How to Find Negative Keywords Before You Launch

Before launch, build your negative list from three sources that know your business: your own website (every "we don't" line is a candidate), your competitors' ads (any offer they sell that you don't is traffic you'll pay for), and Google autocomplete around your core keywords (it shows the learners, DIYers and job seekers). Only block terms no buyer could mean. Put the doubtful ones on a watch list.
Then plan the first review before you press launch. Read every search term on the first day with spend, look for patterns on day three, and settle the watch list on day seven. A pre-launch list is a prediction. The search terms report is where you find out how wrong it was.
A new campaign has no search terms report. Anything you block before launch is a prediction about queries nobody has typed yet. The best predictions come from sources that know your business: your own website, your competitors' ads, and what Google autocompletes around your keywords. A list downloaded from a blog doesn't know your business. Then you plan the first review, because the list will be wrong in places.
This article follows one example all the way through: a contractor that installs Level 2 home EV chargers around Austin, Texas. It has a $50 daily budget and expects to pay about $5 a click.
Why generic negative keyword lists fall short before launch
Generic lists block words that are junk for the average advertiser, not for you. They miss the queries tied to your specific offer, and they sometimes block queries you want.
Take "free," which is on almost every downloaded list. Our EV installer offers a free quote. Someone who types "free ev charger installation quote austin" is a buyer, and a broad negative on "free" throws them away. "Cheap" is on most lists too. But "cheap ev charger installation austin" is someone who wants the job done and cares about price. That may be exactly who a small contractor beats a big one on.
Meanwhile, the generic list misses the queries that will actually cost this advertiser money:
- "ev charging station near me": drivers who need a plug right now
- "tesla supercharger austin": the same, for one brand
- "chargepoint app not working": people using the public network, not installing anything
- "portable ev charger": shoppers buying a cable, not hiring an electrician
- "level 3 charger cost": mostly businesses or curious drivers. A home can't really take one.
No list of 500 common negatives contains these, because they only matter if you install home chargers.
Most guides on this topic say the same three things: brainstorm, download a list, then check the search terms report after launch. Step two is the risky one. Step three needs data you don't have yet. What's missing is a method for the gap between "nothing has run" and "enough has run to judge." The rest of this article fills that gap. If your business sells a service, negative keywords for service businesses by intent covers the intent patterns that come up after launch.
Start with your own site: every "we don't" is a negative keyword
Your website already says what you sell, where, and to whom. Anything outside those lines is a candidate negative. Read the site the way a stranger would, and note every limit you've put in writing.
Go page by page. For the EV installer, this is what turns up.
Service pages. "We install Level 2 chargers in homes." That rules out a family of queries straight away: "commercial," "fleet," "workplace charging," "level 3," "dc fast charger," "charging station business." These people want something the company doesn't do.
Service area page. Austin, Round Rock, Cedar Park, Georgetown, Pflugerville. Location targeting keeps out most searchers in Houston. But someone in Austin who types "ev charger installer houston" for a relative's house still matches. So do people outside the area if you target by interest as well as location. Other big Texas cities go on the list. Wrong-location clicks show up in the search terms long before they show up in the location report, so this is worth getting right early.
FAQ page. "We don't sell chargers on their own, only with installation." That points at "for sale," "buy," "amazon," "home depot." It also flags a harder case: "best ev charger for tesla." That could be a shopper comparing hardware, or a homeowner about to book an installer. That one goes on the watch list, not the block list.
Careers page. If you're hiring, your own site proves Google will connect you with job seekers. "Jobs," "careers," "apprentice," "salary," "hiring," "electrician license." This is generic advice, but your careers page tells you it's urgent for you, not just possible.
Brand pages. The installer fits ChargePoint, Emporia and Tesla Wall Connector units. Here the site saves you from a mistake. "ChargePoint" is both a charger brand you install and a public charging network. Blocking "chargepoint" would cut out "chargepoint home flex installation," which is a buyer. Block the network phrases instead: "chargepoint app," "chargepoint card," "chargepoint near me," "chargepoint account."
Ask Google what it thinks your site is about
Keyword Planner can start from a website instead of a keyword. Enter your own URL and look at the ideas it returns. It's a rough preview of what Google associates with your pages, and that matters more than it used to. Broad match, Performance Max and AI Max all lean on your site's content to decide what you're relevant for.
If Keyword Planner suggests "ev charging station locations" for an installer's site, it's telling you that link exists in Google's view. You'll pay for it unless you exclude it. Don't treat this as proof of what will trigger your ads. Treat it as a list of neighbours worth checking.
Read competitors' ads for the buyers you don't want
Competitor ads show which nearby offers share your keywords. Every offer they advertise that you don't sell is a group of queries you'll pay for unless you exclude it.
You don't need to click anyone's ads to do this. The Google Ads Transparency Center lets you search an advertiser by name, filter by region and read their text ads. The Ad Preview and Diagnosis tool in Google Ads shows the results page for a query and location without creating an impression. Run your five or six core keywords through it and write down every headline.
For the EV installer, three competitors tell three different stories.
Competitor A runs "Solar + EV Charger Bundles." The installer doesn't do solar. Queries like "solar and ev charger installer" and "solar panel ev charging" will match its keywords, so "solar" goes on the list as a phrase negative.
Competitor B runs "Panel Upgrades From $1,500." This is where competitor research stops you blocking too much. Our installer does upgrade electrical panels, because many homes need one before a Level 2 charger. So "panel upgrade" is a keep. It may even deserve its own ad group. The competitor's ad just reminded the advertiser of a service it was about to treat as off-topic.
Competitor C runs "Commercial EV Charging, Tax Credit Help for Businesses." "Commercial" and "business" are already blocked from the site review. "Tax credit" is harder, because homeowners ask about federal credits for home chargers too. Watch list.
Competitor brand names are a decision, not a default
With broad match, your ads can show for competitor names. Someone searching "[competitor] reviews" or "[competitor] phone number" usually wants that company, and you'll pay for clicks that rarely convert. If you aren't deliberately bidding on competitors, add their names as phrase negatives.
There's a real cost. Someone typing "[competitor] vs" or "[competitor] alternatives" is shopping around, and you'd lose them. If that matters to you, block the navigation phrases ("[competitor] login," "[competitor] phone number," "[competitor] careers") and leave the bare name alone.
Competitor pages tell you what not to block
Read competitors' landing pages for their words, not their offers. EV installers write "EVSE," "NEMA 14-50," "240V outlet," "hardwired." These are words your buyers use. Note them as keeps, so a phrase negative like "outlet" (which you might add to stop "outlet mall" searches) doesn't quietly block "240v outlet for ev charger."
Mine autocomplete for the junk around your keywords
Autocomplete shows what people really type after your keywords. Run each core keyword through it with a letter after it, and the patterns of people who aren't buyers show up in a few minutes.
Type "ev charger installation" into Google, add a space, then go through the alphabet: "ev charger installation a," "ev charger installation b," and so on. Then try question prefixes: "how to," "can I," "do I need," "why." Suggestions change by location and over time, so do this yourself rather than trusting someone else's screenshot. For a phrase like this, expect groups along these lines:
- Learners: "course," "training," "certification," "class." Electricians trying to get into the trade.
- DIYers: "diy," "how to install ev charger yourself," "wiring diagram." Some will give up and call someone. Most won't.
- Code researchers: "nec requirements," "code," "breaker size." Mostly electricians. Sometimes a careful homeowner. Watch list.
- Buyers: "cost," "near me," "rebate," "permit," "quote." Keep all of these.
- Mixed: "in apartment," "in garage," "outdoor." Keep these unless your FAQ rules them out.
Scroll down to "People also ask" and "Related searches" on the same pages. They show the questions sitting next to your keyword, and they bring up adjacent topics autocomplete misses. For example, "how much does it cost to charge an ev at home" comes from a car owner working out electricity costs, not someone ready to hire an electrician.
Keyword Planner's "Refine keywords" panel helps here too. It groups ideas by theme, often including brands. You can scan a group like "charging network" and see a dozen related phrases at once.
Why autocomplete matters more with broad match
Your keywords won't only match queries that contain their words. Close variants already stretch exact match (see why exact isn't exact). Broad match and AI Max stretch much further. Autocomplete is a fair map of where that stretching will go, because it shows what people type near your words. If you're turning on AI Max, the controls that actually work on its search terms are worth reading before launch, not after.
Sort into three buckets
Everything you've gathered from the site, competitors and autocomplete goes into one of three buckets:
- Block now. No buyer could mean it. "Ev charger installer jobs." "Charging station near me." "Chargepoint app."
- Watch. A buyer could mean it, but probably doesn't. "Best ev charger for tesla." "Ev charger tax credit." "Nec requirements ev charger."
- Keep. Buyer words, including ones you nearly blocked. "Panel upgrade." "240v outlet." "Free quote."
One test settles most cases. If you can write a single realistic sentence in which a paying customer types the query, it doesn't belong in "block now."
Choose match types and placement before you paste
Use phrase match for most pre-launch negatives, exact match for narrow ones, and broad match only for single words you're certain of. Then put each list where the campaign type will actually read it.
Phrase match is the safe default. The phrase negative "charging station near me" blocks any query that contains those words in that order, such as "tesla charging station near me" or "free charging station near me now." It won't block "home charging station installer near me," because the word order breaks.
Exact match suits queries that are too vague to be worth paying for but sit right next to good ones. The exact negative [ev charger] blocks that bare two-word query, which is mostly shoppers. It leaves "ev charger installer austin" alone.
Broad match negatives behave unpredictably across match types. Before launch, when you can't see what they're catching, keep them to single words that are never relevant, like "jobs" or "salary."
Negatives don't stretch to close variants the way keywords do. Add plurals and obvious variants yourself: "course" and "courses," "certification" and "certified," "installer job" and "installer jobs." For worked patterns, see phrase match negative keyword examples you can copy safely.
Where each list goes
- Search and Standard Shopping: build one shared negative keyword list and apply it to every campaign. Both accept shared lists, so one edit updates everything.
- Performance Max: it does not accept shared lists. It takes up to 10,000 negative keywords at campaign level, so paste the "block now" list directly into each Performance Max campaign. If negatives there don't seem to hold later, here's why that happens.
- Display and Video: these report no search terms, so a search-based negative list is the wrong tool. You'll manage placements and audiences there instead.
Check the list against your own keywords
Before you launch, compare every negative with every keyword. The "free" negative from a generic list against the keyword "free ev charger installation quote" is the classic case. The keyword will sit there with no impressions. Google won't always warn you, and some overlaps are never flagged. See negative keyword conflicts Google Ads won't flag. A plain spreadsheet check takes ten minutes. Without it, you can lose a week wondering why a good keyword is silent.
Plan the first review before you press launch
Put three search terms reviews in your calendar before launch: the first day with spend, day three and day seven. The pre-launch list will have gaps, and early clicks only stay cheap if you catch the gaps early.
Here's the arithmetic for the example. At $50 a day and about $5 a click, the installer gets roughly 10 clicks a day: 70 clicks and $350 in the first week. Suppose 20 of those clicks come from "charging station" variants the pre-launch list didn't catch, like "ev charging austin downtown" or "where to charge tesla austin." That's $100, two full days of budget, spent on drivers looking for a plug. Bidding also learns from those clicks. If you're on Smart Bidding, that early junk skews what the algorithm thinks a good click looks like.
These numbers are made up for illustration. The point holds at any budget: a gap in the list costs the most in the first week, when spend is already low and every click carries extra weight.
Day one: read every term
There will be few enough to read them all. Block anything that clearly belongs in the "block now" bucket and slipped through. Mark any watch-list items that appeared. Don't act on them yet, unless one has already cost more than a few clicks.
Day three: look for patterns, not single terms
By now one query is just noise. A word that keeps coming back is a signal. If "station," "supercharger" and "plug" show up across ten different queries, that's one decision, not ten. A routine for this is in how to analyze a search terms report in 20 minutes.
Day seven: settle the watch list and check your own blocks
Give each watch-list term a deadline tied to money, not to a feeling. If your target cost per lead is $150, a term that has spent $150 (30 clicks at $5) without a lead has had a fair trial. Below that, keep watching.
Also check what you blocked that you shouldn't have. A keyword whose impressions dropped right after you added a negative is the first sign. And remember the report isn't complete: low-volume terms and terms hidden for privacy leave part of your spend without a visible query. A clean-looking report in week one doesn't mean a clean account.
Where a pre-launch list stops and PPCleaner starts
No tool can see queries before people type them, and PPCleaner can't either. It reads the search terms your account has paid for, so it can't build your pre-launch list. It starts working on the first night after launch.
From then on, every night, it reads the search terms reported for your Search, Shopping and Performance Max campaigns. It judges which ones were never going to convert and adds them as negative keywords. It doesn't touch budgets, bids, ads or keywords. You can undo any block in one click for 90 days.
One limitation matters most at launch. PPCleaner never blocks a term that converted in the last 90 days. On an established account, that rule protects your proven buyers. On a brand-new account, nothing has converted yet, so there's nothing for the rule to protect. That's why your "keep" bucket matters: it records what you already know about your buyers before the data does. It's also why it's worth checking PPCleaner's blocks during that first week, and undoing any that hit a term you know is good.
It also can't help on Display or Video, because those campaigns report no search terms. For Search, Shopping and Performance Max, the day-three and day-seven reviews above become a check on decisions already made, instead of a pile of terms you haven't read.
Frequently asked
How many negative keywords should I add before launching a campaign?
There's no right number. Fifty negatives taken from your own site, competitors' ads and autocomplete will do more than 500 pasted from a generic list, because they match your offer. Only add terms no buyer could mean. Put anything doubtful on a watch list and decide after launch, using real spend.
Is it safe to use a downloaded negative keyword list?
Use it as a checklist, not something to paste in whole. Go through it line by line and drop anything a customer of yours might type. "Free" is the usual trap, because it blocks "free quote." Then check the remaining negatives against your keywords so none of them silences a keyword you're paying to run.
Can I add negative keywords to Performance Max before launch?
Yes, at campaign level. A Performance Max campaign takes up to 10,000 campaign-level negative keywords, but it doesn't accept shared negative keyword lists. You have to paste your list into each Performance Max campaign. Search and Standard Shopping campaigns can share one list.
Should I block competitor brand names before launch?
If you aren't deliberately bidding on competitors, block the phrases people use to find that company, such as "[competitor] login", "[competitor] phone number" or "[competitor] careers". Blocking the bare brand name also loses people searching "[competitor] alternatives" who are still comparing options, so decide whether those searchers are worth keeping.
How soon after launch should I check the search terms report?
On the first day with spend, then on day three and day seven. On day one, read every term and block the obvious misses. On day three, look for words that keep coming back. On day seven, settle the watch list using a spend threshold tied to your target cost per lead, and check whether any negative is blocking a keyword you want.
Clean your search terms. Save time and money.
Every night, PPCleaner reviews the searches your campaigns actually paid for, groups them by intent, and excludes the ones that never convert. No spend threshold to wait for, nothing touched but negatives — and one click undoes any of it.
Start 7 days freeSearch, Shopping and Performance Max · Cancel in one click
Related articles

Negative Keywords for Lead Generation That Cut Junk Leads
In lead gen, the costliest search terms are often the ones that convert: job seekers, vendors and existing customers filling your quote form. Here is how to trace each junk lead to its search term and block it without losing good leads.

Google Ads Exact Match Negative Keywords: The Safest Block
An exact match negative blocks one query and nothing else, so it can't kill traffic you didn't mean to block. That precision has a price, and you need to know it before you build hundreds of them.

Phrase Match Negative Keyword Examples You Can Copy Safely
Most guides explain phrase match negatives with one tidy example and stop there. Here are real-looking cases where they block buyers by accident, with the search terms before and after and safer versions to use instead.
