Google Ads Search-Term N-Grams: Find Repeated Intent Across Queries

A search-term n-gram analysis splits reported queries into repeated one-, two-, or three-word sequences, then aggregates their cost and outcomes. Use it to surface patterns hidden across many low-volume rows, but return to the original queries before adding negatives or keywords. An n-gram is a clue, not a targeting decision.

What is an n-gram analysis in Google Ads?

It is an analyst-created view of search-term data, not a special Google Ads campaign type. A unigram is one token, a bigram is two adjacent tokens, and a trigram is three. Aggregating those sequences can reveal a repeated intent modifier even when no single query looks large.

Start from the Search terms report, which Google describes as reported searches that triggered an ad. Privacy thresholds and reporting scope mean the export should not be treated as a complete record of every possible query.

How do you prepare search terms without distorting them?

• Export a declared date range with campaign, ad group, cost, conversions, and value.
• Normalize case and ordinary punctuation while preserving meaningful model numbers, locations, and negation.
• Keep the original query beside every generated n-gram.
• Do not silently remove words that change intent, such as “not,” “near,” or “jobs.”
• Document how plural forms, spelling, and stop words are handled.

The regular search terms report guide explains the row-by-row review. N-grams add a second lens; they do not replace it.

Which patterns deserve attention?

Look for repeated research intent, employment intent, unsupported locations, product categories you do not sell, and high-value modifiers worth their own landing-page review. Compare cost and outcomes, but avoid declaring a phrase “bad” solely because conversions are zero in a thin sample.

A strong positive pattern may support keyword research, ad-group structure, or landing-page copy. A negative pattern may support an exclusion only after the full queries confirm that the phrase is consistently irrelevant.

How do you turn an n-gram into a safe negative?

• Open every material source query containing the sequence.
• Check whether the phrase has more than one business meaning.
• Choose negative match type intentionally.
• Preview which valuable searches the negative could block.
• Add it at the narrowest useful level and log the decision.

Google notes that negative match behavior differs from positive keywords and that report-added negatives default to exact match in the documented flow. Use the official negative-keyword workflow alongside the negative keywords guide.

How often should you rerun the analysis?

Run it after enough new search-term data accumulates to reveal patterns, and after material changes to match types, products, geography, or landing pages. Compare periods using the same normalization rules. Include the review in the Monday pass when query volume warrants it.

Keep the analysis reproducible and the account change human-readable. A compact phrase table is useful only when each action can still be traced back to real searches.