Search terms
(close variant)
(close variant)
(close variant)
(close variant)
(close variant)
(close variant)
(close variant)
(close variant)
N-gram analysis
Find recurring search-term patterns and preview negative keyword impact.
Reviewing a full Search Terms report query by query can burn a lot of PPC time, especially when the account has many low-volume variants. N-gram analysis is a practical way to compress that review: it groups repeated words and phrases, shows their combined metrics, and turns scattered search-term patterns into candidates you can actually evaluate.
What n-gram analysis means
In MirachSEM, search-term metrics are grouped by every repeated 1-word, 2-word, and 3-word phrase found in the report. Those phrases are the n-grams. It is a simple mechanical grouping method, but it is useful because each group already gives you a keyword-like label for the pattern behind a set of queries.
Phrases that spend money without conversions are natural negative keyword candidates. High-traffic phrases can point to SEO article ideas, landing-page gaps, or separate campaign and ad group themes. Strong converting phrases may deserve their own keyword coverage. N-gram analysis does the grouping; the PPC specialist still decides what each group means for the account.
Connect the source data
MirachSEM runs inside the browser, so by default it can only read the rows currently available on the page. For n-grams, that means the initial analysis is based on the visible Search Terms report page. It can be enough for a small report, or for a quick directional check after sorting and filtering, but it is not a complete account-level analysis.
For a full review, download the current Search Terms report as a CSV from Google Ads and upload that file in the Source section of the N-grams tab. After upload, MirachSEM calculates n-gram counts and metrics from the CSV. If the report changes in Google Ads, download a fresh file before making decisions; an old CSV can make the panel disagree with the live report.
Review the active list impact
MirachSEM also estimates what your current active negative keyword list would have removed if it had been applied during the report period. Terms removed by active list shows the matched search terms and the spend, clicks, conversions, and CPA that would have been removed with them.
When you enable Simulate list removal, candidate n-gram metrics are recalculated as if the active list had already been applied. This helps avoid double-counting the same wasted traffic and prevents you from spending review time on queries that are already covered by another negative keyword.
Configure the n-gram list
Use N-gram controls to set the match mode, filters, and sorting before you start working through candidates. The right setup depends on the review goal. For example, the Waste filter keeps phrases whose matched search terms have no conversions, which is useful when the session is focused on negative keyword cleanup.
Match mode changes which search terms are counted for the same n-gram. For buy leather sofa, Exact counts only that exact query, Phrase also counts queries where that phrase appears inside a longer search term, and Broad counts queries where all three words appear anywhere in the search term.
Add a negative from an n-gram
Once the filters are set, the workflow is similar to a normal Search Terms review: move through the rows, decide whether the traffic is useful, and choose what to do with it. The difference is that each row represents a rough group of related queries instead of one query at a time.
Click Add on an n-gram that should be affected by your negative list. In the review screen, you can edit the keyword before adding it. For example, if free breakfast has spent without converting, you may decide that the real blocker is just free, not the full phrase.
You can also change the match type before saving. Editing the keyword or switching match type changes the set of search terms that would be removed, so MirachSEM updates the preview metrics and the matched query list before you add the negative keyword to the active list.