Why Your Google Ads Leads Are Low Quality

The account is generating leads. Cost per lead looks respectable, the graphs point the right way, and yet sales keeps reporting the same thing: dead phone numbers, bounced emails, enquiries from students, job-seekers and bots. If that sounds familiar, you have a Google Ads lead quality problem, and it is one of the most misdiagnosed problems in paid media.

The reflex is to blame the audience. Wrong keywords, wrong people, “Google Ads doesn’t work for our industry”. Occasionally that is true. Far more often the account is doing exactly what it was told to do: the instructions were wrong.

The core position

Low-quality leads are often a measurement architecture problem, not an audience problem.

Why Google Ads lead quality is a measurement problem

Smart Bidding optimises towards whatever you count as a conversion. It does not know what a good customer is; it knows what fires your conversion tag. If the tag fires on spam, the algorithm becomes a machine for buying more spam, efficiently, automatically and at scale. Every optimisation it makes takes the account further in the wrong direction, and it does so with confidence, because by its own definition it is winning.

Below are the ten causes we check when we audit lead quality inside our Ads engine, grouped the way we work through them: traffic first, then measurement, then the feedback loop between your sales team and Google.

Traffic faults: the wrong people are clicking

1. Broad or irrelevant search queries

Broad match keywords paired with Smart Bidding give Google licence to interpret intent, and it interprets generously. A campaign bidding on “commercial cleaning services” can end up matched to “cleaning jobs near me”, “how to clean office carpet” and “free cleaning checklist template”.

Open the search terms report and read a month of real queries: not the summary, the queries. If a meaningful share of them could never become a customer, your lead quality problem starts here. Broad match is only safe when your conversion signal is clean enough to teach the algorithm what a good match looks like, which (as the measurement section below usually reveals) it is not.

2. Search Partner traffic

The Search Partner network extends your ads beyond Google search to a long tail of third-party sites and embedded search boxes. Some of it is fine. Some of it is where form spam lives. Because partner placements are opaque, you cannot see which sites sent the traffic, only that conversions from the segment look suspiciously cheap.

Segment your campaigns by network. If Search Partners shows a flattering cost per lead that sales never manages to validate on the phone, untick it and watch your lead quality recover. Cheap conversions that never close are not cheap.

3. Incorrect location settings

Google’s default location setting includes people who have shown interest in your targeted area, not just people physically in it. That is how a Manchester plumber ends up paying for clicks from another continent: someone abroad searching “plumber manchester” counts as interested.

For any business that serves a physical area, switch the setting to presence only, then check the user location report for leakage. It is a two-minute fix that regularly explains a stubborn tail of unreachable leads.

4. Weak negative keyword coverage

Negative keywords are the only permanent instruction you can give Google about who you do not want. Most underperforming accounts have a few dozen scattered negatives added in a panic months ago; healthy accounts have structured shared lists that grow every week.

Build lists by intent category: job-seekers (jobs, salary, careers, hiring), learners (course, how to, tutorial, certification), freebie hunters (free, template, DIY, cheap), and adjacent-but-wrong services. Then mine the search terms report on a weekly cadence. Negative coverage is a compounding asset: every exclusion improves every future auction.

Measurement faults: Google is learning from the wrong signal

Traffic faults waste money. Measurement faults are worse: they teach the algorithm to repeat the waste.

5. Conversion tracking that rewards spam

The most common setup we find counts every form submission as a conversion: bot fills, gibberish entries, duplicate submissions, competitors doing research. Some accounts fire the conversion on a thank-you page URL that crawlers can reach directly, so a “conversion” does not even require a form.

Every one of those events lands in Google’s learning system labelled “this is what success looks like”. Smart Bidding then goes looking for more people who behave like your spam. Validation, deduplication and bot filtering are not tracking hygiene extras; they decide what your budget optimises towards. It is the same discipline we apply to revenue measurement in our ROAS tracking guide: the number you feed the machine matters more than the number on the dashboard.

6. Forms that collect no qualification data

A name-and-email form maximises submissions and tells you nothing. You cannot separate a buyer from a browser, so you cannot score leads, so every fault above stays invisible until sales complains.

Add one or two qualifying fields: company size, budget band, timeframe. Raw conversion rate will dip; qualified volume usually holds or climbs, because the field itself filters out people who were never going to buy. The form is not just a capture mechanism, it is the first stage of qualification.

7. Optimising for leads instead of qualified revenue

Target CPA on raw form fills gives the algorithm one job: find the cheapest possible form-fillers. It will do that job brilliantly, and the cheapest form-fillers are rarely your best customers.

The fix is to move the optimisation target down the funnel: assign values to conversion stages (enquiry, qualified lead, opportunity) and bid to value rather than volume. A £5 form fill and a form fill that becomes a £5,000 client should not look identical to your bidding strategy, and by default they do.

8. Landing-page message mismatch

If the ad promises one thing and the page delivers another (or delivers something so generic anyone could believe it is for them), you get wrong self-selection: the right people bounce and the wrong people fill in the form.

Qualification copy is targeting. “From £X per month”, “for teams of ten or more”, “minimum project size”: lines like these politely repel the wrong clicker before they cost you a form fill and a sales call. A landing page that converts fewer, better visitors is usually the more profitable page.

If your CRM is filling up with enquiries sales refuses to touch, have Gyrodile trace where the junk enters your funnel. In most audits it is one of the four measurement faults above, and the fix takes days, not quarters.

Feedback faults: sales knows, Google doesn’t

9. Sales feedback never returns to Google

Your sales team already knows which leads answered the phone, which qualified and which turned into pipeline. In most accounts, none of that knowledge ever reaches Google. The platform keeps optimising to the form fill because the form fill is the last thing it can see.

At minimum, run a weekly disposition process: sales marks every lead as junk, unqualified or qualified in the CRM, and someone owns feeding the pattern back into the account, as negatives, exclusions and bid changes. The teams that close this loop manually still beat the teams that close it not at all.

10. Imported offline conversions are missing or broken

The proper version of that loop is offline conversion import: capture the Google click ID (or use enhanced conversions for leads), store it on the CRM record, and upload qualified, opportunity and closed-won events back into Google Ads. Now the algorithm optimises towards leads that become revenue, not leads that fill forms.

Where it exists, it is often broken: the click ID is never written to the CRM, uploads run outside the attribution window, duplicates inflate the numbers, or the imported action is left as a secondary conversion nothing bids against. An import that silently fails is worse than none: it looks like measurement while the account still learns from form fills.

How to build a qualified-lead conversion

If you change one thing after reading this, make it this sequence:

1. Define “qualified” with sales, in writing. One sentence, unambiguous, agreed by both teams. For example: right service, right geography, budget confirmed, reachable within two working days.
2. Capture the click ID on every lead. Store the GCLID in a hidden form field and write it to the CRM record, or implement enhanced conversions for leads.
3. Create conversion actions for pipeline stages. Keep the raw form fill, and add qualified lead, opportunity and closed-won as distinct actions.
4. Import offline conversions on a schedule. Automated daily is ideal; a weekly manual upload still transforms the signal.
5. Assign values by stage. Even rough, directionally honest values beat equal-weighting everything.
6. Promote the qualified action to primary once it has volume. Smart Bidding needs data to learn from. If the qualified action fires only a handful of times a month, keep the raw lead as primary and let stage values carry the difference until volume grows.
7. Review search terms against qualified leads, not raw leads. The queries that produce form fills and the queries that produce customers are different lists. Optimise to the second one.

Fix the signal before you blame the audience

Almost everything above is diagnosable in an afternoon with the search terms report, the conversion settings page and an honest conversation with sales. The method matters more than the tools: find the mechanism, don’t guess at the symptom, the same forensic sequence behind the audit that recovered 40% of a client’s lost traffic after Google’s Helpful Content Update.

And if an agency runs the account and none of this architecture exists, that is a conversation worth having this week: measurement is precisely what a competent management fee should buy, as we set out in our breakdown of Google Ads agency pricing.

We’ll identify where your spend turns into noise.