Negative Keyword Lists by Industry: Where to Start

Search "negative keyword list" and you get files of 500 terms. They all contain "porn", "free", "jobs", "craigslist", and roughly 400 words that will never appear in your search terms report. They are mostly padding. And the padding is not harmless — somewhere in a 500-term list dropped into a dental practice account is "cheap", which kills "cheap teeth whitening near me", a search a cosmetic dentist would happily pay $4 for.

A useful list is short, specific, and written for one kind of business. Here is what that looks like for seven verticals, plus the terms in each that look obviously blockable and are not.

Why generic 500-term lists fit nobody

A list that applies to every advertiser can only contain terms that are irrelevant to every advertiser. That is a very small set — adult terms, a few scraper sites, obvious nonsense. Everything beyond it is a bet about your business, and the list author never saw your business.

The bets go wrong in both directions. "Free" is on every list. It is correct for a law firm that bills hourly and wrong for a SaaS company with a free tier, where "free CRM" is often the cheapest path to a paid account. "Used" is on every list. It is correct for a manufacturer selling new equipment and wrong for a dealer whose whole margin is in used stock. "Jobs" is nearly always right, and it is still worth checking: a recruitment agency lives on it.

The second problem is match type. Most published lists do not say which match type to use, and people paste them in as broad. A broad negative blocks any search containing all the words in any order, which sounds precise until you remember it ignores close variants on the positive side but not in the way you expect. Negative match types don't behave like positive ones, and a broad negative "free trial" blocks "trial lawyer free consultation" for a personal injury firm. Phrase negatives are the safe default for almost everything on these lists.

The third problem is that a static list never learns. Your search terms report changes every month because Google's matching changes every month. A list you pasted in January is answering questions from January. The terms that cost you money in March were not in anyone's file.

So: use the lists below as a floor. Then read your own data.

The four intents that waste money in every industry

Before anything vertical-specific, four buckets of intent account for most of the obviously bad traffic in the accounts we audit. They are worth building first because they are nearly universal and nearly safe.

Job seekers. "salary", "jobs", "hiring", "careers", "vacancy", "apprenticeship", "wages", "pay rate", "recruitment", "internship", "cv", "resume". People researching a career in your industry search the same words as people buying from it. A commercial cleaning company pays for "commercial cleaner salary" all day without noticing. This is the single highest-volume waste bucket in service businesses, and it is covered in depth in blocking job-seeker traffic.

Learners. "how to", "what is", "meaning", "definition", "tutorial", "course", "training", "certification", "exam", "pdf", "example", "guide", "wikipedia", "reddit". Careful here: "training" is the product for a training provider, and "course" is the product for an education business. For everyone else it is a student.

Free and cheap substitutes. "free", "freeware", "no cost", "open source", "crack", "torrent", "diy", "homemade", "do it yourself", "template", "download". The "free" judgement call depends on your funnel. The DIY ones are safer: someone searching "how to unblock a drain yourself" is not calling a drain company today.

Self-service and support. "login", "sign in", "customer service", "phone number", "complaint", "cancel", "refund", "returns", "track order", "my account". These are existing customers or other companies' customers. They cost you clicks and convert at nearly nothing. Exception: if your support site is the conversion path for renewals, check first.

Four buckets, maybe 50 phrase negatives, and in most accounts they remove more waste than the 450 terms that follow them in a generic file. If you want to know what the removal is actually worth before you do it, calculate wasted spend honestly rather than counting every non-converting click as waste.

Starter lists for seven verticals

Each list below is the industry-specific layer — the terms you add after the four universal buckets. Use phrase match unless noted. Each one ends with the trap: the term that belongs on a generic list and should not be on yours.

Home services (plumbing, HVAC, electrical, roofing)

salary, apprentice, license, licence requirements, how to fix, how to replace, parts, diagram, manual, schematic, wiring diagram, error code, reset, troubleshooting, diy, youtube, home depot, lowes, screwfix, rental, course, certification, union, insurance claim, permit.

Home services bleeds on two things: people fixing it themselves and people becoming tradespeople. "Furnace error code 33" is a homeowner with a phone and a screwdriver. "Plumbing apprenticeship" is someone who will never be a customer. Add brand names of the equipment you service only if you do not service them — a Carrier dealer should not block "carrier".

The trap: cost, price, how much. These appear on home service negative lists constantly and they are terrible negatives. "How much does a new boiler cost" is a buyer with a budget forming. The broader pattern of what local accounts actually waste money on is in local business wasted spend.

E-commerce and retail

review, reviews, vs, versus, alternative, coupon, promo code, discount code, wholesale, bulk, supplier, manufacturer, dropship, amazon, ebay, etsy, temu, aliexpress, second hand, used, repair, parts, size chart, return policy, recall, knockoff, replica, pattern, how to make.

Shopping and PMax pull in a lot of near-miss product language. The marketplace names matter most — "nike air max amazon" is a click you lose at the landing page. Block the marketplaces you do not sell on.

The trap: review and vs. For considered purchases over roughly $200, comparison searches convert. A mattress retailer blocking "vs" loses the entire middle of the funnel. For a $12 phone case, block them. The rule is: block comparison terms when your product is cheap enough that nobody researches it. More patterns specific to product catalogues are in e-commerce search terms that drain budget.

Legal services

free, pro bono, legal aid, salary, how to sue, can i sue, statute, law school, lsat, paralegal, definition, meaning, act 1996, form, template, represent myself, small claims, court date, jail, disbarred, complaint against, report a lawyer, citation, case law, reddit.

Legal has some of the most expensive clicks in Google Ads — $40 and up in injury and immigration is normal — so a single bad term category is real money. "How to write a will template" at $35 a click is a person who has decided not to hire you.

The trap: near me variations and cheap. Some legal lists block price-sensitive language to protect positioning. If you take volume work — traffic tickets, simple conveyancing, uncontested divorce — "cheap divorce lawyer" is your customer.

SaaS and B2B software

free, open source, crack, nulled, github, api docs, documentation, login, sign in, status, down, outage, salary, jobs, tutorial, certification, training, consultant, agency, glassdoor, funding, valuation, acquisition, stock, investor, student discount, for students, resume.

SaaS attracts researchers, job applicants and journalists. "Hubspot funding round" is not a lead. Competitor brand terms deserve their own decision rather than a blanket block — see competitor searches, block or keep.

The trap: free. If you have a freemium tier or a free trial that converts, "free" is your highest-intent modifier. Check the conversion data on "free" terms before blocking the word. This is the single most common self-inflicted wound in SaaS accounts.

Healthcare and clinics

symptoms, causes, nhs, medicare, medicaid, insurance coverage, covered by, icd 10, cpt code, salary, nursing school, medical school, jobs, volunteer, charity, free clinic, home remedy, natural, side effects, recall, lawsuit, malpractice, pictures, images, wikipedia.

Symptom searches are enormous in volume and almost never a booking. "Signs of a cavity" is reading. "Emergency dentist open now" is a patient.

The trap: insurance. If you take insurance and that is a differentiator, "dentist that takes Delta Dental" is a high-intent local search. Only block insurance terms if you are cash-pay only.

Automotive (dealers, repair, rental)

salary, jobs, mechanic course, scrap, scrap value, junk, salvage, write off, for sale by owner, craigslist, facebook marketplace, how to fix, torque spec, fuse box, obd, error code, recall, tax, mot history, vin check, insurance group, 0-60, top speed, wallpaper, games, toy, model car, lego.

Automotive attracts enthusiasts in volume. "BMW M3 top speed" and "Lego Porsche" are not shoppers. Toy and model terms are the easiest win most car accounts never take.

The trap: used, finance, lease, bad credit. Each of these is the core business for somebody. A franchise dealer with a used lot should never block "used". Decide per-term, not per-list.

Education and training providers

free, free course, pdf, answers, answer key, cheat, past paper, jobs, salary, teacher, tutor jobs, scholarship, grant, funding, student loan forgiveness, is it worth it, reddit, review, scam, refund, dropout, coursera, udemy, khan academy, youtube.

Education providers pay for an enormous amount of "is X worth it" and "free X course" traffic. Competitor platform names are worth blocking unless you run a deliberate conquest campaign.

The trap: online, part time, evening, accredited. These read like qualifiers to filter out and they are the exact words a serious adult learner uses.

Shopping and Performance Max change the rules

Negative keywords work differently outside standard Search, and an industry list pasted into the wrong campaign type either does nothing or does less than you think.

Standard Shopping takes negative keywords and shared negative lists exactly like Search, so an industry list applies cleanly. What is different is that Shopping has no keywords on the positive side — your product titles and feed do the matching. That means negatives are your only query control, and a list built for your Search campaigns is usually too narrow for Shopping, where Google will match on any word in your product data. Shopping negatives work differently in ways worth reading before you copy a list across.

Performance Max is the awkward one. It accepts up to 10,000 campaign-level negative keywords, and it does not accept shared negative keyword lists at all. So the industry list you carefully built as a shared list cannot be attached to PMax. You have to push the terms in at campaign level, campaign by campaign. With ten PMax campaigns across an account, that is ten copies of the same list to maintain by hand — which is the practical reason most PMax campaigns have almost no negatives in them. The mechanics and what the cap really means are in negative keywords in PMax and the 10,000 cap.

There is a reporting limit too. Search terms for Search, Shopping and Performance Max all appear in campaign_search_term_view, but PMax reporting is thinner and groups terms into categories you cannot always act on individually. What the PMax search terms report actually shows is less than the Search equivalent, which means industry lists matter more in PMax — they are preventive, because the diagnostic data is incomplete. Display, Video and App campaigns report no search terms at all and take no negative keywords, so nothing in this article applies to them.

Also, search themes are not negatives. Adding a search theme does not exclude anything, and the two systems interact in ways people get wrong — see search themes vs negative keywords.

How to add an industry list without breaking anything

Add the list as phrase match, at account or shared-list level for Search and Shopping, and copy it into each PMax campaign separately. Then check the damage within a week.

The order that works:

1. Export 90 days of search terms first. Before you add anything, pull the report. Run your candidate list against it in a spreadsheet and count how many terms each negative would have blocked, and how much those terms spent and converted. Any negative that would have blocked a converting term comes off the list immediately. This step takes 20 minutes and it is the difference between a list and a liability.
2. Start with the four universal buckets. They are the safest and they usually remove the most.
3. Add the vertical layer, minus your traps. Read the trap paragraph for your industry and delete those terms from the file before you paste.
4. Use phrase, not broad. Broad negatives behave unpredictably across match types and are the main cause of "we added negatives and impressions fell off a cliff".
5. Watch impression volume for seven days. A sharp drop that is not matched by a drop in waste means you blocked something you wanted. Six signs your negatives are hurting performance lists what to look for.
6. Keep the undo path open. Know how to reverse a block before you make 200 of them. How to undo negative keywords covers the manual route; PPCleaner keeps every block reversible in one click for 90 days.

One more thing about size. A list of 500 terms is not five times better than a list of 100. Most accounts never trigger most of their negatives, and a bloated list makes conflicts harder to find when something does break. How many negative keywords are too many goes into where the real limits are.

Where industry lists run out

An industry list handles the predictable waste. The waste that actually eats your budget is unpredictable, and it is specific to your products, your brand name and whatever Google decided to match you to last month.

Here is the shape of it. A list author can guess that a locksmith does not want "locksmith salary". No list author can guess that your brand name is one letter from a video game, that your product name is also a type of fish, or that your broad match keyword started pulling "second hand" variants in week three for no visible reason. Those terms do not appear in any published file because they only exist in your account. And they are typically where the money goes — not in the obvious buckets, which are usually a modest slice, but in the long tail of near-miss terms that each spend $8 and never repeat.

That is a reading problem, not a list problem. Which means the real job is frequency: how often you check search terms determines how much a bad match costs before you catch it. Checking monthly on a $10,000-a-month account means a term can burn for 30 days. Checking nightly means it burns for one.

There is also the part Google hides. The search terms report does not show every query — low-volume and privacy-filtered terms are aggregated away, so some spend has no visible term attached to it at all. What the search terms report hides matters when you are trying to reconcile campaign spend against the terms you can see. You cannot negative what you cannot see, and no industry list changes that.

PPCleaner exists for the part after the list. It reads your search terms every night across Search, Shopping and Performance Max, judges which ones were never going to convert, and adds them as negatives — including pushing them into each PMax campaign individually, since shared lists are not an option there. It never touches budgets, bids, ads or keywords, and it never blocks a term that converted in the last 90 days. The industry list is still worth building on day one. It is just the floor, not the ceiling.

If you are starting from a cold account and do not know where the waste is yet, find wasted spend first and build the list from what you find. A list derived from your own report beats any list derived from your industry.