First, measure quality from what happened to the leads
The cost per lead in Ads Manager counts form submissions; it says nothing about whether those people could be served. Before changing anything in TikTok, pull the last few weeks of TikTok leads from the CRM with their outcome: reached or not, qualified or not, booked or sold. Use the stage definitions the business already has, or agree them now with sales; the guide to MQL, SQL and lead stages sets out how a Sales-Qualified LeadA lead that sales has reviewed and accepted as a genuine opportunity to pursue. is defined and handed over. Without an agreed definition of quality, the argument about lead quality cannot be settled.
Then split the outcomes by campaign, by conversion location (instant form or website form), by form version and by ad. A KPI tree shows where the leak sits, from leads to reached leads to qualified leads to sales, and the Lead-to-Customer RateShare of leads created in a period that became paying customers. at the end is the figure the business lives on. Leads that cannot be reached point to the form or the follow-up. Leads that are reached but cannot be served point to the offer or the form's questions. A count in Ads Manager far above the leads the CRM received points to the event. A quality that fell after a change to the campaign points to what delivery is learning from. Compare each split with the business's own earlier months, not with a rate from another advertiser.
Two related problems have their own pages. Taps that never become any form submission are diagnosed in TikTok clicks but no conversions, and a website lead event that never arrives at all in TikTok pixel not firing. This page starts once leads are arriving and the trouble is what they are worth.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Many leads cannot be served: wrong area, no budget, no real need | Instant form questions; CRM disqualification reasons | An instant form with nothing that screens anyone out | Add a qualifying question about location, timing, need or budget |
| Leads are real people but did not know what they asked for | The video's opening and closing lines against the form's offer | A video offer that does not say who it is for | State the offer, who it suits and where, early and on screen |
| Ads Manager counts far more website leads than the CRM received | When the SubmitForm event fires, and whether it is deduplicated | A website form event that fires on the button | Fire the event on successful submission, deduplicated with the server |
| Many leads never answer; the earliest calls do better than later ones | Time from submission to first contact, by lead | Leads left waiting in Ads Manager | Sync leads to the CRM and contact them quickly |
| Quality fell as volume rose; delivery favours the quickest form-fillers | What the ad group optimises for and what the CRM reports back | Delivery learning from form counts with no CRM outcome | Report qualification back from the CRM and judge on it |
| Website leads are either too few to learn from or too loose to use | Landing page form fields and completion rate | A landing page form that asks too little or too much | Rebalance the fields: enough to qualify, no more |
Fix, verify and prevent each cause
An instant form with no screening question
The TikTok Ads setup for lead generation explains why feed leads arrive with less intent than search leads: the person was watching something else and enquired on impulse. An instant formAn ad format that collects contact details inside the platform, prefilled from the user's profile. that asks only for a name and a phone number, pre-filled where TikTok holds the details, makes that impulse effortless and tells the business nothing about whether it can help.
Fix: add one or two short qualifying questions, multiple choice where possible, on whatever decides whether the business can serve the lead: location, timing, the size of the job or a budget range.
Verify: in the CRM, the share of leads disqualified for the reason the question covers falls, and qualified leads hold or rise even if total leads fall.
Prevent: write the qualifying criteria with sales before the form is built, and treat any form without a screening question as unfinished.
An offer the video never explains
On TikTok the video does part of the qualifying, as the lead generation setup puts it. A video that never says who the offer is for, where it is available or what happens after the form invites people who cannot use it, and they fill the form anyway because it is easy. The content formats and hooks guide covers how to make the opening seconds carry the point.
Fix: state the offer, who it suits and where it applies early in the video, out loud and on screen, and say what happens after the form.
Verify: leads from the revised ad are reached and qualify at a higher rate than leads from the old one in the same campaign, read in the CRM.
Prevent: review every lead ad's script against the qualifying criteria before production, not after the first poor batch.
A website event that counts taps
For website leads, the ad group optimises towards the SubmitForm or Contact event, and the pixel and Events API tutorial is explicit that the event should fire at the real moment of submission, not on the button that starts it, with a server copy sharing an event identifier so the two are deduplicated. An event that fires on the button counts people who tapped and abandoned, or who hit a validation error, and teaches delivery to find more of them.
Fix: move the event to the confirmed submission, such as the thank-you state after the form succeeds, and deduplicate it with the server event.
Verify: with test submissions, one event arrives per real submission, and Ads Manager's daily lead count moves into line with the leads the CRM received.
Prevent: test the lead event on every form release, alongside the pixel checks in the tutorial.
Leads that wait until they forget
The lead generation setup warns that the impulse behind a feed lead fades faster than a search lead's, and that a TikTok lead left for a day is often a lead lost. Instant form leads that sit in Ads Manager until someone downloads them reach a person who no longer remembers the ad. Lead Response TimeAverage time between a new lead arriving and the first contact from sales. is part of lead quality, not separate from it.
Fix: use the download or CRM connection the help centre documents to put each lead in front of a person quickly, and agree who calls, how fast and how many times. The inbound lead handling playbook sets out routing, response and feedback, and the lead routing QA checklist tests that TikTok leads reach an owner.
Verify: the time from submission to first contact falls, and the reached share rises in the CRM.
Prevent: run a test lead through the route after every change to the form or the CRM, and report response time beside lead counts.
Delivery that learns only from form counts
Delivery optimises towards the result it is given. When that result is a form submission and nothing else, it finds the people most likely to fill forms, who are not always the people most likely to buy. The TikTok measurement hub notes that lead events can be connected from a CRM so that a lead's later qualification is reported back, and the help centre documents the current options.
Fix: record qualification and sale against each TikTok lead in the CRM, connect those outcomes back to TikTok where the integration allows, and judge campaigns on cost per qualified lead; the guide to connecting the CRM to ads and analytics explains click identifiers and the offline conversionUploading outcomes recorded outside the website, such as closed deals, to an ad platform. routes in general.
Verify: the qualified outcomes appear in Ads Manager against the right campaigns, and cost per qualified lead, worked out with the CPA calculator, is read each week.
Prevent: make the CRM outcome the campaign's indicator before launch, so form counts are never the target.
A website form out of balance
A landing page form with only an email field produces leads too loose to use; one with many fields produces too few leads for delivery to learn from and loses good prospects to effort. The form design guide covers the trade between fields, qualification and completion. On TikTok the lead generation setup tips the balance towards one or two more fields than a search landing page would use, because the people filtered out were unlikely to become customers.
Fix: keep the fields that decide qualification, drop the ones sales never uses, and use clear choices rather than open text on a phone.
Verify: the form's completion rate, read with the conversion rate calculator, and the CRM's qualified share move together in the direction the change intended.
Prevent: change one field at a time, and keep a record of each version with its dates so outcomes can be matched to it.
Prevention: a weekly quality read shared with sales
Lead quality drifts quietly, so read it on a schedule. Once a week, take the TikTok leads from the CRM with their outcomes, split by campaign, form version and ad, and put the reached, qualified and sold counts beside the lead count and cost per lead from Ads Manager. Share the read with whoever calls the leads and ask for the reasons behind the disqualifications, because those reasons are the diagnosis table filled in by the people closest to it. Record every change to forms, questions, offers, events and routing with its date, so a change in quality can be matched to its cause. Accept that a small share of fake or mistaken entries comes with instant forms, and watch only for a share that moves against the business's own history.
The same symptom on other platforms has its own pages: Meta lead ads bringing low-quality leads and LinkedIn Lead Gen Form leads of low quality. The TikTok Ads troubleshooting hub lists the other TikTok diagnoses.
Worked example
Wrennolf Kitchens judges TikTok on booked surveys
Wrennolf Kitchens, an invented kitchen-fitting firm working within one county, runs a TikTok Lead generation campaign with an instant form asking for name and phone number, at £1,200 a month. In March the form brings 240 leads at £5 each. The CRM shows the office reached 92, and 11 booked a survey visit: £109 per booked survey. The disqualification reasons say most of the rest lived outside the area served or wanted a single cupboard door replaced.
In April the firm makes four changes. The form gains a postcode question and a budget-range question with 4 choices. The video's opening line names the area served and says the offer is a free design survey for a full kitchen. Leads sync to the CRM, and the office calls each one the same working day. Booked surveys are recorded in the CRM and reported back to TikTok against the campaign. In May the form brings 120 leads at £10 each; the office reaches 98 and 30 book a survey, at £40 per booked survey. Ads Manager's cost per lead doubled, and the firm now judges the campaign on cost per booked survey alone.
Common mistakes
- Judging the campaign on cost per lead. It rewards the easiest form and the most casual enquiry; judge on qualified or sold leads from the CRM.
- Changing the audience first. Read the CRM outcomes by form and ad before touching targeting; the form, the offer and the follow-up are cheaper to fix.
- Firing the lead event on the button. Count confirmed submissions, deduplicated with the server.
- Downloading leads every few days. The impulse behind a feed lead fades; route leads to a person quickly.
- Chasing every fake entry. A small share comes with instant forms; diagnose only a share that moves against the business's own history.
- Leaving sales out of the read. The people who call the leads know why they failed; their reasons are the quickest route to the cause.
