sales generated)
- Spend > $5 (meaningful waste)
These are your "bleeding keywords" — terms that Amazon's algorithm thinks are relevant, but that real shoppers aren't converting on. Sort by spend descending. The top 10 terms are usually responsible for 60-80% of total waste.
Step 3: Categorize the waste
Not all zero-conversion terms are equal. They typically fall into three categories:
Category A: Completely irrelevant
Search terms that have nothing to do with your product. Example: your product is a premium leather wallet, but you're getting clicks from "cheap wallet under $5." These should be added as exact-match negative keywords immediately.
Category B: Related but wrong intent
Terms that are related to your product category but indicate a different buying intent. Example: "phone case dimensions" (research intent, not buying intent). These are candidates for phrase-match negatives.
Category C: Promising but underperforming
Terms that seem relevant but haven't converted yet. These might need more data before making a decision. If a term has fewer than 15 clicks, it may simply need more time. Consider moving these to their own exact-match campaign where you can control bids precisely.
Step 4: Take action — the 3 priority moves
Based on your categorization, make three specific changes this week:
- Negate the top 3-5 Category A terms as exact-match negatives in your auto campaign. This stops the biggest bleed immediately.
- Add phrase-match negatives for Category B patterns. If "dimensions" is a common non-converting modifier, negate the phrase.
- Promote your best Category C terms to exact-match manual campaigns. This gives you bid control on terms that might convert with the right placement and bid.
Step 5: Measure the impact
After one week, pull another Search Term Report and compare:
- Has total auto-targeting spend decreased?
- Is ACoS trending down?
- Are the negated terms actually gone from your reports?
Repeat this process weekly. The waste compounds if you don't. One new irrelevant term at $3/day is $90/month — and auto-targeting generates new ones constantly.
The compounding cost of inaction
Here's what makes auto-targeting waste particularly dangerous: it compounds silently.
- Week 1: 5 wasted terms × $3/day = $15/day = $105/week
- Week 4: 15 wasted terms (new ones added by Amazon) × $3/day = $45/day = $315/week
- Month 3: Without intervention, auto-targeting waste can represent 30-40% of your total SP spend
The sellers who maintain low ACoS on auto-targeting aren't doing anything magical. They're doing this review process consistently — catching waste before it compounds.
Automating the process
The manual review process works, but it takes 2-3 hours per week per account. At scale (multiple products, multiple campaigns), it becomes unsustainable.
This is exactly why we built TermPilot's Waste Radar. It automates the analysis above — scanning your Search Term Reports, categorizing waste patterns, and delivering your 3 priority actions every week. What takes hours manually takes 30 seconds with Waste Radar.