By: Matt Shenton, Paid Search Expert at Croud
If you're managing Google Ads, you've probably noticed a big shift over the last few years: everything has gotten much, much broader. While this means more opportunities to reach potential customers, it also means a flood of search term data. How do you make sense of hundreds of thousands of search terms to find what's truly driving performance?
The answer lies in Natural Language Processing (NLP), specifically a technique called n-gram analysis. This powerful method can help you uncover hidden opportunities and insights within your Google Ads data, making your campaigns more efficient and effective.
In this article, we'll explore:
The Challenge of Broad Match in Google Ads: Why the increase in search term data demands a new approach.
What is N-gram Analysis? A simple explanation of how this NLP technique works.
Applying N-gram Analysis to Your Google Ads Data: A practical guide to uncovering high and low-performing phrases.
Actionable Insights from N-grams: How to use your findings to improve ad copy, landing pages, and bidding strategies.
The Future of NLP in Paid Search: Why understanding textual data remains crucial in an AI-driven advertising world.
Over the past five years, Google Ads has moved towards broader matching. While campaign types like Performance Max and AI Max for Search are contributing to this, the adoption of broad match keywords has been a significant driver. This shift has led to an explosion in search term data.
Imagine this: an account that might have had 10,000 search terms five years ago could now be generating 250,000. That's a 25x increase in the amount of raw, textual data you're sifting through. On one hand, this is fantastic – it means you have a wealth of information directly from people searching for your products or services. You can see exactly how they're searching and what their intent might be.
However, herein lies the problem: as humans, we simply can't manually review 250,000 search terms to identify patterns or performance trends. This is where natural language processing comes to the rescue.
Natural Language Processing (NLP) is a field of artificial intelligence that helps computers understand and process human language. N-gram analysis is a specific NLP technique that breaks down text into sequences of "n" consecutive terms.
Let's look at an example to make this clearer: Imagine a search term: "I want to learn Photoshop."
1-gram (unigrams): Breaks it into single words: "I", "want", "to", "learn", "Photoshop"
2-grams (bigrams): Breaks it into two-word phrases: "I want", "want to", "to learn", "learn Photoshop"
3-grams (trigrams): Breaks it into three-word phrases: "I want to", "want to learn", "to learn Photoshop"
4-grams: "I want to learn", "want to learn Photoshop"
5-grams: "I want to learn Photoshop"
By repeating this process across all your search terms, you can generate millions of these n-grams. When you combine this with your performance data (clicks, conversions, cost per acquisition), you can start to identify which specific words or phrases are performing well, and which are draining your budget.
So, how do you put this into practice?
Export Your Search Term Report: Download your comprehensive search term report from Google Ads. Make sure it includes performance metrics like conversions and cost.
Use an N-gram Script or Tool: There are many n-gram scripts available online (a quick search for "Nils Rumen n-gram script" will give you a good start), or you can even use AI tools to help generate one. These scripts will take your raw search terms and produce a table of n-grams with their frequency and aggregated performance data.
Analyze the Output: The output will typically show you the n-gram, its length (e.g., 2-word, 3-word), its part of speech (e.g., noun, verb), its frequency (how many times it appeared), and the combined performance metrics for all search terms containing that n-gram.
Once you have your n-gram analysis, you can start digging for insights. Think of your n-grams as individual puzzle pieces that, when grouped, reveal the bigger picture of your account's performance.
1. Identify High CPA N-grams (Budget Drainers)
These are the phrases that are costing you a lot but aren't converting efficiently. Common examples often include:
High-level research terms: Phrases like "how to [product type]" or "what is [service]" might indicate users who are still in the early stages of their research and not ready to convert. While valuable for creative or landing page inspiration, they might be inefficient for direct conversion campaigns.
Irrelevant terms: Even with broad match, some irrelevant queries can slip through. N-grams can quickly highlight these. For a fashion brand, for instance, an n-gram like "sustainable" might have a high CPA if the brand's sustainable offerings aren't competitive or clearly communicated.
Action: Add these high-CPA n-grams as negative keywords to prevent your ads from showing for these less profitable searches. Consider if the topic is still relevant for content marketing or upper-funnel strategies.
2. Discover Low CPA N-grams (Performance Boosters)
These are the gems – phrases that are driving conversions at a low cost.
Specific product or service categories: You might find that terms related to a particular sub-product or service line consistently perform well. For example, "luxury leather handbag" might have a much lower CPA than just "handbag."
Promotional terms: Phrases like "free voucher" or "discount code" often indicate high purchase intent.
Actions:
Prioritize these terms: Consider increasing bids for keywords containing these high-performing n-grams.
Create dedicated ad groups: If an n-gram points to a specific sub-category, create a dedicated ad group with tailored ad copy and landing pages to capture this demand more effectively.
Integrate into creative: Highlight these winning phrases in your ad headlines and descriptions.
3. Optimize Ad Copy and Landing Pages
Beyond just negative keywords, n-grams offer insights for improving your entire ad experience:
Ad Copy: If you see a lot of "how-to" or question-based n-grams, consider crafting responsive search ad headlines that directly address these questions. This makes your ads more relevant to the user's intent.
Landing Page Content: Your landing pages are going to be increasingly important, especially with AI-driven campaigns like AI Max for Search. If your n-gram analysis shows a high volume of searches around a particular topic (e.g., "eco-friendly materials" for a fashion brand), ensure your landing page comprehensively answers these queries with clear, factual content and relevant schema markup. This helps Google understand the context of your page and match it with relevant AI Overviews.
While n-gram analysis has been around for a while, its importance is growing as Google Ads relies more heavily on AI and broad matching. Landing page content, in particular, will serve as a key input for Google's AI to understand your offerings and match them with user queries.
NLP techniques can help you:
Understand Google's Perspective: By analyzing your own landing page text with n-grams, you can gain insight into the topics and themes Google's AI is likely extracting.
Competitive Analysis: Run n-gram analysis on competitor landing pages to understand their topical focus and identify potential gaps or opportunities for your own content strategy.
As long as we have access to textual data from search terms and landing pages, natural language processing will remain an invaluable technique for uncovering actionable insights in the ever-evolving world of paid search.
Matt Shenton is a Paid Search expert and Biddable Director at Croud. He’s the host of The Paid Search NYC Podcast and organizes regular Paid Search events in New York City, bringing the PPC community together to share ideas, trends, and strategies.
You can find him on LinkedIn