Stop Googling Keyword Research Tips: Here’s the Process That Actually Lowers CPC
You opened Google Ads at 9 a.m. and watched the spend counter climb. By noon, $80 gone. By 5 p.m., $210. You checked the conversions column three times because you thought the tracking was broken. It was not broken. You spent $500 in a week and got zero phone calls, zero form fills, zero purchases. Google Ads keyword research is the reason that happened, and it is also the reason it does not have to happen again.
The guides you read told you to find high-volume keywords with low difficulty and add them to your campaign. They skipped the part where Google’s matching system decides which searches actually trigger your ads, how much each click costs, and why the search terms report hides the queries that waste your budget. Google Ads keyword research in 2026 is not about finding keywords. It is about controlling which searches you pay for and which ones you exclude before they drain your account.
Here is the reframe: keyword research is a cost-control mechanism first and a growth lever second. Every keyword you add is a spending decision. Every negative keyword you add is a saving decision. Most small business owners who burn $500 in a week do so because they built a keyword list without a negative keyword list, ran broad match without Smart Bidding, or accepted Google’s AI recommendations without understanding what changed. The process below fixes all three.
What Google Ads Keyword Research Actually Means in 2026 (It Changed)
Google Ads keyword research in 2026 means something different than it did even 18 months ago. Google launched AI Max for Search campaigns in May 2025, a one-click feature suite that converts your keywords to broad match and adds keywordless targeting to find queries your keywords would miss. Google’s own data says advertisers who activate AI Max see 14% more conversions at a similar cost per acquisition, and that number jumps to 27% for campaigns still using exact and phrase match. The old advice of “exact match only, negative everything else” is now incomplete because exact and phrase match keywords are not eligible to show ads inside AI Overviews. Only broad match and keywordless targeting can trigger ads in those AI-powered placements.
This changes how you research keywords. You are no longer trying to enumerate every possible search query and wrap it in exact match brackets. You are building a keyword structure that gives Google’s AI enough signal to find the right searches and enough negative keywords to block the wrong ones. The keyword list is the steering wheel. The negative keyword list is the brake. You need both.
Broad Match Is Not What It Was in 2020
Broad match used to be a budget incinerator. You would add “plumbing repair” as broad match and your ad would show for “plumbing school,” “plumbing jokes,” and “plumbing salaries.” That is why every guide from 2018 to 2023 told you to avoid it. But broad match in 2026 uses every signal Google has: your landing page content, your other keywords, your ad creative, the user’s device, location, time of day, and search history. With Smart Bidding enabled, Google only enters the auction when it predicts the click will convert at your target cost. Without Smart Bidding, broad match is still dangerous. With it, broad match is how you find searches you would never have thought to bid on.
The catch: Google is making broad match harder to avoid. Starting in September 2026, campaigns using the campaign-level broad match setting will automatically be upgraded to AI Max. If you have been holding out with exact match only, that option is narrowing. The question is not whether to adopt broad match. It is whether your negative keyword list and bidding strategy are strong enough to make broad match profitable.
The Recommendations Page Is Not Your Friend
Google’s Recommendations page shows an Optimization Score from 0 to 100 percent. When you see 60%, it feels like a failing grade. It is not. The Optimization Score measures whether you have reviewed recommendations, not how well your campaign performs. Dismissing a recommendation gives you the same score increase as applying it. You can dismiss every suggestion and sit at 100%.
The recommendations themselves push you toward more spend, not less. “Add new keywords” expands your reach into queries Google thinks are relevant. “Raise budget” suggests spending more. “Use broad match” opens the floodgates. Some of these are useful. Many are not. Turn off auto-apply recommendations immediately. Go to the Recommendations tab, find Auto-Apply Settings, and make sure every box is unchecked. Google should not be adding keywords or changing your bids while you sleep. Every change to your account should go through your review, the same way every post on your WordPress site should go through an approval gate before it goes live.
The Search Terms Report Hides Your Biggest Waste
Here is the problem that makes keyword research harder in 2026 than it was five years ago: Google does not show you every search term that triggered your ads. A search term is only reported when it meets a privacy threshold based on sufficient search volume across all Google searches. An independent analysis of over $20 million in Google Ads spend found that hidden search terms waste up to $0.85 for every $1 spent, with 52% higher CPCs and 44% lower click-through rates on hidden queries compared to visible ones. You are paying for clicks on searches you cannot see and cannot negative-keyword out of.
Google’s Ads Liaison Ginny Marvin has stated the threshold increase was “solely privacy-driven.” Whether you accept that explanation or not, the practical consequence is real: a chunk of your spend goes to queries you cannot identify. This is why keyword research cannot stop at the keyword list. You need negative keywords, account-level exclusions, and brand controls to block the categories of waste you can predict, even when you cannot see the individual queries.
The 4 Data Sources You’re Ignoring (That SEMrush and Ahrefs Can’t Give You)
SEMrush and Ahrefs are keyword discovery tools. They tell you what people search for and how hard it is to rank organically. They cannot tell you what searches actually triggered your ads, what your competitors are bidding on, or what your real customers type when they are ready to buy. Four first-party data sources give you that information, and none of them require a $139 per month subscription.
| Data Source | What It Tells You | Why SEMrush and Ahrefs Cannot Give You This | How to Access It |
|---|---|---|---|
| Google Ads Search Terms Report | The exact queries that triggered your ads, with cost, clicks, and conversions per query | Third-party tools estimate search volume. Only Google knows what actually triggered your ads and what you paid for each click. | Campaigns menu, Insights and reports, Search terms |
| Google Ads Auction Insights | Which competitors bid on your keywords, their impression share, and how often they outrank you | SEMrush shows organic competitors. Auction Insights shows who you are paying to beat in the ad auction. | Select a campaign or keyword, click Auction Insights |
| Google Search Console | The organic queries your site already gets impressions and clicks for, with CTR and average position | This is your own first-party organic data. No third-party tool has your actual impression and click counts. | Google Search Console, Performance, Search results |
| Customer Intake Forms and Call Transcripts | The exact words your paying customers use when they describe their problem, not what keyword tools think they search for | Keyword tools show aggregate search behavior. Your intake forms show the language of people who actually bought from you. | Your CRM, contact form submissions, recorded sales calls |
How to Use Each Source in Your Keyword Research
The Search Terms Report is your most underused cost-control tool. Download it weekly during your first month of advertising. Sort by cost descending. Every query that spent more than $10 without a conversion is a negative keyword candidate. Every query that converted is a keyword to isolate in its own ad group with a dedicated budget. This report shrinks over time as you add negatives, which means your waste shrinks with it.
Auction Insights tells you whether your CPC problem is a keyword problem or a competition problem. If your impression share is 15% and three competitors each hold 20% or more, you are being outbid, not out-researched. No amount of keyword research fixes that. You either raise your bids, improve your ad quality to increase Ad Rank, or find less competitive keywords. If your impression share is 85% and you are still not converting, the problem is your landing page or your offer, not your keywords.
Google Search Console shows you which queries your site already gets organic impressions for. These are free signals that tell you what Google thinks your site is relevant for. If you are getting organic impressions for “emergency drain cleaning” but your ads target “plumbing services,” you are spending ad budget on a broader keyword when a more specific one already has organic traction. For a deeper look at how to turn organic query data into a keyword strategy, our guide on AI-assisted keyword research techniques covers how to cross-reference Search Console data with keyword difficulty scoring.
Customer intake forms are the data source that no tool will ever replicate. When a customer fills out your contact form, they describe their problem in their own words. “My water heater is leaking from the bottom” is a keyword. “Water heater making popping noises” is a keyword. These are long-tail queries with clear buying intent that keyword tools often show as zero volume. We covered this exact gap in our analysis of what keyword volume numbers actually mean: long-tail terms frequently show zero volume in tools but generate real impressions and clicks once you target them. Your customers’ words are your best keyword research data, and they are sitting in your inbox.
Build Your First Negative Keyword List Before You Spend a Dollar
Your negative keyword list is more important than your keyword list. Every keyword you add is a decision to spend money. Every negative keyword you add is a decision to save money. If you build the keyword list first and the negative list later, you have already paid for the clicks you should have blocked. Build the negatives first.
Google supports negative keywords at four levels: account, campaign, shared list, and ad group. Account-level negative keywords apply to all campaigns and are the right place for terms no campaign should ever buy. Campaign-level negatives block terms wrong for one campaign but fine for another. Shared lists let you create a themed list once and apply it across campaigns. Ad group negatives route searches between ad groups, like blocking “repair” in a “replacement” ad group so repair searches reach the right destination.
Your Starter Negative Keyword List
Here is a starter list that covers the searches that rarely convert for service businesses. Review every term against your own offer before adding it. “Free” blocks “free estimate” for a roofer. “Used” blocks a refurbished store’s best searches. Add only the terms that genuinely do not apply to your business.
- Jobs and hiring: jobs, careers, hiring, salary, internship, apprenticeship
- Free and DIY: “for free”, freebie, diy, “how to”, tutorial, template
- Education: school, course, certification, training, “classes near me”
- Research and comparison: reviews, “vs”, alternatives, “competitors of”
- Used and secondhand: used, secondhand, “pre owned”, refurbished, craigslist
- Wholesale: wholesale, bulk, distributor, manufacturer
- Brands you do not sell: competitor brand names you do not carry
Add these as account-level negative keywords so they apply to every campaign from day one. Google recommends at least 3 negative keywords per ad group, but a well-structured account should have 50 to 100 negatives at the account level before the first ad runs. This is the cheapest work you will do in Google Ads. Every negative you add before launch saves you the cost of a click you would have paid for otherwise.
One caution: single-word negatives are blunt instruments. Adding “repair” as a negative blocks “repair or replace water heater,” which often ends in a replacement job. Use two or three word phrases for precision: “repair manual” or “diy repair” instead of just “repair.” Review your negative keywords monthly against your search terms report to make sure you are not blocking queries that actually convert.
How to Read Auction Insights Without an MBA
Auction Insights shows six metrics. You only need to understand three of them to make decisions. Impression share tells you what percentage of eligible auctions you won. Overlap rate tells you how often a specific competitor showed up in the same auctions as you. Outranking share tells you how often you beat that competitor. The other three metrics (position above rate, top of page rate, absolute top of page rate) are useful for fine-tuning but not essential for a first read.
The Three Metrics That Matter
Impression share is the number that tells you whether your problem is budget, competition, or relevance. If your impression share is below 20%, you are barely showing up. Check the “Search Lost IS (Budget)” and “Search Lost IS (Rank)” columns at the campaign level to see why. If you are losing 60% of impressions to budget, you need to increase your daily budget or narrow your keyword targeting. If you are losing 60% to Ad Rank, you need to improve your ad copy, landing page experience, or bid amount. These are different problems with different solutions, and the data tells you which one you have.
Overlap rate tells you who your real competitors are, not who your SEO competitors are. If Competitor A overlaps with you 80% of the time and Competitor B overlaps 15%, Competitor A is your direct auction rival. Look up their ads in the Google Ads Transparency Center to see what ad copy they are running. If their ads promise something yours do not, that is a messaging gap, not a keyword gap.
Outranking share is the metric that tells you whether your bids are competitive. If Competitor A outranks you 70% of the time, you need a higher bid, better ad quality, or both. If you outrank Competitor A 70% of the time, your bids are fine and your conversion problem is elsewhere. Check the landing page.
How Often to Check Auction Insights
Check Auction Insights monthly, not daily. The data needs time to accumulate, and checking it too often leads to reactionary bid changes that destabilize Smart Bidding. Export the report to a spreadsheet each month and track the trend lines. You are looking for new competitors entering the auction, existing competitors scaling up, and your own impression share trend. A 5 point month-over-month change in either direction is worth investigating. A 1 point change is noise.
Your First $100: A Test Budget That Pays for Itself
You spent $500 with zero conversions and now you are skeptical. Fair. Here is how to spend $100 in a way that tells you whether Google Ads can work for your business, before you commit to a real budget. This is a research exercise, not a profit exercise. The $100 buys you data, not customers.
Step 1: Research (30 Minutes, $0)
Pull your top 5 keywords from your customer intake forms and Google Search Console. These are searches your business already has organic relevance for. Add 5 to 10 related keywords from Google’s Keyword Planner, which is free inside your Google Ads account. Filter by keywords with 100 to 1,000 monthly searches. Avoid keywords above 1,000 searches for your first test. High volume means high competition and high CPC, and you do not have the budget to learn there.
According to WordStream’s Google Ads benchmark data, the average CPC across all industries is in the range of $4 to $6. At $100 and an average $5 CPC, you will get roughly 18 to 20 clicks. That is enough data to identify whether any keyword produces a conversion or a meaningful engagement signal. If your industry has higher CPCs (legal and home improvement, for example, tend to run well above average), your $100 buys fewer clicks, so narrow to 3 keywords instead of 5.
Step 2: Build the Campaign (45 Minutes)
Create one Search campaign with one ad group per keyword theme. Set your match type to phrase match for your first test. Phrase match gives you more control than broad match while still capturing close variants. You are not ready for broad match until you have a negative keyword list built from real search terms data, which you will not have until after this test runs.
Set your daily budget to $10. At $10 per day, your $100 lasts 10 days. Set your bidding strategy to Maximize Clicks, not Maximize Conversions. You do not have enough conversion data for Smart Bidding yet. Maximize Clicks will get you the most clicks for your budget, which gives you the most search terms data to analyze.
Write two ads per ad group. Make them specific to the keyword theme. If the ad group targets “emergency drain cleaning,” the ad headline should say “Emergency Drain Cleaning” not “Full Service Plumbing.” Specificity raises your click-through rate, which raises your Ad Rank, which lowers your CPC. Write a different landing page URL for each ad group if you can. Sending “emergency drain cleaning” clicks to your homepage is the fastest way to waste $100. Send them to a page that talks about emergency drain cleaning.
Step 3: Read the Results (After 10 Days)
After 10 days, download the search terms report. Sort by cost per conversion ascending if you got any conversions. Sort by cost descending if you did not. Here is how to read what you see.
Kill signals: a search term with 8 or more clicks and zero conversions. The intent does not match your offer, or your landing page does not match the intent. Add it as a negative keyword. A keyword with CPC above $8 and zero conversions after 15 clicks. The auction is too expensive for your margin. Pause it and look for a more specific variant. A search term that is semantically related to your keyword but describes a different service. “Water heater installation” when you only do repairs. Negative keyword it.
Scale signals: a search term that produced a conversion at a cost per acquisition below your target. Move it to its own ad group, raise its bid, and write a dedicated ad for it. A keyword with a click-through rate above 5% and a cost per click below your industry average. The messaging and intent are aligned. Increase its budget share. A search term you did not bid on directly but that converted well through broad or phrase match. Add it as an exact match keyword to control it precisely.
| Metric | Kill Signal | Scale Signal |
|---|---|---|
| Clicks without conversion | 8+ clicks, zero conversions | 1+ conversion within first 10 clicks |
| Cost per click | Above $8 with zero conversions | Below industry average with clicks converting |
| Click-through rate | Below 2% after 50+ impressions | Above 5% with relevant search terms |
| Search term relevance | Query describes a different service | Query matches your offer exactly |
| Cost per conversion | Above your target CPA after 15+ clicks | Below your target CPA |
What the $100 Test Tells You
If you got at least one conversion from $100, you have a signal that Google Ads can work for your business. Scale the winning keywords, add negatives for the losers, and increase your budget to $300 to $500 per month for the next test cycle. If you got zero conversions but your search terms report shows relevant queries and your CPCs are reasonable, the problem is likely your landing page. Fix the landing page before spending more on ads. If your search terms report is full of irrelevant queries, your keyword targeting is too broad. Add negatives and tighten your match types.
If your $100 disappeared in two days on a handful of expensive clicks and the search terms report shows queries you never wanted to target, you have a keyword research problem. That is what this process is designed to fix. The $100 test is not a one-time exercise. Run it every time you add a new keyword theme. The alternative is the $500 test you already ran, where you learned nothing because you had no structure to read the results.
Once your keyword research produces a list of winning search terms, the next bottleneck is turning those terms into content. Your ad data tells you what people search for. Your blog should answer those same queries organically, so you eventually stop paying for clicks you can earn for free. If you want to see how that pipeline works, our breakdown of AI-driven SEO strategies for small businesses covers how to turn keyword research into published posts without an agency retainer. For the cost-per-post math behind that approach, our analysis of content ROI with AI tools shows why the agency model breaks small business budgets.
Try ClearPost free for 7 days. AI does the heavy lifting, you approve every post before it goes live. No long onboarding, no agency overhead, cancel anytime. See what 30 SEO-optimized posts a month looks like compared to the 4 you are getting now.
Frequently Asked Questions
Should I use broad match or exact match keywords in Google Ads in 2026?
Use both, but in different roles. Broad match with Smart Bidding finds new queries you would never think to bid on, including AI Overview placements that exact and phrase match cannot access. Exact match gives you precise control over your highest-value keywords. The old advice of exact match only is incomplete because Google’s AI Max for Search, launched May 2025, uses broad match and keywordless targeting to expand reach. Your negative keyword list is what makes broad match safe.
What is a good impression share in Google Ads auction insights?
There is no universal target. A 20% impression share might be fine in a competitive market with many advertisers, while 80% might be expected in a niche with few competitors. The number that matters is the trend. If your impression share drops 10 points month over month, a competitor has entered or scaled up. If it rises, your bids or ad quality are improving. Compare your impression share against Search Lost IS (Budget) and Search Lost IS (Rank) to understand why you are missing impressions.
How much should I budget for a Google Ads keyword test?
$100 is enough for a first test if you target keywords with 100 to 1,000 monthly searches and use phrase match. At an average CPC of $5.42 across industries, $100 buys roughly 18 to 20 clicks. That is enough to identify whether any keyword produces a conversion or a clear engagement signal. If your industry has higher CPCs, narrow to 3 keywords instead of 5. If you get at least one conversion, scale the winning keywords and increase your budget to $300 to $500 for the next test cycle.
What metrics signal I should kill a keyword vs scale it?
Kill a keyword if it gets 8 or more clicks with zero conversions, if its CPC is above $8 with no conversions after 15 clicks, or if the search terms it triggers describe a different service than what you offer. Scale a keyword if it produced a conversion below your target cost per acquisition, if its click-through rate is above 5% with relevant search terms, or if a search term you did not bid on directly converted well through phrase or broad match.
Why does my Google Ads search terms report not show all my search queries?
Google only reports search terms that meet a privacy threshold based on sufficient search volume across all Google searches. An independent analysis of over $20 million in ad spend found that hidden search terms waste up to $0.85 per dollar spent, with 52% higher CPCs on hidden queries. Google states the threshold is solely privacy-driven. You cannot see these queries, which is why account-level negative keywords and brand controls are essential to block categories of waste you can predict even when you cannot see the individual queries.