Semantic DNA (N-Gram Analysis)

Your Search Terms report shows 2,000 queries. You review them one by one, add negatives, and move on.

The problem: You're looking at trees and missing the forest.

Example:

You have 47 different search queries containing the word "free":

Reviewing them individually, some look okay (free trial intent). But in aggregate, the word "free" across all 47 queries costs $3,200/month with 0.3x ROAS.

N-Gram Analysis reveals this pattern in one row: free | 47 queries | $3,200 spend | 2 conversions | 0.3x ROAS

One negative keyword ("free") eliminates $3,200/month in waste across 47 queries.

That's the power of pattern-level analysis vs query-level analysis.

Unigrams (single words):

Most useful for finding broad waste patterns ("free", "cheap", "jobs") or broad opportunity signals ("enterprise", "premium", "urgent").

Bigrams (two-word pairs):

The sweet spot for actionable insights. "Free trial" is very different from just "free." Bigrams capture intent more precisely.

Trigrams (three-word sequences):

Highly specific. "Best enterprise CRM" or "free project management" — these often map directly to ad group themes.

The N-Gram Table:

Signal classification:

🟢 Opportunity: ROAS >2x account average, 5+ conversions

🟡 Mixed: ROAS near account average, needs investigation

🔴 Waste: ROAS <0.5x account average OR 0 conversions with >$500 spend

What it does: Classifies n-grams by inferred user intent based on keyword modifiers.

Intent Groups:

🛒 Transactional: Contains buy, purchase, order, pricing, cost, subscribe, signup

Expected ROAS: Highest (users are ready to convert)

🔍 Informational: Contains how, what, why, guide, tutorial, learn, difference

Expected ROAS: Lowest (users are researching, not buying)

⚖️ Comparison: Contains best, top, review, vs, compare, alternative

Expected ROAS: High (users are evaluating options, close to purchase)

📍 Local: Contains near me, in [city], local, nearby

Expected ROAS: Varies (high for services, lower for products)

Why this matters:

If 60% of your spend goes to Informational n-grams with 0.5x ROAS, you have a targeting problem. Your broad match keywords are attracting researchers, not buyers.

Action: Add Informational modifiers as negative keywords ("how to", "what is", "tutorial") to focus budget on Transactional and Comparison intent.

But be careful: Some Informational queries convert at high rates if your landing page serves the educational need and then converts. Check before excluding.

The most powerful feature: Finding n-grams that appear across many queries and consistently waste money.

The Waste Leaderboard:

Ranks n-grams by total wasted spend (spend on queries with <1x ROAS containing this n-gram).

Example Waste Leaderboard:

Total waste from top 5 n-grams: $8,770/month

Adding these 5 words as negative keywords eliminates waste across 121 search queries simultaneously.

Compare this to manual query review: You'd need to identify and negative-match 121 individual queries. With N-Gram Analysis, it's 5 negative keywords.

One-Click Actions:

The flip side of waste detection: Finding n-grams that consistently produce high ROAS and should be promoted.

The Opportunity Leaderboard:

Actions for high-performing n-grams:

Cross-reference with Creative Lab: If an n-gram drives high ROAS but your ad copy doesn't mention it, you're missing an alignment opportunity.

What it shows: How n-gram performance changes over time.

Visual: Sparkline charts showing ROAS and spend trends for each n-gram over 90 days.

Why trends matter:

A waste n-gram today might have been profitable 3 months ago. A profitable n-gram might be declining.

Trend patterns:

📈 Rising Star: Low spend initially, ROAS climbing. Worth expanding into a dedicated ad group.

📉 Fading Giant: Was profitable, now declining. Audience saturation or competitive pressure. Consider reducing exposure.

🔄 Seasonal: Performance peaks and dips cyclically. Adjust bids seasonally rather than permanently excluding.

📊 Stable Performer: Consistent ROAS over 90 days. Your foundation keywords—protect and maintain.

Monthly review: Sort by "ROAS Change (30d)" to quickly identify n-grams with the biggest positive or negative shifts.

N-Gram Analysis is a monthly exercise. You need 30 days of data to identify reliable patterns.

Third Monday of Every Month (30 minutes):

Step 1: Waste Scan (10 min)

Open the Waste Leaderboard. Add the top 3-5 waste n-grams as negative keywords. Check the "Review Queries" before adding to avoid false positives (e.g., "free trial" might be profitable even though "free" is waste overall).

Step 2: Opportunity Scan (10 min)

Open the Opportunity Leaderboard. Any new high-performing n-grams? If an n-gram has >5 conversions and >3x ROAS, consider a dedicated ad group.

Step 3: Intent Audit (5 min)

Check the Intent Pattern breakdown. Is Informational spend increasing? If so, tighten keyword match types or add intent-based negatives.

Step 4: Trend Check (5 min)

Sort by ROAS Change (30d). Any rising stars to expand? Any fading giants to reduce?

Success Metrics:

Use N-Gram Analysis when you want to:

Don't use N-Gram Analysis when you want to:

Who Should Use This:

✅ PPC specialists (advanced keyword optimization)

✅ SEO teams (search intent insights transfer to organic)

✅ Content strategists (understanding what language your audience uses)

❌ Executives (too granular)

❌ Beginners (start with Search Hygiene for query-level review)

How Often: Monthly. Weekly only if you just launched broad match campaigns and need rapid waste containment.