Google Ads Keyword Count: Why 8 to 20 Per Ad Group Wins

Use 8 to 20 closely related keywords per ad group. For a small Australian account, that adds up to somewhere between 50 and 200 keywords in total. Those two numbers are the short answer, and they hold for an account spending AUD 1,500 a month as well as one spending AUD 40,000. A bigger budget changes how many ad groups you build, not how many keywords sit inside each one.

How many keywords should you use in Google Ads?

The right count depends on three things: your monthly budget, your match types and how many conversions the account produces. The number of search terms that exist in your market barely matters. A plumber in Geelong and a national retailer both run ad groups of roughly the same size, because the limit on a useful ad group is how much data each keyword can collect.

Use these ranges to sanity check your own account:

• Per ad group: 8 to 20 keywords that share one search intent. Fewer than 5 usually means you have split too finely. More than 30 usually means the ad group has lost its theme.
• Per campaign: 40 to 200 keywords across 5 to 15 ad groups, assuming one campaign per product line, service or location.
• Per account: 50 to 200 keywords for a small local business, 500 to 2,000 for a mid-size account. Large ecommerce accounts run higher, but rarely above 10,000 active keywords.
• Negative keywords: often more than your active list. A mature account commonly carries two to three times as many negatives as positives.

Nothing in those numbers says a bigger budget needs more keywords. Extra spend buys more impressions on the keywords you already have.

Keywords also belong to the ad group, not to individual ads. Every ad in an ad group serves all of that group’s keywords, which is why a tight theme matters so much. How many ads to run alongside those keywords is a separate question, covered in our guide to how many ads belong in one ad group.

What are the official Google Ads keyword limits?

Google’s account limits page allows up to 20,000 ad group targeting items per ad group and 5 million across an entire account. That is roughly a thousand times more than a sensible ad group needs. These are technical ceilings built for the largest advertisers in the world, not recommendations.

The limits worth knowing:

• 20,000 ad group targeting items per ad group, which covers keywords, placements and audience lists combined.
• 5 million ad group targeting items per account, across all campaigns.
• 20,000 ad groups per campaign and 10,000 campaigns per account, including paused ones.
• 10,000 negative keywords per campaign.
• 5,000 keywords per negative keyword list, with up to 20 shared lists per manager account and 20 per child account.

The gap between the ceiling and the sensible number is the whole lesson. Google gives you room for 20,000 keywords in one ad group, and you should use between 8 and 20 of them. Treating the limit as a target is one of the most expensive structural mistakes an account can make.

Does Google recommend a specific number?

No. Google has never published a recommended keyword count per ad group, and its official guidance on building a keyword list avoids numbers altogether. What it does say is to “make sure your keywords directly relate to the theme of your ad and the page you’re directing your customers to”, that “keywords with 2-3 words tend to work most effectively”, and that you should try “grouping your keywords into themes”.

Read that closely and the real rule appears. Google is describing thematic tightness rather than quantity. The number of keywords in an ad group is a consequence of how tightly you have themed it, never a target you set first.

Why more keywords usually means worse performance

Every keyword you add divides the same budget into smaller pieces and gives Google’s bidding less data per keyword to learn from. Past about 20 keywords in an ad group, that trade tends to go against you.

Your budget splits thinner across every keyword

Take an illustrative AUD 3 a click. Competitive categories such as legal, insurance and finance run far higher, so treat this as a floor rather than a benchmark.

At that price, AUD 1,500 a month buys about 500 clicks, before you add what PPC management costs on top. Spread 500 clicks across 200 keywords and each one averages two and a half clicks a month. No keyword will ever collect enough data to show whether it works. Spread the same clicks across 30 keywords and each averages 17, which is where a usable signal starts.

Smart bidding starves without conversion data

Google’s Target CPA help page recommends measuring performance over the last 30 days, including at least 30 conversions. That threshold applies to the bid strategy as a whole, but the model still learns faster when conversions concentrate rather than scatter.

Forty conversions a month across 40 keywords hands the algorithm one conversion per keyword. Concentrate the same 40 conversions into 12 keywords and each gets three or four, in patterns it can act on. An over-long keyword list is one of the most common reasons smart bidding underperforms.

Duplicate keywords compete against each other

When two keywords in different ad groups can match the same search, Google picks one and leaves the other idle. Coverage has not doubled. You have created a second keyword with none of the first one’s history, plus a reporting problem that hides which of them is running.

This happens far more often than you would expect, because match types stopped being literal years ago.

Reporting stops being actionable

A 30 keyword account can be reviewed properly in 20 minutes. A 2,000 keyword account cannot be reviewed at all, so it gets managed by aggregate and the decisions that matter never get made. Most of those 2,000 keywords will show zero impressions, and the handful driving results stay buried.

This is the state most accounts are in when they come to us for a Google Ads audit. The fix is rarely a new campaign; most of the work is deleting what is already there.

How match types change the number you need

Broad match needs the fewest keywords and exact match needs the most, because each match type covers a different share of search volume per keyword. Pick your match types first, then set your keyword count.

Broad match: 3 to 8 per ad group

One broad match keyword paired with smart bidding can cover hundreds of search variations by itself. In this setup, 3 to 8 keywords per ad group is often plenty, and adding more creates overlap rather than reach.

Broad match only behaves this way when conversion tracking works properly and a real negative keyword list is in place. Without both, it spends money on searches that were never going to convert.

Phrase match: 8 to 15 per ad group

Phrase match covers meaningful variation while keeping a boundary around intent, which makes it the sensible default for most Australian small business accounts. Since 2021 it has also absorbed the old broad match modifier, which Google retired as a separate match type.

Exact match: 15 to 25 per ad group

Exact match is the narrowest per keyword, so a topic takes more of them to cover. It also leaves the least room to discover new search terms, which is why exact-only accounts tend to plateau.

Check one caveat before building that list. Exact match has not been literal since 2018. Google’s close variants definition now lets an exact match keyword trigger on misspellings, plurals, reordered words, added or removed function words, synonyms, paraphrases and searches with the same intent, and there is no way to opt out. Twenty-five exact match keywords written to capture every phrasing of one idea will mostly be duplicates.

Are single keyword ad groups still worth using?

For most accounts, no. Single keyword ad groups (SKAGs) depended on exact match meaning exact, and that behaviour is gone. The approach made sense a decade ago. It stopped making sense once Google’s matching became semantic.

The timeline explains why older advice now fails:

• 2014: Exact match starts including plurals, misspellings and similar close variants.
• 2017: Exact match expands to reordered words and function words.
• 2018: Exact match starts matching queries with the same meaning, including implied words and paraphrases, as Search Engine Land reported at the time.
• 2021: Broad match modifier is folded into phrase match and disappears as a separate option.

An ad group built around one exact match keyword will match a wide range of queries anyway. You carry the admin overhead of hundreds of ad groups and get none of the control they were meant to provide.

Intent grouping replaces SKAGs. Build one ad group for each thing the customer is trying to do. “Emergency plumber Melbourne”, “24 hour plumber Melbourne” and “after hours plumber near me” express one intent, so they belong in one ad group with one set of ads and one landing page.

Keyword counts by budget and business type

Monthly spend sets a practical ceiling on how many keywords can generate useful data. Work backwards from the budget rather than forwards from a keyword list. Notice that the per ad group range stays the same in every row; accounts grow by adding ad groups, not by inflating them.

Account type and monthly spend	Keywords per ad group	Ad groups	Total keywords	Notes
Local business, AUD 1,000 to 3,000	8 to 20	2 to 5, in one or two campaigns	20 to 60	About 330 to 1,000 clicks a month at AUD 3 a click. A keyword cull usually beats an expansion. Let phrase match handle variations.
Mid-size account, AUD 5,000 to 20,000	8 to 20	10 to 40	150 to 600	Enough traffic to segment by service line, location and funnel stage. Negative keyword work starts to matter more than adding keywords.
Large ecommerce or lead generation	8 to 20	Many, by product category, brand and margin tier	1,000 to 10,000	Shopping and Performance Max handle much of the volume without keywords, so the Search list often shrinks over time.

Trades, allied health, local legal and hospitality accounts in the first band tend to improve after a cull, not an expansion. Focus on terms with clear commercial intent. In the middle band, you will gain more from tightening what you exclude than from broadening what you include, which is why most pay per click management work at that level is subtraction rather than addition.

How to set your keyword count

Keyword count sits downstream of two decisions that matter more: how tightly your ad groups are themed, and how much conversion data the account generates. Get those right and the count settles itself.

Start with conversion volume

Count last month’s conversions and divide by 3. That figure is roughly the maximum number of keywords you can meaningfully optimise right now. Forty conversions a month supports around 13 keywords worth your attention, not 300. For a brand new campaign with no history, launch with 10 to 20 keywords per ad group and expand later from the search terms report, not from a keyword tool.

Group by search intent, not by keyword string

Put keywords together when the person typing them wants the same outcome, even if the words differ. “Buy office chairs online” and “ergonomic desk chair delivery” share an intent. “Office chair repair” does not, however similar the words look.

Cap every ad group at 20 keywords

Set a hard limit and treat breaching it as a signal to split the ad group, not to raise the limit. An ad group that creeps past 20 usually contains two intents that deserve separate ads and separate landing pages.

Prune every 30 days using the search terms report

Open the search terms report monthly. Add the converting terms you were not targeting, and pause keywords with meaningful spend and no conversions. Keywords averaging fewer than 20 to 30 searches a month rarely justify their place.

Review the count quarterly as the account matures

A healthy keyword list stabilises, then shrinks. If an account has been running for two years and still gains keywords every month, it is chasing volume rather than profit.

How many negative keywords should you use?

More than your active keywords, often two to three times as many. Negatives are the cheapest performance lever in Google Ads, because each one redirects budget away from a search that was never going to convert. Build the list before you expand, not after.

The 10,000 negatives per campaign and 5,000 per shared list that Google allows are ample. Start with a shared list at account level for universal exclusions such as “free”, “jobs”, “salary”, “course” and “DIY”, then add campaign-level negatives for terms specific to each service. Cover job seekers, DIY researchers, free-seekers, competitor brand names you cannot service and locations you do not operate in. In Australia that also means excluding overseas city names that share spellings with local suburbs, a leak that goes unnoticed in many accounts.

Negatives matter more the more broad match you run. A broad match account without a serious negative list is not a keyword strategy. It is a donation.

If your account runs hundreds of keywords on a modest Australian budget, the fastest improvement available is to remove the keywords that have spent money without ever producing a conversion and give that budget back to the ones that have. A new bid strategy can wait.

If you would rather someone else did that pass, our Google Ads management work for Australian businesses starts with exactly this: a structure audit, a keyword cull, and a negative list built properly. We do it for clients in Sydney, Melbourne and Brisbane, and everywhere in between.

Tell us what your account is spending and we will tell you how many keywords it should be running.

Frequently asked questions

How many keywords do I need for Performance Max? None, because Performance Max does not use keywords. Google lets you add up to 50 search themes per asset group as optional signals, and exclusions and negative keywords still apply to them.

Does keyword count affect Quality Score? Not directly, because Quality Score is calculated for each keyword on its own. It hurts indirectly, because a bloated ad group weakens ad relevance, one of the three components Google scores.

How long before you know whether a keyword is working? Give it roughly 100 clicks or 30 days, whichever comes first. Below that, the numbers are noise and any decision made from them is a guess.