Every query runs against
data nobody else has.
ThoughtLeaders has spent seven years building the only complete record of who advertises on YouTube and what actually worked. Competitors can guess at these numbers from public scrapes. Only we have measured them, classified them, and connected them across seven years of real outcomes.
The numbers, live from our database.
The Brand Extractor.
We process every brand mention in every video we track — across transcripts, descriptions, and on-screen content — and classify each one as sponsored or organic. This is the engine behind every CLI command and the most defensible thing we own.
Why sponsored vs. organic matters.
Most data providers can tell you that a brand was mentioned. We can tell you whether they paid to be mentioned. That one bit of metadata is the difference between a database and a marketplace. It powers every repeat-sponsorship calculation, every category analysis, every CPM benchmark the CLI gives you.
Why repeat sponsorships are the real signal.
Click-through rates lie. Vanity metrics lie. Impression counts lie. Repeat purchases don't. When a brand comes back to the same channel for a second deal, it's the most honest success signal in the industry — they paid, they waited, they measured, and they decided to do it again.
$ tl brands --category vpn --window 12m --md brand unique_channels total_deals repeat_deals avg_price ────────── ─────────────── ─────────── ──────────── ───────── Surfshark 15 58 43 ← elite $775 NordVPN 9 9 0 ← no tail $831 ProtonVPN 1 5 4 $3,000 CyberGhost 2 4 2 $1,325 ↳ Read: Surfshark is running the strongest retention program in the category (43 repeat deals across 15 channels). NordVPN has 9 deals across 9 channels and zero repeats — every deal is a one-shot. Two similar brands. Opposite results. Only the CLI can show you this.
61,008 brand profiles.
Our brands table tracks every brand we've observed advertising on YouTube, with their complete marketing strategy rolled up across every tracked channel. Every sponsorship. Every organic mention. Every repeat. Every dollar.
$ tl brands Manscaped --md total_sponsorships: 55 first_deal: 2020-10-28 most_recent: 2025-08-19 unique_channels: 30 repeat_rate: 45.5% ← strong retention signal avg_deal_price: $1,583 recent_deals: • ScreenCrush 2025-08-19 $2,800 • ScreenCrush 2025-08-19 $2,800 (repeat) • Capt. Panda 2025-04-07 $600 • Forrest Kritzer 2025-03-21 $450 • Theme Park Express 2025-02-25 $340 ↳ Manscaped is active, returning to proven channels, and hitting a 45.5% repeat rate — almost half of their relationships have resulted in a second or third deal.
Three data sources. One CLI.
Postgres
Deals, brands, channels, pipeline state, users. Every write lands here first. Queried by tl deals, tl brands, tl channels.
Elasticsearch
Video transcripts, Brand Extractor output, current channel/video metrics, demographics. Queried by tl uploads.
Firebolt
7.4B video snapshots + 1.1B channel snapshots. View curves. Growth trajectories. Evergreen calculations. Queried by tl snapshots.
The CLI abstracts all three. Your agent doesn't need to know which store a field lives in — it just asks.