Negative Keyword Strategy for Contractor Google Ads Accounts
Negative keywords prevent contractors from wasting ad spend on job seekers and DIY searches.
Contractor Google Ads accounts are losing money faster than they did even two years ago, and the reason is in how Google now matches queries to ads. A roofer bidding on "roof repair" can end up paying for a click on "roofing salary." A landscaper running ads for lawn care can end up footing the bill for someone reading a "lawn mower review." These are the predictable result of changes Google has made to how broad match and close variant matching work. Broad match is now the default setting when a contractor sets up a new Search campaign with Smart Bidding, and close variant matching pulls in synonyms Google decides are similar enough to the keyword, so a contractor bidding on one clean service term now shows up in auctions for queries that have nothing to do with hiring a professional. Exact match, which used to mean exactly that, now routinely fires for searches that share "the same meaning or intent" as the keyword even when the words themselves are different. AI Max for Search, which rolled out broadly in the second and third quarters of 2025, adds another layer by generating keyword variations on its own, and the expansion it produces has caught even advertisers who watch their accounts closely off guard. For a contractor paying for clicks on "plumber" in broad match, the same budget also buys clicks on "plumber salary," "how to become a plumber," and "plumber near me free estimate," none of which turn into a job.
Negative keyword match types
Negative keywords follow their own set of rules, and those rules run backward from what most advertisers expect based on how positive keywords behave. Getting the match type wrong is the single most common reason a negative keyword list fails to stop the spend it was built to stop.
Three match types exist for negatives: broad, phrase, and exact, and each one earns its place and carries its own risk. Negative broad match blocks any query that contains all the words in the negative term, in any order, which makes it useful for cutting off an entire theme at once, things like job-seeker traffic, DIY searches, or anyone hunting for something free. Broad match cannot distinguish a bad use of a word from a good one. The word "free" is a textbook example: add it as a broad negative and it will block "free consultation" right alongside "free estimate for siding job," taking out a legitimate lead along with the junk. Negative exact match is at the other end of the spectrum. It blocks only the precise query typed in, nothing more, nothing less. That makes it slow and narrow, but safe. It is the right tool when a contractor needs to cut one specific query without touching anything close to it, though it will do nothing to stop every other variation of that same bad intent from appearing the following week.
One rule changes everything about how these lists need to be maintained: negative keywords do not get close variant matching. A positive keyword will automatically catch misspellings and near-matches because Google fills in those gaps. A negative keyword will not. If a contractor blocks "plumber salary" as an exact negative, a search for "plumbers salary" or "plumber salery" can still slip through and trigger the ad, because the negative only blocks what it was literally told to block. Every variation has to be added by hand. The costliest mistake occurs when a contractor applies broad match to a term that has mixed intent, cutting off an entire theme of traffic when only one particular use of a word was ever the problem.
The three intent categories that determine which queries a contractor should exclude
Before anyone touches a negative keyword list, the question to ask about any search query is never whether the keyword sounds bad. A query should be excluded only if the person typing it was never going to pay for a job. Sorting queries into three tiers of intent makes that call consistent across an entire account.
Tier one covers high-intent, location-aware searches: a city name paired with a service, or words like "emergency" and "same day" attached to the job. These convert at the highest rate of any query type a contractor will see, and they should never be excluded under any circumstance. Tier two covers symptom or problem searches that lack a location, things closer to "water heater not working" than "water heater repair" paired with a city name. These convert at a lower rate than tier one, but they still represent real prospects, people with an actual problem who just haven't narrowed their search yet. The right move here is adjusting bids up or down, not cutting the traffic off with a negative keyword.
Tier three is where exclusions belong, and it breaks into four categories that apply to nearly every contractor account regardless of trade. Employment intent covers searches like jobs, careers, salary, hiring, resume, and apprenticeship, and a search for "roofing salary" costs exactly as much to show an ad against as a search for "roof repair near me," while producing nothing in return. DIY and educational intent covers searches like how to, DIY, tutorial, step by step, instructions, and video, since a homeowner looking up a tutorial has already decided to do the work without hiring anyone. Freebie and price-research intent covers words like free, cheap, cost of, and price list when they show up as standalone searches rather than attached to a specific service request: someone who wants an actual quote is not the same as someone fishing for the lowest number on the internet with no plan to hire a contractor. Product and supply purchase intent rounds out the list, and it is specific to each trade: a landscaper paying for a click on "mulch delivery" or "fertilizer purchase" is funding someone's trip to Home Depot, not a landscaping job.
The trade-specific lists make this concrete. A roofing contractor should be watching for and excluding roofing materials, shingles purchase, roofing supply, DIY roofing, how to shingle, roof pitch calculator, shingle brand comparisons, roofing jobs, and roofing salary. A landscaper should be excluding landscaping supplies, mulch delivery only, soil purchase, fertilizer purchase, weed killer product, landscaping jobs, lawn mower review, and landscaping software. A general contractor should be excluding plans and permit when they appear as standalone searches, along with contractor salary and construction jobs. None of these lists will ever be complete, because new variations of the same bad intent keep appearing in search terms, and catching what a static list misses is the job of the search terms report, covered later. When jobs, careers, salary, and hiring all turn up together in an account's search terms, the right move is blocking the whole employment-intent category at once, not adding four separate keywords one at a time.
Structuring negative keyword lists across account, campaign, and ad group levels
A single negative keyword list slapped onto every campaign in an account causes nearly as much damage as having no list. The fix is a tiered structure, where exclusions that apply everywhere live at the account level and exclusions that only make sense in one specific context stay isolated to that campaign or ad group.
Campaign-level negatives do most of the actual work in a well-run account, because they catch terms that are waste in one campaign but a real signal in another. Pricing is the clearest example: a search containing "pricing" might be waste in a broad prospecting campaign aimed at cold traffic, but that same word is a legitimate high-intent signal in a bottom-of-funnel campaign built around people close to hiring. Ad group-level negatives handle the most granular job, directing traffic to the single most relevant ad group inside a campaign by excluding overlapping terms from the ad groups that are a worse fit for that specific query.
The layering works top to bottom. Ad group negatives only govern traffic inside that one ad group. Campaign negatives govern everything inside that campaign. Account negatives govern the entire account at once. Keeping that hierarchy straight is how a contractor avoids both gaps in coverage and accidental overblocking. Applying account-level negatives too liberally in a complex setup carries real risk: a term blocked across brand, competitor, and service campaigns all at once can wipe out valuable traffic in the one campaign where that word actually meant something different.
Google Ads supports shared negative keyword lists, which let a contractor build one list and apply it across multiple campaigns at once. This is the mechanism that lets a contractor running separate campaigns for roofing, gutters, and siding maintain a single employment-intent exclusion list, covering jobs, careers, salary, and hiring, without typing the same entries into three different campaigns and keeping them all updated by hand.
Performance Max campaigns and the 2025 negative keyword limit expansion
Performance Max campaigns need to be handled separately from standard Search campaigns, because the way PMax matches queries and the limited control advertisers have historically had over it make it a distinct source of wasted spend. The changes made through 2025 improved that picture considerably, even if they did not fix it fully.
Before December 2024, adding a campaign-level negative keyword to a PMax campaign meant contacting Google support directly or filling out a specifically formatted request form, just to get one term blocked. That friction meant a large share of PMax campaigns ran with no campaign-level exclusions at all, simply because the process to add them was too slow to bother with. In March 2025, Google raised the maximum number of PMax negative keywords allowed per campaign by a factor of one hundred, moving it from a ceiling so low it barely mattered to a limit that actually supports a real exclusion list. That same month, Google also released the PMax Search Terms Report, giving advertisers visibility into what PMax was actually matching against, visibility that simply did not exist before. In August 2025, Google added shared negative keyword list support for PMax. A contractor can now take the same structured exclusion list built for search campaigns and apply it directly to a PMax campaign.
Even with those improvements, PMax still carries real limitations for a local home service business. Traffic inside PMax spreads across search, Shopping, display, YouTube, Gmail, Discover, and Maps. The high-urgency, high-intent queries that actually drive contractor jobs get diluted across channels where intent runs much lower. Advertisers also lose the keyword-level control that makes a standard search campaign effective for catching a high-intent local query and bidding on it specifically. Dedicated search campaigns, built and structured correctly, consistently beat PMax on the exact query types that matter most to a home service contractor, the emergency and same-day searches that convert at the highest rate. PMax still has a place in a contractor's account. It means that a contractor running PMax should apply the same universal exclusions used in search campaigns, take full advantage of the much higher campaign-level negative keyword cap now available, and use the new Search Terms Report regularly to check what queries PMax is actually spending money against.
The search terms report as a weekly operational discipline, not a one-time setup task
A negative keyword list built once at campaign launch starts going stale within weeks, because new irrelevant queries keep appearing as Google's matching behavior keeps shifting. The search terms report is the only place a contractor can see, in plain terms, what people actually typed before an ad showed up, and checking it on a fixed schedule is what turns a list from a one-time setup task into a system that keeps working.
The review should run on a set weekly cadence, treated as a standing task on the calendar rather than something triggered only after a contractor notices a spike in spend. A single irrelevant query left running for a week at cost can burn through a meaningful chunk of budget before anyone catches it, so waiting for a problem to become obvious costs real money. The weekly process itself is straightforward. Pull the search terms report and filter it down to queries that have clicks but zero or near-zero conversions. Look across those queries for a pattern rather than judging each one in isolation: if jobs, careers, salary, and hiring all appear together, that is a signal to block the whole employment-intent category, not a reason to add four separate keywords one by one. Add the irrelevant queries as negatives at whichever level makes sense, the account level for exclusions that apply everywhere, the campaign level for exclusions that only matter in one specific context. Finally, look for the queries doing well: terms that are already converting should get promoted to exact match as positive keywords, which protects that traffic and helps improve Quality Score at the same time. Run that way, every week, a negative keyword list becomes the budget control system it was always meant to be, not just a document.
