Tools
Search Term N-Gram Analyzer
Use this n-gram analysis tool on your Google Ads search terms report: it totals clicks, cost and conversions for every recurring 1, 2 and 3 word phrase, so you can spot the phrases that burn budget and export them as negative keywords. Free, no sign-up.
Result
- Digital Marketing Expert
- Google Partner
- Full Stack Developer
- Last updated
- Based on
- 3 sources
How to use the Search Term N-Gram Analyzer
- Export the report
Open the search terms report in Google Ads, pick a date range of at least 30 days and download the table as CSV or Excel CSV. Copying the rows with their header works too.
- Paste or upload
Paste the text into the box or use the Upload CSV button. The tool finds the delimiter, the header row and the number format on its own; the Columns panel shows the mapping.
- Choose the settings
Type your brand terms, separated by commas, so brand searches do not inflate the averages. Leave the wasted spend threshold empty to use your account cost per conversion.
- Review the phrases
Switch between the 1, 2 and 3 word tabs. Use the Wasted filter for phrases without conversions and the Expensive filter for phrases below the account average; click a column header to sort.
- Export your negatives
Tick the phrases, choose a match type, then copy or download the list. If your selection brought conversions, the tool warns you before you paste the list into Google Ads.
Which metrics does the tool calculate?
First the tool merges rows that carry the same search term, then it splits every term into words. For each phrase it adds up impressions, clicks, cost, conversions and value, and it derives the ratios from those totals.
clicks / impressionscost / clickscost / conversionsconversions / clicksconversion value / costconversions = 0 and cost ≥ threshold (default threshold: account CPA)ROAS when value exists, otherwise CPA, compared with the account average: CPA above 1.5 times or ROAS below 0.67 times means Expensive; CPA at 0.75 times or less or ROAS at 1.33 times or more means EfficientEach phrase takes its share from a search term only once. However, one search term feeds several phrases, so the phrase rows add up to more than the account total. The Expensive and Efficient limits are a practical rule of this tool, not a Google threshold.
What did the tool find in the sample report?
The rows come from the fictional running shoe store report that the Sample data button loads. Account ROAS is 1.78 and the automatic wasted spend threshold is $56.97.
| Phrase (tab) | Terms | Cost | Conv. | Flag and decision |
|---|---|---|---|---|
| used (1 word) | 7 | $490.41 | 0 | Wasted spend: add "used" as a phrase match negative |
| how to choose (3 words) | 7 | $397.70 | 0 | Wasted spend: research intent, add as negative or point it to a guide |
| cheap (1 word) | 5 | $916.70 | 7 | Expensive: ROAS 0.67 against 1.78; lower the bid, keep the traffic |
| free (1 word) | 15 | $330.91 | 7.07 | Do not exclude: every conversion comes from free shipping searches |
| free shipping (2 words) | 10 | $272.61 | 7.07 | Efficient: ROAS 2.46, worth its own keyword |
| trail running (2 words) | 16 | $349.62 | 9 | Efficient: ROAS 2.61, candidate for a separate ad group |
The same word can tell two different stories in two tabs. Before you exclude a single word, search for it in the 2 and 3 word tabs.
Common negative phrase types
The intents below cause wasted spend in many sales accounts. Still, check every row against your own report; a phrase that wastes money in one account can sell in another.
| Intent | English examples | German examples | Turkish examples | Note |
|---|---|---|---|---|
| Job seekers | jobs, salary, career | jobs, gehalt, stellenangebote | iş ilanı, maaş, kariyer | A frequent negative in sales accounts |
| Free seekers | free, gratis | kostenlos, gratis | bedava, ücretsiz | Protect phrases such as free shipping |
| Research | what is, how to, examples | was ist, wie, anleitung | nedir, nasıl | Send it to a content campaign if you run one |
| Used and rental | used, second hand, rental | gebraucht, mieten | ikinci el, kiralık | Exclude if you only sell new products |
| Repair and parts | repair, parts | reparieren, ersatzteile | tamiri, yedek parça | Exclude if you offer no service |
| Competitors and places | competitor names, cities | Wettbewerber, Städte | rakip marka, şehir | Depends on your strategy; check performance first |
What is n-gram analysis for search terms?
N-gram analysis groups the real searches behind your ads into recurring phrases, so you can see where the budget goes. The Google Ads report lists single search terms. In a large account, however, you face thousands of terms and most of them get only a few clicks. That makes a decision on one term risky.
The n-gram method solves this. The tool splits every search term into words, then adds up impressions, clicks, cost and conversions for each phrase of 1, 2 and 3 words. For example, if the phrase used appears in ten different terms, the spend of those ten terms lands on one row. As a result, small leaks become visible.
Talha Aslan and the team run this analysis on a regular schedule in the accounts we handle through our Google Ads management service. The tool runs entirely in your browser, and your report never reaches our server. In the end you see the phrases that waste money and the phrases that deserve more budget in one table.
Where do you get the search terms report?
Open the search terms report in Google Ads; in the current interface it sits under Insights and reports. Pick a date range of at least 30 days, because a short range leaves too little data per phrase. Then use the download icon and choose CSV or Excel CSV. The tool reads both, and it also decodes the UTF-16 file that Excel CSV produces.
Column names change with the interface language. The tool knows English, German and Turkish headers, so Search term, Suchbegriff and Arama terimi all map to the same column. The Spend and Revenue columns of a Microsoft Ads report also map to cost and value. If a header stays unknown, you pick the column yourself under Columns.
For privacy reasons, Google leaves low volume searches out of the list of individual terms. The Google Ads Help page explains this rule. Consequently, the phrase table can total a little less than your account spend. The tool notes Total rows and grouped rows separately and keeps them out of the phrases.
How should you read 1, 2 and 3 word phrases?
Single words give you the widest view. Words like cheap, repair or jobs end up on one row, whatever term they appear in. On the other hand, a single word loses its context. In the sample report the word free looks like a classic negative, but every one of its conversions comes from free shipping searches.
Two word phrases bring that context back. Free shipping gets a row of its own, with all 7.07 conversions and a ROAS above the account average. Three word phrases then come closest to the intent. Especially in long tail accounts, the 3 word tab shows clearly which question patterns eat the budget.
Keep one rule in mind: each search term feeds every phrase inside it once. The term used trail running shoes appears on the used row and on the trail running row. So do not add up phrase rows; their sum always exceeds the account total. Instead, compare each row with the account average on its own.
How does n-gram analysis flag wasted spend?
The key question in n-gram analysis is when a phrase without conversions really counts as waste. By default, the tool uses your account cost per conversion as the threshold. In other words, a phrase that spent as much as an average conversion costs, and brought none, gets the Wasted spend flag. Phrases below the threshold without conversions show Low data instead.
This rule is careful, but it does not fit every account. In an account with a low conversion rate, doubling the threshold is safer. In contrast, a high volume account can lower the threshold and add negatives earlier. Whatever value you type into the threshold box, the table updates its flags right away.
Also allow for conversion lag. A click happens today, while the sale may follow a few days later. Therefore do not exclude phrases that look empty in a report ending yesterday. To see how the saved money fits your overall plan, use the Google Ads budget calculator next to this tool.
Which match type should your negative keywords use?
The tool writes the list in three syntaxes. Phrase match uses quotes: the negative "used shoes" blocks every search that contains those two words in that order. Exact match uses brackets: [used shoes] blocks only that exact search. Broad match has no marks and blocks a search when all its words appear, in any order.
For 2 and 3 word phrases, phrase match is usually the most balanced choice. For single words, a broad negative can shut out far more than you expect. So look at the other tabs before you exclude a single word.
Negative keywords do not cover close variants. That means the negative shoe does not stop the search shoes; singular and plural need a line each. To compare how the match types behave on the keyword side, the keyword match type tool makes the job easier.
Which phrases should you never exclude?
Excluding phrases that convert is the most expensive mistake in this work. To lower that risk, the tool adds up the conversions of your selection and shows a warning when the total is above zero. When you see the warning, replace the selection with longer phrases.
Generic words that describe your product should not become negatives either. For example, running or shoes may look expensive, simply because most of the account contains them. In such a case the problem is not the word but the intent that comes with it. So look for the answer in the 2 and 3 word tabs.
Treat brand searches separately as well. Brand terms usually convert cheaply and make the account average look better than it is. That is why the settings include a brand box. Once you fill it in, the real performance of generic searches shows up and the thresholds become fairer.
What should you do with expensive and efficient phrases?
An Expensive flag does not mean negative. These phrases convert, only at a higher cost than the account average. As a first step, lower the bids on keywords that contain the phrase, or move it into its own ad group and test a better landing page. To find out what the phrase really earns, set your target with the ROAS calculator.
Efficient phrases, on the other hand, are growth chances. In the sample report, trail running brings a ROAS well above the account average. Add a phrase like that as an exact match keyword, feature it in your ad copy and protect its budget.
Leave phrases with the Low data flag for the next round. They have not spent enough yet to support a decision. Meanwhile, the conversion rate calculator helps you check your targets.
How often should you repeat the analysis?
For accounts with a high daily spend, a weekly check works well. Smaller accounts need a monthly check, because phrases take time to collect meaningful data. If you use the same date range each time, comparing results becomes much easier.
Instead of adding negatives campaign by campaign, build a shared negative keyword list. That way you manage phrases like jobs or repair in one place for every search campaign. After that, add only the new phrases after each analysis.
Updating your budget plan alongside is just as important. Our article on how to set a Google Ads budget explains where to move the money you stop wasting. To turn efficient phrases into new keywords, read our guide on finding keywords that drive sales.
Common mistakes in n-gram analysis
- ✕MistakeExcluding a single word right away.✓Do this insteadCheck the 2 and 3 word tabs first; a word like free may convert inside free shipping.
- ✕MistakeDeciding on a few days of data.✓Do this insteadUse a report of at least 30 days and read the last days carefully because of conversion lag.
- ✕MistakeAdding up phrase rows and comparing the sum with the account total.✓Do this insteadOne term feeds several phrases, so the rows overlap and their sum exceeds the account total.
- ✕MistakeForgetting singular and plural in negatives.✓Do this insteadNegative keywords do not cover close variants; add shoe and shoes on separate lines.
- ✕MistakeLeaving brand searches in the analysis.✓Do this insteadType your brand terms into the settings; brand conversions make the account average look better than it is.
Frequently Asked Questions
Are search terms eating your budget?
Talha Aslan and the team review search terms on a regular schedule and build negative lists and match types around your account.
Related Articles
Blog
How to Set a Google Ads Budget for B2B Service FirmsRead article →
