Skip to main content
A play tells you someone watched. It does not tell you who sent them, or whether that person converts. The Sources tab does both: it classifies every session to one traffic platform, then ranks affiliates, domains, and pages by revenue instead of volume. Open a video and click Sources. Two more pieces of the same picture live on General: the Traffic Quality card (how much of this traffic is real) and the Custom Data breakdown (any URL parameter you send, broken into rows).

​
How a session gets classified

Every session gets exactly one traffic_source value, decided once on the first event and carried through the whole session, including the conversion. TrackPlay checks seven signals in a fixed order and stops at the first one that answers. A later, weaker signal never overrides an earlier, stronger one.
A declared channel and a delivery network are not competing signals, they refine each other. A gclid proves the click came from Google Ads and nothing more. The {network} parameter proves it was delivered on YouTube specifically. Both facts can be true at once: a campaign labeled youtube can still show ad_network = video_partners, meaning it ran on a Google video-partner site, not on youtube.com itself. TrackPlay keeps both rather than collapsing them into one.

​
Direct is a fact, Unknown is a gap

Once a session clears the classifier with no referrer, no UTM, no click id, no affiliate id, and no share link, TrackPlay checks whether it actually captured attribution data for that session at all. If it did, and every signal came back genuinely empty, the session is labeled Direct. That is a real claim: this viewer typed your URL, used a bookmark, or opened a message with no tracking parameters. If the session carries no attribution data to check in the first place, the label is Unknown, never Direct. TrackPlay does not guess.
Affiliate hoplink redirect chains and in-app browsers on Facebook and TikTok drop document.referrer on the vast majority of clicks that pass through them. An empty referrer on affiliate-driven traffic is expected, not a tracking failure.
If the reported referrer is the same domain as the page the video is embedded on, TrackPlay treats that as internal navigation, not an acquisition source, and falls through to the next signal instead of reporting a referral from yourself.
Email is the one channel where the browser will not tell TrackPlay the truth on its own, so it is worth two minutes of setup. Add utm_medium=email to every link you send. That is the whole requirement:
email, e-mail, mail, and newsletter are all accepted. Add utm_source and utm_campaign if you want the broadcast broken out by name in the UTM Source and UTM Campaign tabs:
Most email platforms can append this automatically to every link in a campaign. In ActiveCampaign, GoHighLevel, Klaviyo and Mailchimp it is the campaign-level link-tracking or UTM setting, so you set it once rather than editing links by hand.

​
Why the referrer is not enough on its own

TrackPlay does recognize webmail referrers, and a click from Gmail, Outlook, Yahoo Mail, Proton Mail, Zoho, AOL, Superhuman, or a self-hosted Roundcube or webmail. host is classified as email even with no tag. That check deliberately runs before the search-engine check, because mail.google.com would otherwise match Google and your newsletter would be credited to organic search. But it only covers people reading email in a browser tab. Two large groups send no referrer at all:
Outlook for Windows and Mac, Apple Mail, and Thunderbird hand the link to the operating system, which opens the browser with no referrer. The session looks Direct.
The Gmail and Outlook apps and Apple Mail on iOS open links in an in-app browser or the system browser. In most cases no referrer survives, so those sessions also look Direct.
Gmail routes clicks through a Google redirect. When a referrer does arrive it is a Google host, which is indistinguishable from search unless the classifier checks the mail hosts first. TrackPlay does. Other analytics tools frequently do not, which is why email traffic so often shows up as Google in them.
Tag the links. The referrer check is a safety net for untagged sends, not a substitute for tagging, and it cannot see the desktop and mobile app readers who are usually the majority of your list.

​
Click id parameters captured

These platform click ids are recognized at step 3 of the classifier. Each one also identifies the ad platform for the Ad Platform column, independent of which specific surface delivered the click.
sccid is Snapchat’s own ScCid parameter, matched case-insensitively. irclickid, cjevent, ps_xid, and gsxid all identify traffic as affiliate-sourced rather than naming a specific ad platform, since they come from affiliate networks (Impact, CJ, PartnerStack) rather than paid platforms.

​
The tables

Sources opens with a tabbed breakdown, then four ranked reports. All five read the same economics for every row: sessions, play rate, average watch time, conversions, conversion rate, revenue, and four money columns.
Traffic Sources and Top Pages are not the same report asking the same question twice. The referring-domain report is empty on almost all of this traffic, because hoplink chains and in-app browsers strip the referrer before it ever reaches TrackPlay. Top Pages groups by the page your player actually loaded on instead, which is populated on effectively every session, so it is the report that answers “which of my pages converts” when Top Referring Domains cannot.

​
Ad Platform, Ad Network, and Ad Channel are three different questions

The Traffic Source dimension is always populated. The other three platform dimensions answer narrower questions and stay empty until you set them up:

​
Revenue columns

ROAS reads as a dash, not 0×, when no payout value has been captured for a source. A zero would claim the source lost money. A dash means TrackPlay does not know what it cost.
Every table is ranked by revenue first, then conversions, then sessions, and every column is sortable. Rows carrying no classification stay in the table labeled Unknown rather than being dropped, so the sessions column still adds up to the video’s real traffic.

​
Traffic Quality

The Traffic Quality card, on the video’s General tab, splits sessions into four categories and breaks them down by country.
Best-effort classification computed at ingest from the visitor IP. Bot (automation or crawler) and datacenter (hosting or VPS) are reliable. VPN detection needs the IP-reputation add-on and is currently rarely available, so most VPN traffic appears as datacenter or human. Anything not positively flagged counts as human.
This is a fail-open design: an unclassified session defaults to human rather than to bot, so Traffic Quality never inflates a fraud number it cannot back up.

​
Custom Data breakdown

Also on the video’s General tab. TrackPlay captures every URL query parameter your landing page carries into custom_data, whether or not it maps to a first-class column. The Custom Data breakdown lets you pick any captured key (utm_source, affid, a plan name, anything) and see one row per value, with plays, unique plays, play rate, conversions, and revenue. The top 50 values are shown individually, with an “Other” row summing the rest. Click a row to apply it as a filter across the rest of the tab. A gclid proves a click came from Google Ads. It cannot tell you whether that click was served on YouTube, on Search, or on the Display network. Without more information, YouTube Ads spend and Search spend are the same number in every report. Google has the answer. It does not send it unless you ask for it.
1

Open Ad Tracking

Go to Integrations → Ad Tracking. The setup is one screen, and it applies to every video in the workspace at once.
2

Copy the final URL suffix

Copy this string exactly. Do not translate the {'{'}...{'}'} tokens: they are Google’s own ValueTrack macros, and Google expands them at click time.
3

Paste it into Google Ads

Settings → Account settings → Tracking → Final URL suffix. You can also set it per campaign instead of account-wide.
4

Check the status badge

The Ad Tracking screen shows whether the suffix is expanding. It reads your last 30 days of Google Ads traffic and reports one of three states, refreshed every 5 minutes: Active (a percentage of Google sessions carrying the network parameter), Not detected (Google traffic exists but none of it carries the parameter), or No Google Ads traffic. A workspace running only Meta correctly reads “no traffic,” never a broken-looking 0%.
5

Optional: declare the channel directly

Append &ef_channel={_ef_channel} to the suffix, then add a custom parameter named _ef_channel on each campaign, with a value like youtube, google_search, google_display, performance_max, or demand_gen. Only add this if you actually define the _ef_channel custom parameter on the campaign. Without it, Google leaves the macro unexpanded and you get no channel data at all.
Two real limits, and no workaround exists for either. Performance Max only ever reports that a click was Performance Max. It cannot say whether the surface was YouTube, Search, Display, Discover, Gmail, or Maps, because Google does not expose that information at the network-parameter level. Demand Gen does not support the network parameter at all, in any form. For both, the declared-channel parameter above is the only signal that answers the question, and it is only as accurate as what you typed into the campaign.

​
Refresh and caching

A date range that ends in the past is immutable, so TrackPlay caches it hard: 24 hours. A date range that touches today is still accumulating events and refreshes every 10 seconds. The Ad Tracking coverage badge is cached separately, for 5 minutes, since it answers a setup question rather than a live report.