Often because the campaign is doing exactly what it was told. If it bids toward clicks or every form fill, neither says anything about quality, and Google’s automation will keep finding more of the same as efficiently as it can, junk included. To get better leads, Google has to learn which ones turned out to be good, which means getting what your sales team learns back into the ad account. One distinction first: if the leads are the wrong people, start with targeting and what the campaigns count as success. If they’re the right people but too expensive, the question is whether they turn into enough profitable business to justify the cost.
I review Google Ads accounts in my practice and the agency I run manages them, but everything below is something your own team or current agency can check.
In this short video I go through what I’d check in the account itself.
Start with the last few months of leads
Before changing anything, pull the leads from the last 30 to 90 days and have someone go through them with sales. For each poor lead, note why: wrong size, wrong service, job seeker, spam, existing customer, outside your area. Then check a sample of the “junk” yourself. Read the form, listen to the call recording if there is one, and see how quickly someone followed up.
Reviewing them tells you which problem you actually have, because job seekers and spam need different fixes. And it tests the sales team’s judgment. A lead can look like junk because nobody called it for three days, or because the buyer wasn’t ready this quarter. If good leads are being mishandled, no change to the ads will fix it.
Read the pattern before you fix anything
The shape of the numbers usually points to where to look first:
| What you’re seeing | Where I’d look first |
|---|---|
| Lots of form fills, almost none qualified | What the campaigns count as a conversion, search terms, spam |
| Good-fit inquiries, few sales meetings | Follow-up speed and how sales qualifies |
| Plenty of the right visitors, few inquiries | The landing page, the offer, the tracking |
| Qualified leads, few closed deals | Sales process, the lead definition, the offer |
| Expensive clicks, but profitable customers | Possibly nothing; check the economics first |
| Few leads, but they close well | Search demand, budget and reach |
Some checks can happen at the same time. Obvious junk searches don’t need to wait for a CRM connection before you exclude them.
Check what the campaigns are bidding toward
Then open the ad account and look at three things for each campaign:
- The bid strategy. Is it bidding for clicks, for conversions, or for conversion value?
- Which conversion actions the campaign actually bids on. An account can record many actions (form fills, calls, downloads, page visits). An action generally has to be marked primary and included in the goals the campaign uses; custom goals have exceptions. Don’t assume everything recorded is steering bids.
- What those actions actually are. If the action that counts is any form submission, the campaign is rewarded equally for a job seeker and a qualified buyer. Check calls too: an account may count any call over a certain length as a conversion, even if it was a vendor pitch or a support request.
This is where junk compounds. If spam form fills count as conversions, they contaminate the data Google uses to decide which clicks are worth paying for, and the system can get very good at producing submissions that are worth nothing to you. Looking at one account recently, I said to the team that I was really starting to wonder if we were training Google to go after trash, because we were optimizing toward conversions and not conversion values. Check for duplicates too: on that account a leftover backup form conversion was still switched on, so each junk fill was being rewarded twice. The fix the team settled on was to keep spam and sales pitches out of what goes back to Google as a conversion, while still letting them reach the CRM so someone can see them.
Decide what a good lead is
A lead worth having is two things at once:
- Fit: is this the right kind of company, in the right size, place and situation, for what you sell?
- Intent: did they do something that shows they actually want to buy?
Many campaigns only reward intent, because a form fill or a phone call is easy to count. That ignores fit, which is how you end up with a sales team full of tire kickers. The lead you want is both: the right company asking for something real.
Sales and marketing need to agree on that definition, in writing, before anything else changes. If they don’t, marketing will keep celebrating volume while sales keeps complaining about quality, and both will be right by their own measure.
Tell Google which leads were good
Google’s bidding can only learn from what it’s told. So what your sales team knows, which leads were real and which turned into deals, has to get back into the ad account. This is usually called offline conversion tracking: your CRM records what happened to each lead, and that result is sent back to Google.
For that to work, each lead has to be connected to the ad interaction that produced it. That can mean capturing Google’s click ID on your forms, or using what Google calls enhanced conversions for leads, which matches details such as the email address someone submitted. Google has specific requirements and time limits for these imports, so ask your team to confirm they’re using the current setup and that the sales stages you care about still fall inside the window.
Once the matching is set up properly, the upload itself can start simple. Early on it can be a spreadsheet: someone marks each lead as good or bad, and the good ones go back to Google as a way of saying “find me more people like these.” With a CRM, it can be automated. Either way, send it on a regular schedule, ideally daily, not as an occasional clean-up. Then check it works with one question: of the leads sales marked qualified last month, how many actually showed up in Google Ads? If sales qualified twenty and Google received five, the loop isn’t doing what you think.
High-ticket B2B sales cycles are long, and that creates a problem. If Google only hears about closed deals, it’s working blind for months. So track earlier signals as well: a lead the sales team has confirmed is real, a first meeting, an opportunity opened. Then choose carefully which of them the campaigns bid on. Bidding for a properly defined qualified lead can work without putting a value on every stage. Others have enough data to give each stage an estimated value based on how often it turns into revenue. Don’t invent values just to switch on value-based bidding, and if you send several stages of the same deal, make sure nothing counts the same expected revenue twice.
Then make sure the campaigns actually use that information. Values only steer a value-based bid strategy, and a stage only steers conversion bidding when it’s in the campaign’s goals. The order matters too. Bidding for conversions aims to get as many of the chosen conversions as the budget allows, which in practice tends to mean the easiest ones to get; bidding for value aims for the most value. Switch to value while junk is still being counted, and you’ve told Google junk is worth chasing. Clean the signal first; value bidding comes later, if your volume supports it.
One caveat. If your account produces only a handful of qualified leads a month and fewer deals, better signals help, but the system may still have too little to learn from. In that case, bid toward the earliest stage that’s both frequent and meaningful, such as a lead sales has confirmed is a real prospect, and use the later stages to check whether that stage predicts revenue.
Check where your ads are showing
Once the campaigns are aimed at the right outcome, look at where the clicks come from:
- Search terms. Read the searches that triggered your ads. A quick test I use: sort them by impressions and ask whether the people behind the top searches look like customers. Google doesn’t show every search, especially in small accounts, but you’ll see enough to spot job hunters and services you don’t offer. Exclude those. Be careful with research searches, though; in a long B2B buying process, some of them are your future customers. A short list of the right keywords beats a long list of maybes; I’d be fine bidding on two keywords if we knew they were the right ones.
- Broad match. Broad match lets your keywords trigger ads on a wide range of related searches. It can find good buyers you’d never have targeted, and a lot of noise.
- AI Max. This is a set of features inside Search campaigns that can match ads to searches without your keywords, rewrite ad text and send people to other pages on your site. Each feature has its own setting, so check which are on and what searches and pages they’re producing.
- Search partners. Search campaigns can also show on Google’s partner sites. Check whether that’s on, and whether those leads hold up.
- Performance Max and Demand Gen. These campaign types show ads beyond search, on YouTube, websites and apps. Look at results by campaign and channel, and at whatever placement reporting is available, with one caveat: the placement reports don’t give a complete breakdown of which sites produced your leads. Where you can see the sites, ask whether your buyers would actually be there. Cheap clicks and form fills from junk sites aren’t a reason to keep them.
- Retargeting. Retargeting works when the right people visited. If the campaigns are sending you the wrong visitors, you’ll also spend money following those wrong visitors around the internet.
- Location. Check whether your campaigns target people in your area, or also people who have shown interest in it. The second setting can bring in inquiries from places you don’t serve.
None of these is wrong in itself. Judge each by whether it produces qualified opportunities.
Check whether the landing page matches the search
This is the one to look at first if your complaint is cost rather than quality: clicks feel expensive, and too few of the people who click do anything.
Someone searches for something specific, your ad promises it, and the page they land on should deliver that same thing in the first few seconds. Landing-page specialists call this message match. When the search, the ad and the page drift apart, visitors have to work out whether they’re in the right place, and someone who can’t tell quickly has little reason to stay.
Google gives you a clue. Each keyword’s Quality Score includes a landing page experience rating of below average, average or above average. Google calls Quality Score a diagnostic, not something used in the auction itself, though related quality factors do affect where your ads rank and what you pay (Google Ads Help, “About Quality Score,” 2026). A below-average rating on important keywords is a reason to investigate, not proof the page is the problem, and a good rating doesn’t mean the page converts. The rating also won’t tell you why, so compare it with what visitors actually do and how many qualified leads the page produces.
Keep the cost in proportion, though. Much of what you pay per click comes from competition, alongside your bid and quality. Google Ads runs as an auction, and the more advertisers bidding on the same searches, the more each click costs. A better page helps at the margin. It won’t make a crowded keyword cheap.
What a matched page looks like is simple to describe. If I were building an account for a firm with four services from scratch, each service would get its own campaign or ad group, with its own ad copy and its own landing page, worded the way that searcher would ask the question, instead of sending every ad to the homepage or a general services page. As a made-up example, say someone searches “marketing consultant for law firms.” An ad promising marketing help for law firms that lands on a page headed “We help businesses grow,” with testimonials from a dentist and a software company, makes that person translate. A page that names law firms in its headline and shows work for law firms answers the search before they have to wonder.
Two quick tests help. Put the ad and the page in front of a few people who resemble your buyer for about 20 seconds, then ask what you offer and who it’s for. If they can’t say, a stranger from Google can’t either. And because paid traffic arrives fast, it’s a good place to split-test the page itself: headlines, imagery and which form fields people actually fill in. Keep a winner only if it brought in qualified leads, not just more form fills.
The page can also do some qualifying. If it reads as though it’s for anyone, anyone will fill it in. Say who you work with and roughly what size of work you take on, and ask a question or two on the form that helps sales tell a real buyer from a browser.
Just don’t expect the page to fix lead quality by itself. A matched page can lower what each lead costs and lift how many visitors convert. If the campaign is still rewarded for any form fill, a better page can simply deliver more of the wrong leads, more cheaply. Quality comes from what the campaign is told to find and what sales sends back, which is why this check comes after those.
Then reduce spam before it distorts the numbers
Beyond what it does to bidding, spam wastes your team’s time and muddies the numbers. A spam filter on your forms helps; just make sure rejected submissions can still be reviewed, and that spam never counts as a valuable conversion in Google Ads. One industrial company found that after it added a spam filter, nearly all of its paid leads looked like real prospects, and most of the poor ones left were existing customers looking for support.
A person on your side still has to look at the leads. Some spam looks like real people filling in the form, and good-fit buyers sometimes arrive looking like strangers. The right answer for those is to qualify them, not filter them out.
What to expect when you fix it
Lead volume may drop and cost per lead may rise. That’s fine if the cost of each qualified lead falls. A made-up example:
| Made-up comparison | Before | After |
|---|---|---|
| Cost per form fill | $100 | $200 |
| Share that are qualified | 1 in 10 | 1 in 2 |
| Cost per qualified lead | $1,000 | $400 |
The second scenario looks worse on cost per lead and is much better for the business. It’s also a good question to ask internally: is marketing being held to form fills, qualified opportunities or customers won? Cost per qualified lead still isn’t cost per customer, so follow it through to opportunities and deals.
Tightening too far has its own cost. One company that cut its targeting hard saw volume fall further than it wanted, and planned to check whether some of the searches it had excluded had once produced good leads. Watch quality and volume together.
Don’t change every campaign at once, either. Changes to bidding goals can disrupt delivery while the system adjusts, so agree what you’re testing, how long you’ll give it and what would count as better.
It also takes time, and the effort varies. Changing conversion settings and exclusions is quick. Connecting a CRM to Google Ads properly, and keeping it working, can be real work. On one account, it stalled because qualified leads were being marked in the CRM and never sent back to Google. Someone has to own that connection.
When Google Ads isn’t the answer
Better signals can’t create demand that isn’t there. Some B2B markets have very few people searching for what you sell, or the searches that exist come mostly from people wanting something different. If the account is set up properly, sales is following up well, and the cost of each qualified opportunity is still too high for what a customer is worth to you, the conclusion is that this channel doesn’t fit your market right now. Judge it on cost per qualified opportunity, how many of those opportunities close, and what a won customer is worth (the subject of how much you can afford to spend to win a customer), over long enough to cover a sales cycle.
I review Google Ads accounts, tracking and the path from lead to CRM in my Google Ads consulting. Ask about a review.
