The Native Ad Supply Chain: Nodes, Evidence and Limits
Follow one captured native ad from a publisher's "you may also like" box to its landing page, reading the visible widget, SSP, exchange and click-tracker clues, and what captured URLs cannot show.

Open the network tab on almost any news article and scroll down to the "You may also like" boxes. Those few thumbnails are native ads. Behind them sits a chain of middlemen most marketers never look at: a publisher monetizing the slot, an SSP packaging it, an exchange running the auction, a winning buyer, a stack of click trackers, and several redirects later, an advertiser's landing page.
Markup and captured URLs can reveal some serving and tracking participants. They do not necessarily identify every auction party, the payer or where fees ultimately go.
This guide explains common serving and destination nodes using illustrative paths and historical capture statistics. It does not provide a complete verified transaction trace for every node.
What is the native ad supply chain?#
The native-ad supply chain includes parties involved in selling inventory, selecting and serving an ad, measuring delivery and resolving destinations. These processes are related but are not necessarily one visible URL chain. Some participants cannot be inferred from browser evidence alone.
If you want the wider regulatory and library context for why this data is visible at all, start with our pillar on ad transparency. This piece goes straight to the plumbing underneath it.
The anatomy of one impression#
The table separates common roles and potential evidence. Do not read it as a universal sequence of click redirects.
| Node | What it does | What you can read from it |
|---|---|---|
| Publisher widget | Renders the "recommended" units on the page | Widget vendor, publisher ID, placement/slot ID |
| SSP / supply platform | Monetizes the slot, holds the auction inventory | SSP domain, account/publisher mapping |
| Exchange / marketplace | Matches bids to the slot in real time | Exchange host, auction macros, deal IDs |
| Demand buyer | Wins the slot and serves the creative | Campaign/creative IDs in the served URL |
| Click tracker(s) | Logs and redirects the click | Tracker domains, click macros, affiliate IDs |
| Landing page / pre-lander | Where the user actually arrives | Real advertiser domain, offer, funnel type |
The point that trips people up: you do not need to click a live ad to see most of this. The technical identifiers (widget scripts, impression beacons, slot IDs, the click-macro host) are already present in the rendered page and the ad-serving URLs at the moment of capture. That is the whole premise of a native ad library. Record the impression as it renders, classify the intermediaries, and resolve the destination, without firing a real click that would charge the advertiser or pollute their analytics.
Distinguish the displayed brand, destination domain, intermediary and legal advertiser. The final domain is evidence about the destination, not necessarily the entity paying for the ad.
Node 1: the publisher widget#
The chain starts where the user sees it. Native ads are not banner slots. They are recommendation units injected by a widget, a small script the publisher embeds that calls out to a content-discovery network and fills the slot with a mix of editorial and paid recommendations. This is the native ad widget layer.
From the widget alone you can usually pull three things. The widget vendor, because the script domain tells you which content-discovery network powers the unit. The publisher account, a property ID baked into the widget config that maps the slot back to the site monetizing it. And the placement ID, the specific position on the page, which matters because the same advertiser often buys different placements at very different prices.
The cited June 2026 sample contains more Taboola records than Outbrain or MGID records. Unequal collection coverage prevents converting that into the probability of encountering each network on a US or UK news site.

A screenshot alone cannot establish the supply chain. Compare tools by the actual evidence fields they return; this article does not substantiate what most competing tools capture.
Node 2: the SSP and the exchange#
In most cases the widget vendor is also acting as the SSP, the supply-side platform that holds the publisher's inventory and exposes it for buying. In native, the SSP and the content-discovery network are frequently the same company wearing two hats. One face to the publisher (monetize my recommendation slots), one face to advertisers (buy this inventory).
Auction-related hosts and parameters may help classify a path, but their presence does not prove an open auction, and their absence does not prove a direct deal. Resolve that question with documented transaction evidence where available.
The number and commercial roles of intermediaries vary. A visible redirect count does not reveal every party or its fee.
A note on the players, because outdated claims circulate. The native landscape consolidated hard in 2025. Outbrain acquired Teads in early 2025 and rebranded the combined company as Teads, while Taboola built its performance platform Realize on top of its existing publisher network. The Taboola and Outbrain merger that made headlines in 2019 was actually called off in 2020, so anyone telling you "they merged" is wrong. The practical takeaway is simple: which corporate entity owns an SSP shifts over time, so classify by the technical fingerprint in the URL, not by a brand name you happen to remember.
Node 3: the demand buyer and the served creative#
Captured identifiers can help group records when their meaning and stability are known. Avoid assuming that an identifier uniquely represents one advertiser or campaign across every source.
Observation span and the number of captured publishers describe sample presence. They do not establish continuous delivery, spend, profit or campaign scale.
The cited snapshot has an observation window of roughly 28 days. A first-to-last span at that limit does not establish daily observations; daily-presence claims require the underlying dated records.

A quick caveat to keep honest. That 28-day figure is the depth of our observation window, not the true lifetime of the ad. The general industry lore about "90-day winners" that never die is a separate claim, and it is not something our index measures. What we can say from our own data is that the creatives still running at day 28 are vastly outnumbered by the ones that vanished inside a week.
A time series can show where and when the collection observed a creative. Even sightings across many sites do not establish that the advertiser made money.
Node 4: the click tracker chain#
Here native diverges most sharply from search or social, and here the real money hides. A native click rarely goes straight to the advertiser. It passes through a click tracker, an intermediary URL whose job is to log the click, attach identifiers, and 302-redirect onward. In affiliate-heavy native, that is not one redirect but a chain.
First comes the network click tracker, the SSP's own redirector carrying click macros. Then, optionally, an affiliate-network tracker, for example a CPA network logging the click for attribution. Sometimes a cloaking or rotation layer that splits traffic between offers or geos. Then the advertiser's own analytics or redirect before the page finally renders.
An identified affiliate-tracking host can support a hypothesis about the path. It does not by itself prove the commercial relationship or classify every participant's business model.

Preserve the captured path and distinguish visible branding from destination and payer identity. Additional records may be required to identify who purchased the placement.
Inspect captured metadata and stored traces without clicking live ads. Do not assume that requesting a tracker server-side is harmless: it can still log an event. Use the repository's approved capture and resolution workflow.
Node 5: the landing page and the pre-lander#
The final node is the destination, and it answers the questions the rest of the chain only hints at.
A captured destination can reveal an offer, pre-lander or checkout. Its domain does not necessarily identify the payer, advertiser or legal seller; verify those roles separately.
A quiz-style headline suggests a destination to investigate, but does not prove the actual path. Inspect a stored capture before describing a quiz or offer sequence.

Combine the available artifacts and mark missing or inferred nodes. Capturing a destination does not prove that every supply-chain participant is known.
Putting the whole trace together#
Walk the chain end to end on one impression and a flat thumbnail becomes a story.
Illustrative path: a publisher renders a widget; captured hosts suggest serving infrastructure; creative identifiers group some observations; a stored destination trace includes a tracker and an advertorial. Each attribution needs its own artifact. This example is not a measured transaction and does not prove an auction type, payer or profitable campaign.
That is the difference between knowing an ad exists and understanding the business behind it. It is also why the supply-chain view underpins every other competitive workflow: reverse-engineering a competitor's funnel, sourcing winning angles, spotting arbitrage, auditing your own placement quality. For a primer on the broader tooling category that makes this readable, see our explainer on the ad transparency tool category.
How to read your own traces#
Inspect page markup, recorded requests and existing stored traces without triggering a live ad click. Mark uncertain hosts and missing nodes explicitly; do not follow arbitrary tracking links as a supposedly harmless investigation step.
That is what OpenAdLibrary is built for. As an open, low-cost native ad spy tool, it captures live public native ads, records the real creative at full quality, classifies the supply chain node by node, and follows each click to the landing page, without clicking live ads. The free tier shows the 20 newest ads with no card, and paid access is a flat $29.99/mo against the $80 to $400/mo charged by legacy rivals. Start free and run your first trace today.







