Guide · Lead quality diagnosis
Junk leads from paid ads: find the cause before you cut the budget
“The leads are junk” describes a symptom with five different causes. Spam, job-seekers, poor-fit companies, casual researchers and leads nobody reached each need a different fix, and cutting budget fixes none of them.
Why do paid ads bring junk leads?
Short answer
Usually because the campaign is optimising to form submissions, so the platform finds whoever submits most cheaply: students, job-seekers, bots or poor-fit businesses. Broad keywords or audiences, soft offers and easy prefilled forms make it worse. Classify the junk first, because spam, wrong fit and unreachable leads each need a different fix.
Ad platforms are extremely good at one thing: finding more people who complete the action you report. If the action is “submitted a form”, the platform has no way to know that half the submissions came from people who will never buy. It sees cheap conversions and looks for more of them.
Four settings then widen the net:
- Broad reach. Broad match keywords, expanded audiences and automatic placements find more people, including people with little intent.
- Soft offers. A free guide or catalogue attracts researchers, students and competitors as readily as buyers.
- Frictionless forms. Forms pre-filled from a profile take seconds to submit, sometimes without much thought.
- No qualifying questions. Nothing on the form separates a plant manager from a trainee.
None of this is a malfunction. It is the system doing what it was told. That is why the first job is diagnosis, not a budget cut: the same channel can be both your largest source of junk and your cheapest source of qualified leads.
Original framework
The five types of junk lead
Tag every bad lead with one type. The type points to the cause, the first check and who owns the fix.
| Junk type | How it shows up | Most likely cause | First check | Owner |
|---|---|---|---|---|
| Spam or bot | Gibberish names, repeated submissions, fake numbers | Unprotected website forms; some placements | Form logs, placement report, whether the conversion fired on spam | Marketing |
| Invalid: student, job-seeker, vendor | A real person, but not a buyer | Broad keywords matching “jobs”, “course” or “PDF”; careers-style wording in ads | Search terms report; ad copy | Marketing |
| Wrong fit | A real business you can’t serve: too small, wrong region, wrong buyer type | Targeting too broad; no qualifiers on the page or form | Location and company-size targeting; form fields | Marketing |
| Wrong intent | Fits, but is researching rather than buying | Soft or gated offers; pre-filled forms | Offer and form type by campaign | Marketing |
| Unreachable | Valid, but never successfully contacted | Slow follow-up, wrong contact field, calling at the wrong time | Time to first contact; number of attempts | Sales |
“Unreachable” is on the list so it can be separated out, not blamed on ads. A valid lead that nobody called for two days is a follow-up problem.
In what order should you diagnose junk leads?
Short answer
Work from measurement outwards: tracking first, then search terms and placements, then audience, then offer and creative, then the form, then follow-up. Each earlier check can make later ones unnecessary, and the cheapest, safest fixes come first. Bids and budget come last, because changing them restarts learning.
Tracking
Does the conversion fire only on a real submission? Are spam and duplicates counted? If spam counts as a conversion, bidding is being rewarded for it.
Search terms and placements
Which queries and placements produced the junk? Invalid leads usually cluster in a handful of terms.
Audience and location
Is expansion switched on? Is location set to people in your area, or also people interested in it? Is company size or seniority targeted where the platform allows?
Offer and creative
Does the ad promise something a buyer wants, or something anyone would download? Does the wording read like a job advert?
Form
Is there one question that separates buyers from everyone else, such as company size, project type or quantity?
Follow-up
How fast did sales reach out, by which channel, and how many times? This decides whether “unreachable” is really junk.
Template
The 7-day junk audit
A week of part-time work with a spreadsheet, no CRM needed. It turns an argument into a table.
Columns
- Lead ID and date
- Platform, campaign, ad group or ad set
- Search term or audience, where known
- Form type: website form, Google lead form, Instant Form, Lead Gen Form
- Junk type: spam, invalid, wrong fit, wrong intent, unreachable, or valid
- Time to first contact, in hours
- Note, one line
The week
- Days 1–2: export the last 30–60 days of leads with their source fields.
- Days 3–4: sales and marketing tag each lead together, one type each.
- Day 5: pivot junk share by platform, campaign and form type.
- Day 6: for the biggest type, run its first check from the taxonomy.
- Day 7: agree one change per type, an owner and a re-check date.
| Platform | Spend | Leads | Invalid | Wrong fit | Spam | Unreachable | Valid | Cost per valid lead |
|---|---|---|---|---|---|---|---|---|
| Google Search | ₹2,60,000 | 220 | 70 | 25 | 5 | 20 | 100 | ₹2,600 |
| Meta Instant Forms | ₹1,40,000 | 180 | 18 | 35 | 27 | 20 | 80 | ₹1,750 |
| Total | ₹4,00,000 | 400 | 88 (22%) | 60 (15%) | 32 (8%) | 40 (10%) | 180 (45%) | about ₹2,222 |
Sales’ complaint was about Meta, which produced 27 of the 32 spam leads. The audit shows something else. Google produced most of the invalid leads, 70 of 88, and they cluster on broad match terms around “HR software” that also matched “HR jobs” and “HR course” searches. Meta’s valid leads cost ₹1,750 against Google’s ₹2,600. Cutting Meta would have removed the cheaper source of valid leads.
The 40 unreachable leads, split evenly, are a follow-up question for sales, not a targeting question for either platform. The decisions from this audit are in the table further down.
Fixes on Google Ads
- Search terms and negatives. The search terms report shows the searches that triggered your ads; Google’s advice is that if a term isn’t relevant enough to what you offer, add it as a negative keyword (Google Ads Help). Negative keywords don’t match close variants, so add synonyms and singular and plural forms yourself; casing and misspellings are handled automatically (Google Ads Help). A B2B starter list: jobs, careers, vacancy, salary, course, training, internship, PDF, free, meaning.
- Location option. Google offers “Presence”, people in your targeted locations, and “Presence or interest”, which also reaches people interested in them. Google recommends the broader option for Search but notes Presence suits advertisers who want to restrict reach to people physically in the area (Google Ads Help). If wrong-region leads are a large junk type, test Presence.
- Lead form qualifying responses. On Search campaigns, qualifying questions on Google lead forms collect needs or purchase readiness before submission, are mandatory for the user, and automatically create a “Lead form – Response qualified (Google-hosted)” conversion (Google Ads Help).
- Invalid clicks are not junk leads. Google filters clicks its automated protections judge invalid out of your metrics and billing (Google Ads Help). A human student filling in your form is a real click and a real submission; only targeting and forms stop it.
More on waste and search terms: Google Ads AI for lead generation and reducing CPA without buying worse leads.
Fixes on Meta
- Form setting. By default, Meta lead ads are optimised for the volume of leads. A higher-intent setting adds a step where the person reviews and confirms their answers before submitting (Meta for Developers). Expect fewer, more deliberate leads.
- Custom questions. Add one multiple-choice question that screens for fit, such as company size, role or project type. Multiple choice is easier to act on than free text.
- Audience and placement checks. Break down leads by placement and age in Ads Manager and match them to your audit tags. If spam concentrates in one placement, that is evidence; a hunch isn’t.
- CRM optimisation eligibility. Meta can optimise Instant Form campaigns towards a CRM stage through Conversion Leads, but only with at least 200 leads a month, a stage reached by 1–40% of leads within 28 days, and at least daily uploads (Meta for Developers). Below that, the form is your main quality lever.
More on Meta for sales-assisted businesses: Meta Ads AI.
Fixes on LinkedIn
- Audience expansion. It shows ads to members with attributes similar to your target audience, is enabled automatically where available, and is switched off by clearing “Enable Audience Expansion” in the ad set’s Audience section. Its clicks and impressions are blended into your reporting (LinkedIn Help). For tightly defined B2B buyers, check whether it is on before anything else.
- Job function, seniority and company size. LinkedIn maps job titles to function and seniority, and estimates company size mainly from Page admin input. Location can target permanent profile location or recent IP location (LinkedIn Help). Wrong-region junk often comes from recent-location targeting.
- Lead Gen Form questions. Forms allow up to three custom questions, in multiple choice or single-line text, and hidden fields for campaign parameters (LinkedIn Help). Use one custom question to screen and hidden fields to make the audit possible.
- Qualified leads goal. With the Lead Generation objective, LinkedIn can optimise to leads you mark qualified and share within 30 days, recommending five or more within two weeks (LinkedIn Help).
For a UAE industrial trading company whose LinkedIn campaign offered a gated product catalogue, the audit tagged most leads as wrong intent: 150 leads in a month, 9 qualified, 6%. Engineers downloaded the catalogue; few had a live requirement. The fix is not a new audience but a new offer: an RFQ-led form for the in-market segment, keeping the catalogue for retargeting. Re-check the qualified rate on the RFQ cohort once sales has reviewed it. More in LinkedIn Ads AI.
How do you stop training the algorithm on junk?
Short answer
Stop reporting junk as success. Make sure spam, duplicates and test submissions never fire the conversion, then send a later stage, such as qualified lead, that junk can’t reach. Once the platform learns from leads sales accepts, it stops rewarding the cheap submissions that caused the problem, provided the stage has enough volume.
Negatives and form questions remove junk you can see. The bidding signal removes junk you can’t predict in advance. The two work together: without a better signal, the platform keeps searching for new sources of cheap submissions after you block the old ones.
Which stage to send, and whether each platform has enough volume to use it, is covered in which lead stage to optimise for. To compare channels on what matters, use the cost per qualified lead calculator.
Decisions
Which changes to make now, and which to wait on
From the India audit above. Not every finding is a change, and not every change should happen this week.
| Finding | Decision | Action | Why |
|---|---|---|---|
| 70 invalid leads on Google, clustered on job and course searches | Act | Add job, course and salary negatives | Clear pattern, low risk, no learning reset |
| 35 wrong-fit leads on Meta | Act | Add a company-size question to the Instant Form | Screens out small firms without blocking anyone who fits |
| 27 spam leads on Meta | Investigate | Check whether spam fires the conversion; break down by placement | The cause isn’t known yet |
| 40 unreachable leads across both | Review | Agree a first-contact time with sales and re-check in a month | A follow-up question, not a targeting one |
| Meta form just changed | Wait | Hold other Meta edits for two weeks before judging | Let the ad set settle so the form change can be read cleanly |
Zephra labels its own recommendations Act, Review, Investigate or Wait; see how it works.
Failure modes to avoid
- Cutting the channel with the most complaints without comparing its cost per valid lead.
- Blocking free email domains in India or the UAE. Many genuine business buyers, especially in smaller firms and family businesses, use personal addresses.
- Counting unreachable as junk. It hides a follow-up problem and blames ads for it.
- Making five changes at once. When results move, you won’t know which change did it, and several may have reset learning.
- Leaving spam in the conversion count. Every spam submission rewards bidding for finding more.
- Buying a fraud tool before reading the search terms and placement reports. Most B2B junk is real people who aren’t buyers.
India, UAE and US patterns
- India. Broad industrial and software terms attract students, job-seekers and resellers in large numbers. Many buyers prefer WhatsApp or a phone call to email, so a lead that ignores email isn’t necessarily junk. Collect consent on the form under the DPDP Act 2023.
- UAE. Agents and intermediaries enquire on behalf of end buyers, so decide in advance whether an intermediary counts as valid. GCC-wide targeting pulls leads from markets you may not serve. Personal data rules differ between onshore UAE and the DIFC and ADGM free zones.
- US. Bot submissions on website forms are the more common spam pattern. Consent rules for calls and texts affect how quickly and by which channel sales can follow up.
Consent and contact rules by market are in the lead generation compliance guide. Form design is in form and capture optimisation, and follow-up in the sales handoff guide. To keep the diagnosis running every week, set up a sales feedback loop with reason codes.
What to measure next
- Junk share by type, platform and campaign, monthly.
- Cost per valid lead and cost per qualified lead, by platform.
- Median hours to first contact, and the unreachable share.
- Share of conversions sent to each platform that sales later rejected.
Zephra looks for wasted spend in search terms, placements and audiences, and uses lead quality from CRM data or a spreadsheet to judge campaigns on the leads sales accepts, where configured. See lead quality optimisation.
Questions, answered
Junk leads FAQ
01How do I stop students and job-seekers filling in B2B lead forms?
Check the search terms report for job, course, salary and definition queries and add them as negatives. Remove careers-style wording from ads, and add a buyer-type or company question to the form. Then stop counting those submissions as conversions, so bidding learns to avoid them.
02Are junk leads a sales problem or a marketing problem?
Both, and the audit tells you which. Spam, invalid and wrong-fit leads point to targeting, offer or form issues. Leads that are valid but never reached point to follow-up speed, contact data or calling rules. Tag a month of leads by type before changing budgets, so each team fixes its share.
03Is it click fraud?
Sometimes, but less often than teams assume. Google filters clicks it judges invalid out of metrics and billing. Most B2B junk comes from real people who aren’t buyers. Check search terms, placements and submission patterns before concluding it is fraud.
04Should I switch off Meta for B2B lead generation?
Only after comparing its cost per valid or qualified lead with your other channels. Meta can produce more spam and still deliver qualified leads more cheaply. Tighten the form and screen with a question first, then decide on the numbers.
05Should I block free email addresses on B2B forms?
Not in India or the UAE, where many genuine buyers use personal addresses. Screen on company size, role or requirement instead. In markets where business addresses are the norm, test the block on one campaign and compare qualified leads before rolling it out.
06How quickly do junk lead fixes work?
Negative keywords and location changes affect new traffic almost immediately. Form changes show within days but need a matured cohort to judge. Changing the bidding signal takes longest, because the platform has to learn again.
07Do I need a CRM to audit junk leads?
No. A spreadsheet export of leads with their campaign, form type and a junk-type column is enough. The 7-day audit above is designed for exactly that.
08Can Zephra detect junk leads?
Zephra looks for waste in search terms, placements and audiences, and uses lead status and qualification from CRM data or a spreadsheet to assess lead quality by campaign, where configured. It recommends changes labelled Act, Review, Investigate or Wait, within the permissions and approvals you set.
Your accounts
Want the junk diagnosis done on your own accounts?
The free ad account assessment covers wasted spend, campaign structure, performance constraints and measurement gaps across Google, Meta and LinkedIn. 10–15 minutes, no CRM required, no campaign changes.