First, read the audience's source, rule, date and owner against the activity behind it
Write one line about the audience. Its source: the website pixel, a customer file, or engagement with the brand's videos and profile (app audiences depend on app event measurement and sit outside this page). Its rule: the event or pages, the watch depth or interaction, or the exported segment. Its date: the day it was created and, for a file, the day of the export. Its owner: the Business Center or personal login that holds it and the ad accounts it has been shared with. Beneath it, write the activity the audience should contain, each from its own source: the views and interactions Ads Manager reports for the videos feeding an engagement list, the visits to the rule's pages in analytics, and the customers in the business's own records. The distance between the two lines is what needs explaining.
Two readings come before any diagnosis. TikTok sets the size an audience must reach before it can be used for targeting or as the seed of a Lookalike AudienceAudience similar to a seed list., and publishes those minimums, with the period each source keeps people, in its Ads Help Center; they change, and this page quotes none of them. A list created recently may also still be filling, and an engagement list holds only what has happened since its creation, so judge it again after its flight has run, before rewriting the rule.
Several patterns look like a shrunken audience without being a fault.
| What the audience shows | Why it may be expected | What to check in its place |
|---|---|---|
| A list created at a flight's launch looks thin after its opening days | The flight is young, and an engagement list has collected only since its creation | The same list again once the flight has run |
| A deep-watch list sits well below the list built from brief views | A deeper depth is a stricter rule on purpose | Whether the deep list can still feed the ad group meant to use it |
| A website audience below the sessions analytics shows for the same pages | A visitor who refused advertising consent joins no audience, while analytics may still count that visit | Whether visitors who accepted are being recorded |
| A customer-file audience well below the rows uploaded | Each platform matches only part of a file, and there is no rate to aim at | The last upload of this segment to TikTok |
| A buyer seed too small for a lookalike in a young shop | A seed should come from the customers the business wants more of | Whether a wider group of buyers, not every visitor, reaches the help centre's minimum |
Some lists are small and correct: a shop with few buyers, a rung finer than its traffic can supply. Such a list needs a different job rather than a repair, and the timing cause below sets out what the audiences guide advises. A neighbouring fault looks similar from a distance: an audience large enough to use whose ad group still barely spends. That is a narrow audience leaving delivery little room, not a list too small to use, and the TikTok delivery diagnosis for ads that spend little or nothing reads it from the ad group's status and delivery history.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| An engagement list from a finished flight holds a fraction of its viewers | The list's creation date against the flight's dates | Audience defined after the flight it was meant to collect | Create the next flight's lists before launch; merge the thin rung meanwhile |
| A website audience stays thin while site visits are healthy | The rule's event in Events Manager: whether it arrives, and from which pages | Rule built on an event few visitors trigger, or one that never arrives | Repair the event, or base the list on an earlier event |
| A viewer or visitor list far below the flight's views or clicks | The rule's depth against how far viewers watch; its pages against the live ads' destinations | Rule narrower than the traffic | Add a shallower list beside the deep one; point the page rule at the pages in use |
| A viewer list grows slowly at a sensible depth | Watch time and completion for the ads feeding it | Creative that does not hold viewers to the chosen depth | Improve the videos through creative tests, not by lowering the depth |
| A website audience far below the visits although the rule's event fires | A clean-browser test of both consent choices; the market's consent rate | Consent and blockers keeping visitors out | Repair wiring that loses accepted visitors; treat refusals as a floor |
| A customer-file audience far below the rows it held | The file's identifiers and export date; the last upload of this segment | Customer file with few identifiers or stale contacts | Add identifiers the business may lawfully use; replace with a fresh export |
| The ad account cannot choose the audience, or uses a recent copy | Which Business Center or login holds the list, and which ad accounts it is shared with | Audience never shared to the ad account, or rebuilt without its history | Share it from the business's Business Center, or rebuild there and bridge the gap |
Fix, verify and prevent each cause
An audience defined after the flight it was meant to collect
On timing, the TikTok audiences guide is clear about engagement lists: they start gathering people on the day they are created, and a list created after a flight ends holds nobody from it. The same gap opens part-way through: a viewer list created in a flight's second week holds the viewers of its later days only. Whether a list from another source reaches back before its creation is for the help centre to say; this page does not. Check: set each audience's creation date beside the start and end of the flight that was supposed to feed it.
Fix: the viewers the list missed cannot be added later. Keep the list, create the next flight's audiences before its first impression, and meanwhile give the thin rung a job it can do: the guide advises merging a rung that cannot feed its own ad group into the broader rung above, or using the list as an exclusion instead.
Verify: the next flight's lists grow from its first day, in step with the whole flight's views rather than its last few days.
Prevent: put audience creation on the launch list and name each list after the flight that feeds it, so a missing list shows before any impression is bought.
A website audience built on an event few visitors trigger, or an event that never arrives
A website list contains only the visitors the TikTok pixelAn ad platform's browser code that reports page views and events on a site back to the platform. recorded doing what its rule asks. A rule on a late step in the journey, such as entering payment details, describes a small group by construction; a rule on an event that stopped firing, or was never set up, describes nobody. Check: open Events Manager on the pixel the audience uses and read whether the rule's event arrives, from which pages, and how its count compares with visits to those pages in analytics. The pixel and Events API tutorial lists the standard events and the uses each suits, audiences among them, so an event the setup never included shows as a gap there.
Fix: where the event is absent or arrives without its data, repair it at the source: the TikTok pixel diagnosis traces where the event is lost on its way from the page to the ad account. Where the event works and few visitors reach it, as general practice base the list on an earlier event, such as a product view, and keep the deep event for an exclusion or a narrower rung.
Verify: in the test events view, the event arrives once per action from the pages the rule names, and the list grows from that day. Visitors from the weeks the event was silent were never recorded, so no rebuild brings them back.
Prevent: rerun the test for this event in the test events view after every release that touches the site or the container, as the prevention routine below sets out.
A rule narrower than the traffic
Two kinds of rule leave a list short of the activity it was meant to catch. The first is a watch depth few viewers reach. The consideration setup sets that depth when it defines the viewer lists, and treats a deep list as a stricter selection than one of brief viewers. The second is a page rule that no longer matches where the ads send people. The traffic setup scopes a visitor list to the landing pages a campaign uses; when the ads move to a new page, a rule still aimed at the old one catches only stragglers. Check: set the depth the rule demands beside how far viewers actually watch the videos feeding it, and the rule's pages beside the destination of every live ad.
Fix: keep the deep list and, as the consideration setup recommends, build a shallower one beside it so the follow-up campaign can show which converts; point the page rule at the pages the ads use now. As general practice, correct a page rule before lowering a depth, since a stale page condition is an accident while a depth was a choice.
Verify: the new or corrected list collects in step with the flight's views or the landing pages' visits.
Prevent: whenever an ad's destination changes, review the visitor lists scoped to the old page in the same edit.
Creative that does not hold viewers to the chosen watch depth
When the depth is sensible and the list still grows slowly, the videos feeding it lose people before that point. Check: read watch time and completion through each video for every ad feeding the list, against the account's own earlier ads for the same objective, and find where viewers leave: in the opening, or in the body after it.
Fix: treat it as a creative problem. The guide to testing TikTok ad creative adds new concepts to the live ad group in rotation, works on the idea first and the opening second, and uses organic posts as a cheap first filter before media money is spent; follow that order rather than recutting one video.
Verify: the new ads keep viewers further into the video than the ones they replaced, and the list's growth picks up with them.
Prevent: choose videos for a consideration flight that already held viewers as organic posts, which the consideration setup linked under the previous cause names as the first candidates for paid spend.
Consent and blockers keeping visitors out of website audiences
Lists built from a website hold only the visitors whose choice allowed advertising use, and the privacy and consent guide points out that the size of such a list reflects the banner as much as the demand behind it. A visitor who refuses belongs in no website audience, and blocked requests and short-lived cookies also remove some who accepted. Neither is a fault in the audience. The one to look for is a consent setup that loses visitors who accepted, the browser fault the pixel diagnosis linked above works through. Check: in a clean browser, accept consent on a landing page and confirm the pixel records the visit; refuse and confirm nothing is sent; then set the consent rate the business records for the market beside the audience.
Fix: repair any wiring that loses accepted visitors, following the consent step of the pixel tutorial; treat refusals as a floor; and where exclusions or seeds need the customers the pixel missed, cover them with a customer file, subject to the consent and purpose check in the next cause.
Verify: in the clean-browser test, accepted sessions fire the pixel at once and refused sessions send nothing, and the audience grows from the day of the repair.
Prevent: test both consent choices after any edit to the banner, the consent tool or the container, mark each change as a break when comparing audience sizes over time, and compare markets only with their consent rates beside them.
A customer file with few identifiers or stale contacts
A file reaches customers the pixel never saw, but only the share TikTok can match. Results vary by platform and by the identifiers the file carries, and the cross-platform customer list guide is plain that there is no rate to aim at. A file of email addresses alone carries one identifier; a file exported long ago holds people who have moved address, lapsed or left, and it stays as uploaded until a new file replaces it. The Match RateShare of the records in an uploaded customer or account list that an ad platform matched to its own users or companies. is more than a size: a customer who is not matched can be neither excluded nor part of a seed. Check: note the file's identifiers, export date and segment, and set this upload's result beside the last upload of this segment to TikTok, never beside another platform's figure.
Fix: first confirm with whoever owns privacy that TikTok advertising is a permitted use of these records. HashingConverting an identifier such as an email address into a fixed string with a one-way function, so platforms can match records without the raw value. protects the file on its way to the platform without making it anonymous, because a hashed identifier is still Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier.. Then add the other identifiers the business holds and may lawfully use, of the kinds TikTok accepts as set out in its help centre, and replace the old file with a fresh export rather than appending to it, so people who left the segment leave the list.
Verify: the replacement reads better than the last upload of this segment did.
Prevent: keep each file in the customer list register with its segment, identifiers, export date, replacement date and owner. Advertisers who also upload lists to LinkedIn will find the same split between a weak file and later narrowing in the diagnosis for a LinkedIn list with a low match rate, on a network where work addresses behave differently.
An audience not shared to the ad account running the campaign, or rebuilt without its history
The TikTok Business Center tutorial keeps audiences in the business's own Business Center beside the pixel and the ad accounts, has the business decide which of its people can build audiences and which can pass them to ad accounts, and warns that a list created under one person's login goes when that login goes, with every visitor and engager it had gathered. Two faults follow. A list built in another ad account, an agency's Business Center or a personal login may never have been shared to the ad account the campaign runs in, so the campaign cannot use it, or someone builds a fresh copy there. A copy may not carry over what the original gathered: an engagement copy collects only from its own creation, and a website copy built on a new pixel holds nothing from before that pixel existed. Check: find which Business Center or login holds the audience the campaign uses, which ad accounts it is shared with, and whether it is the original or a later copy.
Fix: where the original sits in the business's Business Center, share it to the ad account running the campaign. Where it sits elsewhere, as general practice ask the holder to bring it under the business; failing that, rebuild it in the business's own Business Center, accept the gap, and merge the rebuilt rung into the one above until it fills.
Verify: a new ad group in the campaign's ad account can choose the audience, and the setup record shows it held in the business's Business Center.
Prevent: create every audience there from the start, with the people allowed to share it named in the setup record.
Prevention: audiences defined on launch day and kept in Business Center
A list's size is shaped before its flight starts: by the day it was created, the rule it was given, the videos and pages that feed it and the place it was built. The routine therefore belongs to the launch. On the day a flight goes live, before its first impression is bought:
- Create every engagement list the flight should feed, at the planned depth, with a shallower list beside the deep one wherever the deep one may run short.
- Scope each visitor list to the pages the flight's ads use, noting which list depends on which page.
- Build each audience in the business's Business Center, then let the people the business has named pass it to each ad account that will run it.
- Name each list with its source, rule, the flight or export that feeds it and its date, so a late or stale list shows in every ad group using it.
- Test the events behind each website list in the test events view, with consent accepted and refused.
- Replace customer files from a fresh export after the consent and purpose check, on the dates the register sets.
While a flight runs, read each list's growth weekly beside the views and visits feeding it; a list that stops growing while its flight carries on points at the rule, the videos or the event. After any release that touches the site, the container or the banner, rerun the browser test for the events website lists are built on, and read the counts by route as the TikTok Events API diagnosis does, since a single release can silence the browser events and their server copies together.
Plan the ladder against the budget as well as the traffic. Each audience is a stage, and the marketing funnel framework asks that every stage have an entry rule that can be counted; a rung whose rule admits too few people is a stage the business cannot yet fill. The TikTok campaign structure guide gives each ad group one hypothesis and warns that a budget spread across many ad groups leaves them learning, and a rung with too few people is the same split in another form. Keep the rungs to what the traffic and budget can feed, and record every list in the Business Center setup record with its creation date, its owner and each ad account it is shared to.
The same reading of source, rule, date and owner works on other platforms, though their settings differ. The Meta diagnosis of an undersized custom audience covers the symptom with Meta's own settings, and the diagnosis for a Google Analytics audience that is not populating covers lists built in analytics and shared to Google Ads.
A harmonica shop's viewer rung that never filled (illustrative)
Snopwexil Harmonicas finds why its viewer rung never filled
Snopwexil Harmonicas, an invented online seller of harmonicas and beginner lesson packs, ran a 4-week Video views flight on demonstration videos, planning a viewer rung and a visitor rung for a traffic campaign to follow. The owner created the viewer audience on day 8, at a very deep watch depth, and the visitor audience on day 15, scoped to the lesson-pack page the first ads linked to. By the end, Ads Manager showed 58,000 video views, yet the viewer audience showed as too small to use, the visitor audience was little better, and a customer file of 2,300 email addresses, exported 14 months earlier, had matched poorly. The source, rule, date and owner lines showed why. The first 7 days, with 21,000 of the views, came before the viewer list existed; about 1 viewer in 9 reached the depth the rule demanded, though watch time held up against the shop's earlier ads; from week 2 every ad pointed at a new beginner bundle page the visitor rule did not include; and the file held email addresses alone, many from a mailing list nobody had cleaned. For the next flight the owner created two engagement audiences on its first day, one deep and one shallower, both named after the flight; pointed the visitor rule at the bundle page; confirmed with the person responsible for privacy that the data could be used, then replaced the file with a fresh export of 1,900 current customers with phone numbers added; merged the viewer rung into prospecting until the new lists filled; and kept every audience in the business's Business Center, shared to the shop's one ad account. Three weeks into that flight both viewer lists and the visitor list showed as usable, the file matched better than the first upload, and the viewer rung returned as its own ad group in week 4.
Common mistakes
- Creating the viewer list after the results come in. It holds only what followed its creation.
- Rewriting a list that is merely young. Read it again once the flight it collects from has run.
- Lowering the depth to fill a list the videos cannot fill. Fix the videos first.
- Leaving a page rule on a retired landing page. Review visitor lists whenever an ad's destination changes.
- Reading every website gap as consent. Confirm the rule's event arrives first; consent and blockers explain only what remains.
- Topping up a stale customer file. Swap in a fresh export, so people who left the segment drop out of the list.
- Uploading before the consent and purpose check. A hashed identifier is still personal data.
- Building audiences in a personal login or an agency's account. They leave with that person or agency.
- Forcing spend through a list too small to use. Merge it into the rung above or keep it as an exclusion.
The TikTok Ads troubleshooting collection lists the delivery, measurement and creative diagnoses that sit beside this one.
