Dynamic keyword insertion: the syntax, the three rules and the checks before you switch it on

Dynamic keyword insertion, which Google Ads calls keyword insertion, is a short piece of code you type into your ad text so the ad updates itself with the keyword that triggered it. Google's help page on keyword insertion, read on 10 October 2026, says it "lets you automatically update your ads with the keywords in your ad group that caused your ads to show", and gives the form {KeyWord:Chocolate}. Two details decide whether it helps you or embarrasses you: it inserts your ad group keyword, not the person's search term, and when the keyword is too long to fit in the ad, Google uses the default text you wrote after the colon.

So the quality of the feature is entirely the quality of your keyword list. Clean, short, well grouped keywords produce ads that read as though you wrote each one by hand. A broad ad group full of long tail phrases and competitor names produces ads you would never approve.

The syntax

Open brace, the word keyword, a colon, your default text, close brace. The capitalisation you use for the word keyword inside the braces is what controls the capitalisation of the inserted text, which is why Google's example is written as KeyWord rather than keyword.

Headline: Buy {KeyWord:Chocolate}
Headline: {KeyWord:Running Shoes} in Stock
Description: Order {keyword:chocolate} online today.

The default text after the colon has to stand on its own, because it is what appears whenever the keyword will not fit. Write it as the phrase you would have chosen if the feature did not exist. "Chocolate" works. "Click here" does not.

What Google's documentation makes clear

• It shows the keyword, not the search term. Google's page states that the person's search term is not always the same as the keyword, and that insertion shows the keyword in the creative, so you can anticipate what will eventually show. That is the reassuring part: everything that can appear in your ad is already in your own keyword list.
• Length decides when the default is used. In Google's own example, the keyword "gourmet chocolate truffles" is too long to fit the headline, so the ad falls back to "Chocolate".
• It works the same on all match types. Broad, phrase and exact behave identically, because the insertion is driven by the keyword rather than the query.
• Your ads still have to comply with Google Ads policies after insertion. Google repeats this point for keywords added from recommendations, which is the main route by which a keyword you never wrote ends up in your ad text.

A worked example

Ad group keyword	Headline with {KeyWord:Chocolate}
dark chocolate	Buy Dark Chocolate
sugar free chocolate	Buy Sugar Free Chocolate
gourmet chocolate truffles	Buy Chocolate, because the keyword is too long for the headline

Read the right hand column out loud for every keyword in the ad group before you launch. That one exercise catches almost every problem the feature can cause.

Where it earns its place

Insertion pays off when you have far more keywords than you could write ads for by hand, and when those keywords are already specific. A retailer with hundreds of product terms, a recruiter with hundreds of job titles, a parts supplier with model numbers, a service business covering many towns: in each case the inserted phrase is the thing the searcher typed and the thing on the landing page, so the ad, the click and the page agree.

It also helps when you are testing demand across a wide vocabulary and want the ad copy to keep pace with the ad group while you learn which terms convert. Once you know, the winning terms usually deserve their own ad group with hand written copy.

Where it goes wrong, and how to prevent it

• Misspelled keywords. If you added misspellings deliberately, they will appear in the ad. Remove them from any ad group that uses insertion and rely on close variant matching instead.
• Brand and trade mark terms. A competitor's name or a manufacturer's brand sitting in the ad group can be inserted into your headline. Keep those keywords in separate ad groups with fixed copy.
• Grammar that breaks. "Buy {KeyWord:Chocolate} for sale cheap" reads badly for half your list. Build the sentence so any keyword slots in cleanly, usually by putting the insertion at the end of the headline.
• Capitalisation. Title case on a long phrase can look like shouting. Sentence case inside descriptions usually reads better.
• Length. Count the characters of your longest keyword plus the surrounding words. If most of the group would fall back to the default, split the group rather than accepting the fallback.
• Policy wording. Terms that touch health claims, financial promises or restricted categories need checking in the inserted form, not just in the default.

Using it with responsive search ads

Insertion goes inside individual headline and description assets, and the system then mixes assets as usual. Two habits keep this predictable. Use insertion in only one or two headline assets rather than all of them, so a single fallback does not make the whole ad generic. And keep at least one hand written headline that states your offer, so the ad still works when the inserted asset is not chosen. Google documents a sibling feature, location insertion, for responsive search ads, which inserts a place rather than a keyword and suits service area businesses.

Does it change Quality Score?

Quality Score is reported as three components, and the one insertion touches is ad relevance, which Google's page on Quality Score describes as how closely your ad matches the intent behind a search. A headline containing the keyword tends to read as more relevant, which is why the feature exists. The bigger lever is still structural: ad groups tight enough that one message fits every keyword in them, and a landing page that delivers what the ad said. Insertion is a convenience on top of that work, not a substitute for it.

The organic search mistake that looks similar

Inserting the query into a page, rather than into an ad, is a different activity with different rules. Pages generated at scale around keyword permutations, with the search phrase swapped into the heading, fall under the scaled content and spam behaviour described in Google's spam policies. Dynamic text on a landing page is fine when it personalises a genuinely useful page, such as showing the selected city on a booking form. It stops being fine when the page exists only to repeat the phrase back.

This page covers keyword insertion in ad text. For what clicks cost and how the auction prices them, read how pay per click pricing works and Google Ads costs. For writing the copy itself, read the brief that produces usable ad copy.

Questions buyers ask about dynamic keyword insertion

What is dynamic keyword insertion?

Code in your ad text, written as {KeyWord:default text}, that Google Ads replaces with the ad group keyword that caused the ad to show. When that keyword is too long for the space, the default text appears instead.

Does it insert what the person typed?

No. It inserts your keyword. Google's documentation states that the search term is not always the same as the keyword, which is what keeps the feature predictable: only phrases you added can appear.

How do I control capitalisation?

By how you capitalise the word keyword inside the braces. Google's own example uses KeyWord, and lowercase inside a description usually reads more naturally than title case.

Is keyword insertion still worth using?

Yes, for accounts with large, specific keyword lists where hand writing every ad is impractical. For a handful of high value terms, hand written copy with a sharper offer does more.

Can it get an ad disapproved?

The inserted ad has to meet the same advertising policies as any other, so check the assembled version of your longest and most sensitive keywords, and keep restricted terms in their own ad groups.

What is the safest way to start?

Pick one tightly themed ad group, put insertion in one headline asset, write a default you would be happy to show every time, and read the assembled ad for every keyword in the group before launch.

The query you cannot bid on yet

There is a growing slice of buyer research that never reaches an auction. People ask ChatGPT, Gemini, Perplexity or Claude who to use in a category, or read Google's AI Overview, and get back a handful of named companies drawn from directory profiles, reviews, comparison pages and press coverage. There is no keyword to insert and no bid to raise, and the names that come back differ by assistant because each one reads different sources.

That makes it worth measuring the same way you measure impression share. Ask the buying question your best keyword stands for, keep your own company name out of the prompt so the answer comes back clean, and note who is named and which pages get cited. AI Knows Us measures those answers and the sources behind them.