A Spark Ad is a TikTok ad that runs from a creator's account rather than yours. The creator generates an authorisation code for a specific post, hands it over, and you put budget behind something that already exists. Meta's version is called Partnership Ads, and the agency word for the whole arrangement is whitelisting. The mechanics differ. The measurement problem is identical.
The appeal is obvious enough. The post has already proven it can hold attention organically, it arrives wearing a face instead of a logo, and the engagement it collects lands back on the original post rather than on a dead ad copy. TikTok's own materials claim Spark Ads deliver meaningfully higher completion and engagement rates than standard in-feed ads. Worth knowing who is making that claim, and that they sell the format.
Where the measurement breaks
Three things go wrong at once, and none of them is anybody being careless.
The ad runs from the creator's handle, so nothing about the click says your brand ran it. Discovery is frequently not a click at all: somebody watches a video on Tuesday, thinks about it, and searches your name on Friday. And the reporting you get back describes the ad rather than what happened after it. Views, completion rate, cost per thousand. Distribution numbers.
- viewsWhat the platform reports back to the brand
- engagementWhat the creator can show in a rate card
- conversionsWhat the finance team asked about
- nobodyWho currently connects the first to the third
The asymmetry is the point. Both parties measure honestly and neither can answer the question the budget was spent on.
So the category has settled on measuring reach, because reach is what the pipes emit. Everyone involved knows the number they actually want is cost per customer per creator. That is the acknowledged gap, and it has stayed open mostly because the tooling to close it did not exist.
What closing it takes
Three pieces, and the third is the one people skip.
First, spend has to arrive at ad level rather than campaign level. A campaign total tells you the campaign worked. It cannot tell you which of the six creators in it paid for itself, and the creator is the unit you actually renew or drop.
Second, the creator's identity has to come from the platform. Not from an ad title that happens to start with an at sign. We are strict about this to the point of being annoying, because a per-creator report that names the wrong person is worse than one that names nobody, and the person it names is real. The full rule set is on how we measure.
Third, you have to ask the buyer. Creator discovery leaks out of click tracking constantly, and no amount of parameter hygiene fixes a purchase that started with a video somebody watched without tapping anything. A one-question survey on the confirmation page catches what the click never saw.
The tagging that survives
None of the above works if the traffic arrives untagged, and this is the part you can fix this afternoon. Put the creator in utm_content, consistently, on every amplified post. Same casing every time. Same handle format every time.
Inconsistent casing is the single most common practical cause of unattributed paid traffic we see in the data, and it costs nothing to prevent and a great deal to repair after the fact. Whatever measures your funnel afterwards, including ours, can only group what the URL actually said.
The term itself is worth understanding properly before you spend against it. We wrote a plain definition: what is paid amplification.