Negative Keywords Google Ads: The SMB Guide That Saves Spend

Why Negative Keywords Are the Most Neglected Lever in Small Business Google Ads

If you are running Google Ads as an Australian small business, industry data consistently shows that between 10% and 30% of paid search spend is consumed by queries that have no realistic chance of converting: that is between $1,200 and $54,000 per year wasted at the $1,000-$15,000 monthly spend range. The fix is not a new bidding strategy or a landing page rebuild. It is negative keywords: the exclusion list that tells Google which searches should never trigger your ads. This article explains exactly what to do and where to start. Negative keywords in Google Ads are the single most under-managed element in the average Australian SMB account. Not bidding strategy. Not ad copy. Not landing pages. The simple, unsexy practice of telling Google which searches should never trigger your ads is what separates accounts that compound results from accounts that quietly bleed budget month after month.

Australian paid search is not cheap territory. The market hit $7.2 billion in 2024, up 10.1% year-on-year, representing 44 cents of every dollar spent on internet advertising in the country (IAB Australia / PwC, 2025). When you’re spending between $1,000 and $15,000 a month in that environment, irrelevant clicks are not a minor inefficiency. They are a structural tax on every campaign you run, and negative keywords are how you stop paying it.

No enterprise complexity. No theory. Just the practice.

How Negative Keywords Actually Work

A negative keyword is an exclusion instruction. You are telling Google: when someone’s search query contains this term, do not show my ad. The mechanism layers on top of your positive keyword targeting, narrowing the pool of eligible queries rather than expanding it.

Match Types Explained

Three match types govern how broadly each exclusion applies.

Negative broad match blocks any search containing all the excluded terms, in any order, regardless of surrounding words. Add “-free” as a negative broad match and you block “free plumber Sydney”, “plumber free quote comparison”, and dozens of other variants simultaneously. This is the right tool for blocking entire intent categories, not individual words.

Negative phrase match blocks searches that contain the exact phrase in the same word order. It is more precise than negative broad match and useful when a word is benign in one context but problematic in another. A trades business might want to show for “apprentice plumber courses” if they run a training arm, but not if they only offer residential call-outs.

Negative exact match blocks only the verbatim query, character for character. It is the scalpel of the three, appropriate for one-off anomalies rather than pattern blocking.

One practical update worth knowing: since 2024, Google’s AI automatically covers misspelling variants for negative keywords (Google Ads Help, 2025). You no longer need to manually add “plumer”, “plummer”, and “plumbing” as separate negatives when you add “plumber” – the system handles close spelling variants itself. It does not, however, handle semantic variants. “Cheap” and “affordable” are semantically related but distinct exclusions you must add separately.

Negatives apply at three structural levels: account-wide, campaign-level, and ad group-level. The most efficient architecture for a small business account is a layered one. Universal intent exclusions sit account-wide. These are terms that are never relevant regardless of campaign: “free”, “DIY”, “jobs”, “careers”, “how to”, “tutorial”, “reddit”. Industry-specific exclusions belong at campaign level. Fine-grained query anomalies get handled at the ad group level. Shared negative keyword lists are the mechanism that makes this manageable – one list, applied across multiple campaigns, updated in a single place.

The Broad-Match-Without-Negatives Problem Is Getting Worse

Broad match is Google’s default match type. It is also the match type most likely to spend your budget on queries that have nothing to do with what you sell. Without a negative keyword list, broad match treats semantic adjacency as a targeting signal. A plumber running broad match on “blocked drain Sydney” can find their ads appearing on “how to unblock a drain yourself” – a query from someone who explicitly does not want to hire anyone.

In our experience reviewing accounts, this is the most common pattern we encounter.

Three forms of leakage result from unmanaged broad match. Budget leakage: irrelevant clicks consuming spend without any possibility of conversion. Attribution leakage: low-intent traffic muddying your conversion data, making it harder for Smart Bidding to optimise toward actual customers. And positioning leakage: ads appearing on queries that put your brand next to searches you do not want to be associated with.

The critical shift in 2025 is that even exact match keywords are no longer safe from this dynamic. Google’s AI-driven “close variants” expansion means that exact match keywords now routinely trigger queries that are not exact matches in any meaningful sense. A keyword like [emergency plumber Brisbane] can now trigger searches that Google’s system interprets as having the same intent, even when the specific words differ. This makes negative keywords necessary across every match type, not just broad match.

If you are adding negative keywords for more than 10% of the search terms appearing in your report, that is a signal your keyword targeting needs reassessment, not more exclusions (Search Engine Land). Negatives fix waste at the edges. They are not a substitute for targeting strategy.

The Performance Max situation deserves specific attention. Until 2025, PMax campaigns supported only 100 negative keywords per campaign – a limitation so severe it rendered systematic exclusion management near-impossible. Google expanded this to 10,000 negative keywords per campaign in 2025 (Google Ads Help, 2025). For small businesses running PMax alongside traditional Search campaigns, this changes the practical calculus entirely. PMax exclusion management is now viable. If you have been running PMax without negatives because the old limit made it pointless, 2026 is the year to revisit that.

Mining the Search Terms Report: A Practical Workflow

The Search Terms Report is the most valuable diagnostic tool in a Google Ads account. It shows you the actual queries – verbatim, character by character – that triggered your ads and generated clicks and spend. Not the keywords you intended to target. The real searches that cost you money.

Access it under Keywords, then Search Terms, in your Google Ads interface. The default sort order is by impressions, which is wrong for our purpose. Sort by Cost, descending. The queries that have spent the most with the weakest results sit at the top. Those are your first targets.

From there, work through a classification exercise. Every query is one of three things: a confirmed valuable match you want to keep, an irrelevant match you should exclude, or a query that belongs in a separate campaign or ad group because it represents a different product or intent entirely.

For the exclusions, resist the temptation to add each one as an individual negative exact match. One query is a data point. A pattern is actionable. If you see “plumber apprenticeship Sydney”, “how to become a plumber NSW”, and “plumber trade school Brisbane” in the same report, you do not add three exact-match negatives. You add “apprenticeship” and “trade school” as negative broad match terms to your account-wide list, and they block hundreds of future variants automatically.

The Six Intent Clusters for Australian SMBs

Build your exclusions around intent clusters, not individual words. The six most common intent clusters for Australian SMB accounts:

• DIY and how-to intent: “how to”, “DIY”, “tutorial”, “guide”, “can I”, “myself”
• Job-seeking intent: “jobs”, “careers”, “apprenticeship”, “salary”, “vacancy”, “hire”
• Free and low-cost intent: “free”, “no cost”, “cheap” (if you do not compete on price), “template”
• Wrong geography: suburb names, states, or regions outside your service area
• Wrong product or service: adjacent categories you do not offer
• Research intent without purchase intent: “review”, “comparison”, “vs”, “best X in Australia” (context-dependent – sometimes you do want these)

The geography cluster is particularly important for Australian SMBs in service-based industries. A Sydney-based business running national campaigns without location-based negatives will regularly pick up queries from Perth, Darwin, or Hobart from users who searched broadly. The solution is either tight location targeting or geographic negative keywords – often both.

Semantic clustering is the 2025 evolution of this practice. Blocking “cheap” as a negative does not reliably block “affordable” or “budget-friendly” under AI-expanded matching. Each intent requires its own keyword cluster. This is more work upfront but the compounding benefit over time is significant – accounts with well-structured semantic negative clusters tend to see improving match quality as the algorithm learns from cleaner data.

The Weekly Review Cadence

Negative keyword management is not a set-and-forget activity. It is a recurring operational task, and the right cadence depends on where the account is in its lifecycle.

For a new campaign or a campaign after a major change (new keywords, budget increase, new ad groups), review the Search Terms Report every two to three days for the first four weeks. New campaigns in particular generate wide query variation as Google’s algorithm explores the intent landscape around your keywords. Early review prevents that exploration from burning significant budget on irrelevant territory.

For an established campaign running on a stable structure, a weekly review is the baseline. Block out 20 to 30 minutes once a week, sort by cost, classify what you find, and update the relevant lists. It sounds minimal. In our experience, this review is one of the tasks most commonly left undone for extended periods in small business accounts.

Treat each review as a two-question classification exercise. First: is this a shared-list problem or a campaign-specific problem? A term that is irrelevant across every campaign you run belongs in a shared negative list. A term that is fine in one campaign but wrong in another is a campaign-level exclusion. Second: is this a one-off or a pattern? One-offs get handled at the specific level. Patterns go to the shared list.

Setting Up Conversion Tracking First

One prerequisite underpins all of this: conversion tracking must be in place before cost-based exclusion decisions are meaningful. If Google Ads is not recording actual conversions, sorting the Search Terms Report by cost tells you which queries spent money, not which queries failed to convert. For Australian SMBs in the $1,000-$15,000 monthly spend range, the three conversion actions that matter most are form submissions, inbound phone calls, and purchases. Each should be configured as a separate action in Google Ads so you can see which query types are driving which outcomes. A common failure mode looks like this: tracking is set to “All conversions” with no action-level breakdown, so the account shows conversions but cannot distinguish between a $5,000 job inquiry and a brochure download. That polluted signal makes it impossible to identify which search terms deserve exclusion and which deserve increased budget. Get at least one primary conversion action (form submit or call) firing correctly before making any exclusion decisions based on cost data. Without it, you risk excluding queries that were actually driving results and retaining ones that were not.

Once a quarter, audit the shared negative lists themselves. Exclusions that made sense when you added them may no longer apply. A business that has added a new service line, changed its pricing position, or moved into a new geography can inadvertently find that old negatives are blocking valuable traffic. The list should evolve as the business evolves.

What Poor Negative Keyword Management Actually Costs

Consider what the numbers look like at the account level. A $3,000-per-month account losing 10% of its budget to irrelevant clicks is writing off $300 every month, or $3,600 every year, in spend that had no realistic chance of converting. At $10,000 per month, that same 10% leakage is $12,000 annually. These are not worst-case figures; they are typical findings in accounts that have not had a negative keyword review in the past 90 days. The recoverable value is real, and it compounds.

The broader context here matters. Australian online retail reached $37.9 billion in turnover in the year to June 2025, up 4.9% year-on-year (ABS, 2025). Competition for paid search traffic in that environment is intense. CPCs are not falling.

When an SMB account runs with poor negative keyword hygiene, the losses are not just financial. Conversion data degrades. Smart Bidding algorithms, which depend on conversion signals to optimise bids, receive a polluted signal – low-quality clicks that do not convert look like failed targeting attempts, and the algorithm adjusts accordingly, often in the wrong direction. Accounts that clean up their negatives frequently see improvement in automated bidding performance as a downstream effect, because the signal quality improves.

Australian CPCs have risen steadily alongside overall market growth, with average CPCs in competitive service categories (trades, legal, finance, health) now ranging from $8 to $40 per click depending on location and query type. There is no viable alternative to Google Ads for paid search at scale in this market, which makes efficiency inside your existing account the primary lever available to most SMBs.

Every dollar wasted on an irrelevant click is a dollar that cannot go toward a search that might convert. At $1,000 to $15,000 per month in spend, even a 10% reduction in wasted clicks compounds meaningfully over a year.

Building a Negative Keyword Architecture That Lasts

A functional negative keyword architecture for an Australian SMB account has four components.

An account-level negative keyword list for universal exclusions: terms that are never relevant regardless of campaign. Keep this list focused. The account-level cap is 1,000 keywords, and most accounts should use far fewer. Reserve this level for the highest-confidence, broadest exclusions.

Shared campaign lists for industry-specific exclusion clusters. A home services business might have a shared list for “DIY and self-service intent” applied across all service campaigns. A professional services firm might have a “employment and careers intent” list. These lists support up to 5,000 keywords and should be structured by intent category, not alphabetically. For context on sizing: most SMB accounts in the $1,000-$15,000 monthly spend range need 50-200 exclusions in total across their shared lists. If you are approaching 500 or more, the account has likely been over-engineered, a warning sign that individual query anomalies are being added where broader intent-cluster exclusions should be doing the work instead. The 2025 PMax expansion to 10,000 negatives per campaign is the ceiling for that channel specifically; shared Search campaign lists remain governed by the 5,000 limit.

Campaign-level negatives for exclusions specific to one campaign’s targeting. These are the fine-grained exclusions that do not generalise across the account.

A documented review log. This sounds administrative, but it matters. When you or a colleague or a new agency picks up the account in six months, the log explains why certain exclusions exist. Undocumented negatives get removed during “optimisation” reviews, reintroducing waste that took months to identify.

The architecture should be simple enough to maintain on a weekly cadence without specialist intervention. If it takes more than 30 minutes a week to review and update, it has been over-engineered.

Getting Started if You Have Not Done This Before

If you are inheriting an account that has never had systematic negative keyword management, the first step is a historical audit rather than a live review. Pull the Search Terms Report for the last 90 days. Sort by cost. Look at the top 50 to 100 queries by spend. You will almost certainly find patterns that, once you see them, are obvious.

Add the pattern-based exclusions to the shared list first. That single action can stop the most costly recurring waste immediately. Then set up the weekly cadence and let it run.

If you are launching a new campaign for the first time, build a starter negative list before you go live. The six intent clusters described above cover the most common sources of waste. Start there, then refine from the actual data once the campaign generates search term history. To make this concrete: a trades business in Australia (plumber, electrician, carpenter) would typically open with negatives covering “apprenticeship”, “trade school”, “certificate”, “TAFE”, “how to”, “DIY”, “free quote template”, “jobs”, “salary”, and any states or territories outside their service area. A professional services firm (accountant, solicitor, financial planner) would prioritise “free”, “template”, “software”, “course”, “study”, “jobs”, “graduate”, and “DIY”. These vertical-specific starters take 20 minutes to build and can help prevent the most predictable budget losses from the first day a campaign goes live.

The accounts that get this right are not doing anything exotic. They are doing something simple, consistently, over time. That consistency is what turns a mediocre campaign into one that compounds: less wasted spend each month, cleaner conversion data, better automated bidding, and a clearer picture of what is actually working.

If your account is in the $1,000 to $15,000 monthly spend range and you have not audited your Search Terms Report in the last 30 days, that is the right place to start.

Need a Second Opinion on Your Account?

We work with Australian SMBs at exactly this level, and the negative keyword audit is almost always the fastest way to find recoverable value in an existing account. If you want a second set of eyes on what you are working with, we are happy to take a look; reach out via cjco.com.au.