At a glance
The alert that fires when TikTok’s conversion tracking has silently stopped reporting while ads keep spending. Tags fail quietly, the TikTok Pixel and the Events API stop sending purchase events, and TikTok’s smart-bidding throttles delivery on a signal it can no longer measure. It watches for clicks continuing to rise while confirmed conversions collapse below the level you would expect, or for the Pixel and Events API event-match rate dropping below a healthy threshold. TikTok’s iOS attribution is even more brittle than Meta’s, so the goal is to catch the break inside the same iOS attribution window, before a day of spend reports as wasted and the auction starts mis-bidding on a broken signal. Caveat: a tracking break and a genuine demand drop look identical on a ROAS card alone, so this card exists to tell them apart.
| What it watches | The health of TikTok conversion tracking: the browser-side Pixel events, the server-side Events API events, and the match rate between them. It compares the live conversion rate against the recent expected baseline for the same click volume. |
| Trigger condition | Clicks rising while combined Pixel and Events API conversions run below roughly half of expected for more than about 30 minutes, OR the event_id match rate between Pixel and Events API falling below roughly 80%. Both thresholds are tunable per profile in the Sensitivity tab. |
| Why match rate matters | A healthy dual setup deduplicates Pixel and Events API events by a shared event_id. When that match rate falls, either the Pixel or the server side has stopped sending, or the IDs no longer line up, which is the early signature of a half-broken implementation. |
| TikTok’s Events API | TikTok’s server-to-server equivalent of Meta’s CAPI, formerly called Server-to-Server. It carries the conversions the browser Pixel cannot see, especially on iOS. If it breaks, iOS conversions vanish first. |
| iOS 14.5+ ATT impact | TikTok’s iOS attribution is more fragile than Meta’s because TikTok relies more heavily on probabilistic and modeled signals. On any iOS share, the server-side Events API path carries conversions the Pixel can no longer see, so an Events API break is the most damaging failure mode. |
| Chart type | Alert list. Each row names the affected pixel or dataset, the expected vs observed conversion level, and the match rate. |
| What it is not | Not a demand metric. A real fall in sales also lowers conversions; this card is calibrated to fire on the click-rising-conversions-falling pattern that signals measurement failure rather than lost demand. |
| Currency | Not currency-denominated; the unit is a count and a rate. |
| Time window | RT (real-time, evaluated continuously on the standard refresh). |
| Roles | owner, marketing, engineering |
Calculation
Calculated automatically from your TikTok Ads data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading.Worked example
A US apparel brand on Shopify with the official TikTok Shopify app providing the Events API. A theme deploy at 02:00 on 14 Mar 26 removed a script include that the TikTok Pixel depended on. All figures are illustrative.| Time window | Clicks | Expected conversions | Observed conversions | Event match rate | State |
|---|---|---|---|---|---|
| 13 Mar, all day | 4,200 | ~168 | 171 | 92% | Healthy |
| 14 Mar, 00:00 to 02:00 | 360 | ~14 | 13 | 91% | Healthy |
| 14 Mar, 02:00 to 02:35 | 280 | ~11 | 2 | 41% | Alert fires |
| 14 Mar, 02:35 onward | rising | rising | near zero | falling | Alert sustained |
- The alert fires at 02:35, about 35 minutes after the break. Clicks kept rising on the normal overnight pattern (TikTok delivers heavily late at night for Gen-Z audiences), but observed conversions fell well below half of expected and the match rate dropped from 91% to 41%. That divergence is the tracking-break signature.
- The match-rate collapse is the tell. The Events API server events kept arriving (Shopify’s server-side path was untouched), but the browser Pixel stopped, so the deduplicated pairs fell apart. A pure demand drop would lower conversions on both paths together and would not crater the match rate.
- Every minute of delay costs measured ROAS and bidding quality. TikTok keeps spending and optimising on a signal it can no longer see. Because TikTok’s auction reacts fast, within hours it starts mis-allocating budget toward audiences that merely look like they stopped converting.
- The fix is an engineering action, not a marketing one. Roll back the theme deploy or re-add the Pixel script include, then confirm in TikTok Events Manager that both Pixel and Events API events return and deduplicate.
- Recovery is not instant. Once tracking is restored, TikTok’s smart-bidding and the modeled-conversion layer take time to recalibrate. Expect reported conversions to undershoot for a short period after the fix before they normalise.
- Clicks up + conversions near zero + match rate collapsed = tracking break. Engineering fix.
- Clicks down + conversions down + match rate steady = genuine demand drop, not this card.
- Match rate falling slowly over days = creeping implementation drift (consent changes, script reorder). Audit before it becomes a hard break.
Sibling cards merchants should reference together
| Card | Why pair it with Pixel/Events-API Tracking Broken | What the combination tells you |
|---|---|---|
| Clicks vs Conversions | The underlying divergence this alert formalises. | Clicks rising while conversions flatten is the raw shape of a tracking break. |
| ROAS Dropped Below Threshold | A tracking break shows up there as a false ROAS drop. | Confirms the ROAS drop is measurement, not demand, so you do not cut spend in error. |
| Conversion Drop Alert | The conversion-side view of the same event. | Conversions falling while clicks hold separates a tag failure from a demand fall. |
| Conversions Trend | The daily series for context. | A vertical cliff (not a gradual slope) points to a tracking break. |
| GA4 GTM Tag-Fire Failure Alert | The independent tag-health signal from analytics. | If GA4 tags broke at the same moment, the cause is a site-wide deploy, not a TikTok-only issue. |
| GA4 GTM Tag-Fire Rate | Continuous tag-fire health. | A falling fire rate corroborates a measurement problem across channels. |
| Google Ads Conversion Tracking Broken Alert | The same alert on paid search. | Both firing together means a store-wide tracking break, not a platform-specific one. |
Reconciling against TikTok Ads Manager
Where to look in TikTok Ads Manager: The authoritative place to confirm a break is TikTok Ads Manager > Assets > Events > Web Events > your pixel, which shows live event volume, and the Test Events tool, which shows whether Pixel and Events API events are arriving and deduplicating. The Diagnostics view flags dropped or malformed events. Ads Manager itself shows the downstream effect (conversions and ROAS falling) but not the cause; Events Manager shows the cause. Other TikTok views that look related but are not:- Attribution settings: configures the click and view windows. Useful context but not a live health monitor.
- Event Match Quality score: a periodic quality grade, not a real-time break detector. It updates too slowly to catch a same-window failure.
| Reason | Direction | Why |
|---|---|---|
| Detection lead time | Ours fires earlier | TikTok’s quality scores update on a slow cadence; this card watches the live click-to-conversion divergence and fires within minutes. |
| Expected-baseline basis | Variable | The expected conversion level is derived from the account’s own recent ratio. A genuine shift in audience mix can move the baseline, which is why the click-rising condition is required before firing. |
| Match-rate sampling | Marginal | TikTok’s match-quality figures are sampled and lagged; the live match rate this card watches is closer to real time and can read lower or higher in the moment. |
| Time zone | Boundary only | TikTok uses ad-account time zone; this card uses UTC. Relevant only at day boundaries, not for a within-window break. |
| Card | Expected relationship | What causes legitimate divergence |
|---|---|---|
google_analytics.gtm_tag_fire_failure_alert | If both fire at once, a site-wide deploy broke tags | If only TikTok fires, the break is in the Pixel or Events API path specifically, not the whole tag layer. |
shopify.total_revenue | Real orders continue during a tracking break | If Shopify orders keep flowing while TikTok conversions crater, the conversions are real but unmeasured: a tracking problem, confirmed. |
google_ads.conversion_tracking_broken_alert | Both firing points to a shared root cause | TikTok alone points to a TikTok-specific Pixel/Events-API failure. |
Known limitations / merchant FAQs
How is this different from a normal conversion drop? A demand drop lowers conversions on every tracking path at once and leaves the Pixel-to-Events-API match rate intact. A tracking break lowers measured conversions while clicks keep rising and usually collapses the match rate. This card is tuned to the second pattern. If conversions and clicks fall together, look at Conversion Drop Alert instead. What usually causes a break? The common culprits are a theme or template deploy that removes or reorders the Pixel script, a consent-banner change that blocks the Pixel until acceptance, a checkout-flow change that skips the purchase event, an expired or rotated Events API access token, or an Events API gateway outage. After any of these, expect this card to be the first to react. Why does it watch theevent_id match rate?
Because a half-broken setup is the hardest to spot manually. If the Pixel breaks but the Events API keeps firing (or vice versa), total conversions may only dip rather than crater, but the deduplicated match rate falls sharply. The match-rate condition catches the half-broken case that a raw conversion count would miss.
Why is TikTok tracking more fragile than Meta’s?
TikTok’s measurement stack is younger and leans more on probabilistic and modeled signals, especially on iOS. The Events API is essential rather than optional on TikTok, because a Pixel-only setup loses a larger share of conversions to ATT than the equivalent Meta setup would. That is precisely why catching a break fast matters more here.
Can a deploy trip this even though nothing is actually broken?
Occasionally. A consent-mode change that delays the Pixel until after acceptance can briefly look like a break before events catch up. The 30-minute sustained condition is there to filter transient blips, but a confirmed fire still deserves a Test Events check.
How fast can I expect to be alerted?
Within minutes of a sustained divergence, well inside the iOS attribution window, which is the whole point: catching it before a full day of spend reports as wasted and the auction starts mis-bidding.
After I fix it, why are conversions still low for a while?
Once tracking is restored, TikTok’s smart-bidding and modeled-conversion layer take time to recalibrate. Reported conversions can undershoot for a short period after the fix. Confirm the fix in Events Manager > Test Events rather than waiting on the Ads Manager number to fully normalise.
Can I tune the thresholds?
Yes. The expected-conversion ratio and the match-rate floor are both configurable per profile in the Sensitivity tab. Tighten them on high-spend accounts where even a short break is costly; loosen them on low-volume accounts where small samples are noisy.