The form works. The phone rings. The CRM fills up with new contacts every week, and the marketing report looks healthy: cost per lead is down, volume is up. Then you talk to the sales team and hear the other half of the story. The number is wrong, the person never answers, they wanted something you do not sell, they were “just looking”, or they were a student, a supplier, a recruiter. The pipeline is full of names and empty of money.
We hear this complaint almost as often as its opposite, and it is worse in one way: it costs twice. You pay for the lead, then you pay a salesperson to discover it was never one. The good news is that “junk leads” is not one problem but a short list of mechanical causes, and each leaves a different fingerprint in your own data.
Why it happens
The campaign is optimising for the wrong event. Every advertising platform optimises towards whatever you tell it is a conversion. If that event is “form submitted”, the algorithm will find the people most likely to submit a form, and those are not the same people as the ones most likely to buy. The platform is doing exactly what it was asked: over weeks it drifts towards the cheapest submitters it can find, and quality erodes while the cost-per-lead chart keeps improving. The report and the sales team are both telling the truth about different things.
The offer attracts the wrong person. “Free consultation”, “get a quote in one minute”, “download the guide”: the lower the price of admission, the more people walk through the door who had no intention of buying. A promise that is vague about who it is for collects everyone, like a shop with a giveaway at the entrance: the queue is real, the buyers are not in it.
The form asks nothing that would qualify. A form with only a name and a phone number cannot distinguish a customer from a curious visitor, so the first time anyone learns the difference is on a sales call. Every qualifying question left out of the form is one a salesperson has to ask instead, at far higher cost.
The lead goes cold before anyone calls. A person who fills in a form is interested at that moment. By the next morning they have spoken to two competitors, or forgotten, or solved the problem another way. When the call comes a day later, they sound like a bad lead. They were a good lead that was handled slowly. In our experience this cause is the one most often misdiagnosed as “traffic quality”.
Sales and marketing define a lead differently. Marketing counts a submission; sales counts someone with budget, a timeline and a decision-maker. Nobody wrote either definition down, so both teams measure success honestly and disagree completely. Everything between the two definitions lands in the “junk” bucket, with no owner.
Nothing flows back. The platform that brought the lead never hears what happened to it. It keeps learning from submissions, the only signal it receives, and cannot improve on a dimension it cannot see. This is less a separate failure than the reason the others persist.
How to tell which one is yours
An evening with your CRM export, your ad accounts and an honest salesperson is enough for a first diagnosis; no paid tools needed.
Read the last few dozen “junk” leads one by one. Not the totals: the actual records. Sort them into rough piles: wrong need, wrong geography, no budget, unreachable, spam or bot, competitor or recruiter, “we called too late”. The biggest pile is your first cause. If “unreachable” dominates, check how long each waited before the first call; slow response looks identical to a bad lead from the outside.
Trace each pile back to its source. Add the campaign, the ad and the landing page to the same list. If one campaign produces most of the wrong-need pile, the fault is that campaign’s offer or targeting, not the site. If junk is spread evenly across every source, the cause is downstream: the form, the response time or the definition of a lead.
Check what the campaigns are optimising for. Open each active campaign and look at its conversion action. If it is a form submission, a page view or a button click, the platform has never been told what a good lead looks like.
Time the response. Take a handful of recent leads and compare the submission timestamp with the first human contact in the CRM. If that gap is measured in hours or days rather than minutes, you have found at least part of the problem, whatever the traffic quality.
Ask sales and marketing for their definition of a lead, one sentence each. If the sentences differ, that gap is where your junk lives.
The fix, in order
Start with the cheapest changes and the ones fully under your control; cost and lead time rise as you go down the list.
-
Agree one written definition of a qualified lead. Budget range, geography, need, timeline, decision-maker: whichever of these matter to you, write them down and have both teams sign off. Every stage below depends on this sentence existing.
-
Answer faster. Route new enquiries to a phone, not an inbox. Set a first-contact target measured in minutes during working hours, and automate an immediate acknowledgement outside them. This step often removes much of the “unreachable” pile on its own, at no media cost.
-
Add qualifying questions to the form. Two or three that map directly onto your definition: what they need, roughly when, in what country. Ask them in plain language, explain why, and make them required. Fewer people will submit; more of them will be worth a call. Rebuilding forms so they qualify without frightening genuine buyers is a routine part of how we build websites and landing pages.
-
Rewrite the offer for the buyer, not the browser. Say plainly who the service is for, what it costs to start, and what you do not do. A precise promise repels the wrong visitor before they ever fill in the form; it is the cheapest filter you have and the one most often skipped in favour of “more leads”.
-
Record what happens to every lead, and mark the qualified ones. Add a status field in the CRM that sales actually updates: qualified, disqualified with a reason, won, lost. Without it none of the later steps have anything to learn from. Making this reliable rather than optional is the core of our CRM and sales-automation work, not for reporting’s sake but because the next step needs the data.
-
Feed qualification back to the platforms. Every major ad platform can import offline conversions or accept a server-side event when a lead is marked qualified in the CRM. Switch the campaigns to optimise for that event instead of the submission, and from then on the algorithm learns from the salesperson’s judgement rather than from who was most willing to type a phone number. This is the change that turns the trend, and the one we treat as non-negotiable in our paid advertising set-ups. It only works once steps one and five are done.
-
Cut the sources that never convert. With the feedback loop in place you will see which campaigns, placements and audiences produce qualified leads and which produce volume, and can move budget accordingly. Do this last: before the loop exists you would be cutting on gut feeling.
What to measure
Stop reporting cost per lead on its own; pair it with cost per qualified lead and watch the second one. Track the share of new leads that reach “qualified” within a week, per source, and the median time from submission to first contact. Watch these on the campaigns you changed, not on the account total, and give each change enough leads before judging it. A month in which lead volume drops and qualified volume holds is a good month, even though the old report will call it a bad one.
Where we come in
Most of the list above is work an owner and a sales manager can do together in a few evenings, and the written definition of a lead costs nothing but an argument. Where it usually stalls is the plumbing: the CRM field nobody fills in, the offline-conversion upload nobody owns, the form that needs rebuilding rather than editing. That plumbing is what we do. If you want us to look at your lead flow end to end, describe where it leaks and we will start there.
