"Over 45% of our leads are nothing."
That's a direct quote from an immigration firm we sat down with this week. They spend about $30,000 a month on Google Ads and Local Services Ads. Sharp people. Good intake team. And close to half of what Google sends them is someone who thinks they called USCIS, someone looking for free legal aid, or someone applying for a paralegal job.
I hear a version of this on almost every call with a firm running Google Ads. The number changes. The story doesn't.
So let's break it all the way down: why Google does this, the six specific causes we find in law firm accounts, the exact order to fix them in, and how to judge the campaign by the one number that actually matters.
A quick, honest caveat first. None of this is Google's fault. That's going to sound strange coming from someone about to spend 2,500 words on the problem, but it's the whole point. Google is doing exactly what your account told it to do. Somebody told it a phone call equals success. Google found the cheapest phone calls on the internet. Your intake team is now paying for that instruction every morning.
Junk leads are a measurement problem and a match-type problem. They are not a Google problem. Once you see it that way, the fix is obvious.
Cause One: You're Optimizing toward the Wrong Conversion
Almost every Google Ads campaign now runs on Smart Bidding. Google describes it as bidding strategies that "use Google AI to optimize for conversions or conversion value in every auction." That's from Google's own help center.
Read that literally. It optimizes for whatever you defined as a conversion. It has no opinion about whether the conversion was any good.
Now look at what most law firms have marked as a conversion:
- ❌ Form submitted
- ❌ Call from ad
- ❌ Sometimes "clicked the phone number" on mobile
Google's documentation on primary and secondary conversion actions spells out the difference: primary actions are "used for bidding," secondary actions are "for observation only." If raw calls and raw forms are your primary actions, you've told the algorithm that a USCIS misdial is a win.
It will go find you more of them. Efficiently.
A $130 misdial and a $15,000 case count exactly the same in your account. That's the disease. Everything else in this post is a symptom.
Cause Two: Broad Match and Loose Query Matching
The second cause is how Google decides which searches your keywords should match. Every match type has gotten looser over the last few years, and most firms never noticed.
Here's what Google's keyword matching documentation says today:
- Exact match "may show on searches that have the same meaning or same intent as the keyword."
- Phrase match can match "the meaning of your keyword."
- Broad match "may show on searches that are related to your keyword, which can include searches that don't contain the direct meaning of your keywords."
To decide what's "related," broad match looks at "the user's recent search activities," "the content of the landing pages," and "other keywords in an ad group."
So your keyword is "immigration lawyer." Google decides these are related:
- immigration lawyer near me ✅
- uscis phone number ❌
- free immigration help ❌
- green card case status ❌
- immigration paralegal jobs ❌
Google also says it's "critical to use Smart Bidding with broad match" (broad match guide). Notice the circle. Broad match relies on Smart Bidding to filter bad searches. Smart Bidding relies on your conversion data to know what a bad search is. If your conversion data says every phone call is good, broad match has zero reason to avoid "immigration office phone number."
Google's June 2024 query matching update added brand inclusions and exclusions, which Google framed as broad match reach that only serves "on searches that include the brands you select." Great for Nike. Useless for a law firm whose prospects search for a problem, not a brand name.
In our accounts, broad match is something you earn. Google has to know what a qualified lead looks like first. Before that, broad match buys reach, and reach in legal means a lot of people who will never hire you. I go deeper on match types and campaign structure in our Google Ads for law firms guide.
Cause Three: Weak Negatives and the Hidden Search Terms Problem
The third cause is what's missing from the account.
Here's the part that makes people angry once they understand it. Google matches your keywords loosely. It enforces your negatives literally. Google's negative keywords page says it plainly: "negative keywords won't match to close variants or other expansions." Your targeting stretches. Your negatives don't. Google started covering misspellings of negatives in 2024. Plurals, synonyms and rephrasings are still on you.
It gets worse. You can't see most of what you're paying for.
In September 2020, Google changed the search terms report to show only terms searched by a "significant number of users". Everything else got collapsed into one row: "Other search terms." A three-account analysis, one of them a law firm, found 24% to 54% of clicks sitting in that hidden bucket, at higher CPCs than the visible terms. We've managed accounts where we could see 10% or less of the actual searches typed in.
Read that again. Half your clicks, and you don't get to know what people searched.
If you can only see a fraction of the queries, you can only build negatives from a fraction of the queries. The rest has to come from your intake team. They hear the tenth "is this USCIS?" call of the week long before any report shows it.
Most firms treat negative keywords as a launch task. Build the list, set it, forget it. Wrong. It's a daily job. Two bad days of Google chasing bad searches will burn a week of budget, and you won't see it in the report until the money is gone.
Cause Four: Performance Max without Offline Conversions
Now the campaign type the immigration firm was actually complaining about.
Google describes Performance Max as "a goal-based campaign type that allows you to access all Google Ads inventory, including YouTube, Display, Search, Discover, Gmail, and Maps, from a single campaign" (Google Ads Help). You give it assets, a budget and a conversion goal. It decides where to serve. There are no keywords to tighten.
That makes PMax the purest expression of cause one. Whatever you defined as a conversion, PMax chases across six channels at once. Display and Discover produce very cheap phone calls from very confused people.
PMax does get a cheaper cost per click than Search. That's the pitch, and it's true. The catch is that it will do anything possible to hit the goal you gave it. If the goal is "any phone call," it will happily spend your budget on placements that produce spam calls at scale, because that's the cheapest way to hit the goal.
Google has added guardrails. In March 2025 it raised campaign-level negative keywords in PMax from 100 to 10,000. Brand exclusions and account-level negatives apply too. Use all of it.
But negatives trim the edges. The only thing that changes what PMax hunts for is the conversion you feed it. In our accounts, PMax doesn't go live until offline conversions are flowing back with the click ID attached. Before that, it buys spam at a discount. After that, it buys consultations at a discount.
Cause Five: Landing Pages and Ad Copy that Qualify No One
Google grades every keyword on Quality Score: expected clickthrough rate, ad relevance, and landing page experience, which Google defines as "how relevant and useful your landing page is to people who click your ad."
Most people care about that for cheaper clicks. I care about it for a different reason. A page built to be relevant to everyone qualifies no one.
Look at your landing page right now. I'll bet it says some version of "Free Consultation. Call Now." above a three-field form. Name, phone, email, submit. The person who wants their case status fills it out in nine seconds. So does the person with a $10,000 matter. Your form can't tell the difference, so Google can't either.
Google has quietly conceded this point with its own product. Its best practices for lead form assets now recommend "qualifying responses to filter out low-intent prospects," and its qualifying responses documentation lets you mark certain multiple-choice answers as qualifying so submissions can be "filtered or segmented" by intent.
If Google is adding screening questions to its own forms, your landing page can ask two.
The same rule we use on Meta campaigns applies here: the broader the message, the broader the quality of the client. You do not want it to be too easy to fill out the form. For the immigration firm, that means the naturalization page can push straight to a phone call, while the T visa or VAWA page states the eligibility criteria in plain language before the form ever appears.
Copy that names who you help and who you don't will cost you clicks. Those were the clicks you were about to complain about.
Cause Six: Spam and Invalid Traffic
This is the one firms blame first. It's real. It's also much smaller than it feels.
Google defines invalid traffic as "traffic that doesn't represent genuine interest in your business": bots, accidental clicks, competitors clicking your ads on purpose. Google says invalid traffic "detected before the end of the month is automatically removed from your billing," and anything caught later comes back as a credit.
Want to see how much is being filtered? Add the "Invalid clicks" column to your campaigns table. If you suspect a pattern Google missed, you can file a Click Quality Form covering the past 60 days with your server logs and click IDs (Google Ads Help). And Google's Ad Traffic Quality resources state the obvious fix for form spam: "Adding a CAPTCHA to your forms will help validate that the person submitting it is human."
Do those things. reCAPTCHA or a honeypot field kills most bot form fills.
But here's the truth. When a firm tells us half its leads are junk, the junk is almost never bots. It's real people with real problems your firm doesn't solve, delivered by a campaign that was told to find them. That's causes one through five. Not cause six.
The Fix, in Order
The sequence matters here. Skip ahead to "loosen targeting" or "let's try PMax" before the data is clean and you just buy junk faster.
Step One: Fix Attribution
You cannot teach Google anything until every lead carries the click that produced it.
Turn on auto-tagging. Every click gets a GCLID, which Google describes as unique IDs "for every click that comes to your website from an ad" that you save "along with whatever lead information you collect" (Google Ads Help).
Then clean up the online conversions. The immigration firm's setup was a soup: some conversions firing from the site, some imported from GA4, nothing offline. Pick one source of truth. We use the Google Ads conversion tag directly, not GA4 key events imported into Google Ads. The tag updates faster and credits the click more aggressively, which is what you want for last-click bidding. GA4 imports can take up to 24 hours to reach Google Ads. For a bidding algorithm making decisions every auction, a day is forever.
Forms are easy. A hidden field captures the GCLID. Phone calls have no URL, so you need call tracking. CallRail's Google Ads integration "posts conversions using Google's click IDs" and requires the CallRail snippet on your pages, visitor tracking turned on, and auto-tagging enabled.
Then the click ID has to land in your CRM. Lawmatics documents both a CallRail integration and sending offline conversions to Google Ads on a status change via Zapier or webhook. Whether a CallRail call's GCLID reaches the Lawmatics matter automatically is the piece you test in your own account.
I'll be honest. This step is like pulling teeth. It's also the single highest-value thing in this entire post. If you want the longer version of the tracking setup, I covered it in lead tracking for law firms.
Step Two: Define a Qualified Lead with Intake
Sit down with whoever answers the phone and agree on a definition. Not "good lead" as a feeling. A CRM stage:
- Booked consultation
- Consultation completed
- Retainer signed
Then have intake tag every unqualified lead with a reason: wrong practice area, wanted a government office, wanted free help, outside your geography. Those tags become your negative keyword themes and your landing page copy.
Automate the weekly export from the CRM. Nobody spending $30,000 a month wants to maintain a spreadsheet by hand, and nobody will.
Step Three: Feed it Back to Google
Create an offline conversion action called "Booked Consultation." Fire it from the CRM on the stage change, with the GCLID, the timestamp and, eventually, a value. Google recommends uploading at least daily.
Now the important part. Make that action primary. Demote raw calls and forms to secondary. They still get reported. They no longer steer bidding. Google warns that Smart Bidding "will take some time to learn (1-2 conversion cycles in most cases)" after a goal change (Google Ads Help), so expect a few bumpy weeks.
When you have the volume, upgrade to enhanced conversions for leads, which Google calls "an upgraded version of offline conversion import" that adds hashed email or phone data to improve matching.
Start with booked consults, not signed cases. Google's guidance for value-based bidding asks for at least 15 conversions in the last 30 days. Most firms don't sign 15 paid cases a month from Google. Consults happen sooner and more often, and they've already filtered out the USCIS callers.
Step Four: Loosen Up, then Test PMax
Only now do the things your Google rep has been begging you to do.
Move a few ad groups to broad match. Ease off the most aggressive negatives. Then launch a Performance Max campaign pointed at the offline conversion.
Offline conversions let you break every rule you've been living under. Once Google is optimizing toward the booked consultation instead of the phone call, you can stop being paranoid about negatives, bid more broadly, and let PMax hunt across six channels. Because now it's hunting for the right thing.
How to Measure Whether it Worked
Stop reporting cost per lead.
WordStream's 2026 benchmarks put the average cost per lead for attorneys and legal services at $131.63, the highest of any industry they track, with an average CPC of $9.87 and a conversion rate of 5.55%. That number is meaningless on its own. A $130 USCIS misdial is $130 wasted. A $400 lead that becomes a $15,000 case is a bargain.
The only metrics that matter:
- Cost per qualified lead (booked consult)
- Cost per signed case
The catch is lag. A click today might become a consult next week and a signed retainer next month, so this month's cost per signed case always looks terrible. We report it on a trailing 30-day basis and reconcile as matters close.
I know what you're thinking. "Ryan, I don't need a spreadsheet, I can feel when the leads are good." No, you can't. The last thing I want to hear from a client is "it feels better" or "it feels worse." One client was certain he'd signed nothing in two months. His own intake company's reporting said the campaign was fine. Feelings lose to a CRM export every time.
And audit intake itself. Clio's Legal Trends research found that when researchers contacted firms as prospective clients, about 40% picked up the phone and 33% responded to email. Some of the "junk" on your lead report is a good prospect who gave up waiting. That's an intake problem, and we wrote a whole playbook on it for personal injury intake.
When Junk Leads are Actually Fine
One caveat, because not every firm should chase a spotless lead list.
If you run a high-volume practice, your CPL is cheap, and intake is staffed to work through noise fast, a 30% junk rate on a $60 lead can beat a 5% junk rate on a $300 lead. Top-of-funnel volume plays are legitimate, especially where the qualified prospect and the curious one type the same words and only a phone call separates them.
The test is one line of math: monthly spend divided by signed cases. If that number is where you need it and intake isn't drowning, leave the campaign alone.
What's never fine is not knowing which situation you're in. And that's a measurement problem again.
Where This Leaves You
If half your Google Ads leads are nothing, the campaign is delivering exactly what it was asked for. Change the ask.
- Get the click ID onto every lead.
- Get the consult stage flowing back to Google.
- Tighten the page until the wrong people stop filling out the form.
- Then, and only then, let broad match and Performance Max buy cheaper clicks for the outcome you actually want.
Do that and the 45% number drops. Not because Google got smarter. Because you finally told it what a case looks like.
Want Us to Look At Your Account?
This is what we do at WEBRIS. We run Google Ads for law firms every day, and the first thing we'll do is show you what your account is actually optimizing toward, what the hidden search terms are costing you, and how far you are from offline conversions. No pitch deck. Just the numbers. Get your free growth plan here.
