Suitability and performance are measurement problems. We treat them like such.
PeakSpots classifies social video channel by channel and video by video, then keeps every buy inside the lines — on YouTube today, with Meta and TikTok on the way.
Cheap reach is the most expensive thing you buy.
Unmanaged delivery optimizes for available inventory, not for your brand — and the waste compounds every day a campaign runs.
Part of every unmanaged buy lands where you'd never choose to be.
Kids-adjacent content, unlabeled compilations, made-for-ads channels — cheap in the auction, so the algorithm keeps buying them.
Geo targeting is not language targeting.
In multilingual markets, a real slice of delivery lands on channels your creative doesn't speak.
Account averages hide placement problems.
Suitability is decided channel by channel. Measured any other way, it's an assumption.
Encode. Deploy. Keep current.
From your standard to enforcement in weeks — with every step inspectable.
Encode your standard
Your suitability standard, written as explicit criteria and signed off by your team — not a generic vertical template.
Curated deployment
Inclusion and exclusion lists built to that standard, deployed via the Google Ads and DV360 APIs — in your accounts, visible in your UI.
Continuous refresh
New channels are classified as they appear in delivery, drift triggers alerts, and every change ships versioned and logged.
Metrics that survive scrutiny.
Curation makes raw CPM go up — anyone who says otherwise is hiding something. So we define the metrics that matter and report them in the open.
| Metric | Definition | What it tells you |
|---|---|---|
| Suitability rate | Share of impressions delivered on placements that meet your brand standard. | Whether the buy is inside the lines — the headline number. |
| Wastage rate | Share of spend on placements that fail classification. | The money you're currently paying to be somewhere you'd never choose. |
| Language match rate | Share of impressions on channels in your creative's language. | Whether your message is physically intelligible to the people seeing it. |
| Quality CPM | Total cost ÷ suitable impressions. | The true price of the reach you actually wanted. |
Raw CPM up 6%. Quality CPM down 23%. That's not a trade-off — that's the entire point.
Deployed where you already buy.
No new platform to log into — PeakSpots works through the APIs your team already runs, in your own accounts.
Google Ads & DV360, end to end
Shared negative placement lists and ad-group inclusion criteria in Google Ads; channel and video targeting on DV360 line items. Classification in 38 languages, every verdict with a written rationale.
Feed & Reels suitability
The same classification engine extending to Meta placements — rolling out with design partners.
Creator-level curation
Your standard applied to creator content on TikTok — rolling out with design partners.
Results our clients can defend in the boardroom.
See it on your own campaigns.
A pilot on live spend, measured in the open. Commit when the numbers do.
One engine, three jobs: classify, deploy, keep current.
PeakSpots is not a dashboard you check. It's infrastructure that quietly holds your social video buying to the standard you set.
Every verdict comes with a reason.
The engine scores channels and individual videos against your brand's suitability standard — not a one-size-fits-all vertical preset.
Metadata, transcripts, content
Channel metadata, video transcripts, titles, descriptions, and upload patterns are scored together. Made-for-ads and unlabeled compilation behavior is detected structurally, not just by keyword.
Calibrated to your brand, not a template
Suitability isn't universal — a gaming brand and a pharma brand shouldn't share a list. We encode your standard as explicit criteria your team reviews and signs off on.
Auditable by design
Every INCLUDE and EXCLUDE ships with a written rationale. When your CMO asks why a channel was cut, the answer is one click away — not a shrug from a black box.
Lists as living infrastructure.
Curation only works if it's operationally boring: versioned objects, API-driven updates, and zero manual spreadsheet uploads.
Shared negative placement lists
Exclusions are maintained as persistent shared list objects, updated via the Google Ads API and attached across every campaign that should inherit them. Inclusion targeting is written as placement criteria at the ad-group level, so curated buys stay curated even as campaigns are rebuilt.
Channel & video targeting on line items
The same curated universe is synced into DV360 as YouTube channel and video targeting via API — one standard across both buying doors, refreshed on the same cadence, logged in the same change history.
Nothing leaves your ownership
Lists live in your accounts and remain fully visible in your UI. If we part ways, everything we've built stays with you. Access is scoped, logged, and revocable by you at any time.
Versioned, logged, reversible
Every list update is a versioned change with a diff, a timestamp, and a rationale. Roll back any change. Export the full history whenever your governance team asks.
YouTube changes daily. So do your lists.
New placements, classified as they appear
Channels entering your delivery are scored within the refresh SLA — before they accumulate meaningful spend.
Alerts when suitability slips
If your suitability rate trends below threshold, you hear it from us first — with the placements responsible and the fix already queued.
Suitability metrics beside your media metrics
Suitability rate, wastage rate, language match, and quality CPM reported alongside CPV and view rate — one narrative, no parallel truth.
Questions buyers actually ask.
What about Performance Max and Demand Gen?
Honestly: Google is progressively narrowing placement-level control in those formats. We deploy exclusions to the fullest extent each format allows, report the residual delivery we can't control, and concentrate curation where placement signals remain fully addressable — in-stream and in-feed video campaigns. You'll always know which part of your spend is governable, because we tell you.
Won't curation kill my reach and raise my CPM?
Raw CPM typically rises modestly — you're declining the cheapest, worst inventory. Quality CPM (cost per suitable impression) falls, usually sharply. We put both numbers in every report so the conversation is about the reach you wanted, not the reach you got for free.
How is this different from Google's built-in suitability settings?
Platform-level modes are broad brushes: three settings for every brand on Earth. They don't know your standard, don't classify by language at channel level, and don't produce an auditable rationale. We work with those settings, then do the channel-by-channel work they can't.
Who sees my data?
Your delivery data is used to classify your placements and for nothing else. It's never pooled, resold, or used to benchmark competitors. Account access is read-plus-list-write only, scoped per engagement, and revocable by you.
Put the engine on your campaigns.
Your standard, enforced placement by placement.
For brands that treat suitability as an asset — not a checkbox in campaign settings.
A standard, an engine, and receipts.
Criteria written for your brand
Your suitability standard encoded as explicit criteria, reviewed and signed off by your team. The engine enforces exactly that — nothing generic.
Lists that live in your accounts
Curated lists deployed via the Google Ads and DV360 APIs, visible in your UI, versioned on every change — and yours if we ever part ways.
Numbers that survive review
Suitability rate, wastage rate, language match, and quality CPM beside your media KPIs — every verdict with a written rationale.
Three campaigns where curation earns its keep.
An exclusion floor across everything
A shared exclusion list as the account-wide standard nothing may violate — refreshed continuously as new channels enter delivery.
Language-matched delivery
Inclusion lists per market and creative variant, classified from transcripts — not channel country settings, which are frequently wrong.
Inclusion-only precision
Delivery restricted to classified, approved placements — for launches and categories where one bad placement costs more than marginal reach is worth.
Three steps, no leap of faith.
Define the standard
A working session with your brand and media teams; the criteria go into writing and get signed off.
Pilot on live campaigns
Curated lists deploy to a controlled slice of spend. Results reported side by side, in writing.
Scale on evidence
If the numbers win, we extend the standard across your activity. If they don't, everything we built stays with you.
Bring us your hardest campaign.
We'll put the standard in writing and the numbers on the table.
Numbers first. Names on request.
We anonymize clients on the public site by default — the same discretion we'd extend to you. Full references, with names attached, are available in a sales conversation.
Wastage rate collapsed while view rate rose 19%
Switching curation vendors without a performance dip
Language match across a multilingual launch
Six markets. One standard. Wastage cut by four-fifths.
The situation
A global household and personal care advertiser was running always-on YouTube activity across six European markets. Account-level metrics looked healthy. Placement-level classification told a different story: 14.2% of spend was landing on placements that failed the brand's own suitability standard — kids-adjacent content, unlabeled compilation channels, and channels in languages the creative didn't speak.
What we did
We encoded the brand's suitability standard as explicit classification criteria, reviewed and signed off by the client's brand safety lead. The engine classified every channel with meaningful spend across all six markets, in each market's language. Exclusions deployed as shared negative placement lists via the Google Ads API; curated inclusion lists were written at ad-group level for the always-on lines. Refresh ran continuously, with every change versioned and logged.
The results, after one quarter
"They were the first vendor who showed us exactly what they excluded, and why. Our brand safety review went from an argument to a formality."
VP Media · Global CPG advertiserSwitching curation vendors — without a performance dip.
The situation
A consumer electronics advertiser had run its YouTube curation through a large incumbent vendor for years. Performance had plateaued, and the team wanted verdict-level visibility the incumbent couldn't provide — which channels were excluded, and why.
What we did
The brand's suitability standard was encoded as explicit criteria and signed off by its media team. Curated lists then deployed to a controlled slice of live spend, running in parallel with the incumbent's setup for three weeks — same budgets, same flights, different placements. Every list change was versioned and visible in the brand's own accounts throughout.
The results
"The parallel run made the decision for us. Same budgets, same flights — different placements, better numbers."
Media Director · Consumer electronics brandEleven languages, one launch, and creative that finally landed in the right one.
The situation
An automotive manufacturer was launching a new model across European markets with localized creative in eleven languages. Previous launches had a known but unquantified problem: geo targeting was standing in for language targeting, and in multilingual markets a real share of impressions landed on channels in languages the creative didn't speak. Pre-launch classification put a number on it: a 90.7% language match rate — roughly one impression in eleven effectively unintelligible to its viewer.
What we did
The engine classified channel language across the launch footprint — from transcripts and metadata, not just channel country settings, which are frequently wrong or missing. Language-matched inclusion lists were built per market and per creative variant, deployed through DV360 channel targeting for the programmatic lines and ad-group placement criteria in Google Ads for the auction buys. Refresh ran daily through the launch window as new channels entered delivery.
The results, launch quarter
"We'd been paying to show Dutch creative to French-speaking viewers for years and calling it Benelux reach. Now the number is on a dashboard and it's 99 percent."
Regional Media Director · Automotive OEMNotes from the lab.
What we're learning about suitability, measurement, and the shifting mechanics of YouTube buying — written for practitioners, not for search engines.
Quality CPM: the number that ends the cheap-reach argument
Raw CPM rewards buying badly. Here's the metric that corrects for it — and how to introduce it without a turf war.
Read → Practice · 5 min readInclusion or exclusion: which list, when
The two instruments do different jobs. A practical decision framework for choosing between them, per campaign.
Read →Quality CPM: the number that ends the cheap-reach argument
Every conversation about curation eventually arrives at the same objection: "but the CPM went up." It's true. It will. Declining the cheapest inventory on YouTube — unlabeled compilations, made-for-ads channels, content misaligned with your brand — means the average price of what remains rises. If raw CPM is the metric of record, curation loses the argument before it starts.
The problem is that raw CPM answers a question nobody actually asked. It prices all impressions equally: the one delivered on a channel your brand would proudly sponsor, and the one delivered next to content your legal team would escalate. No CFO would accept "cost per unit of anything" in any other procurement context. Media buying tolerates it out of habit.
The correction
Quality CPM is total cost divided by suitable impressions — the impressions that landed on placements meeting your brand's standard. It reprices the buy in units of the thing you were actually trying to purchase.
A worked example. An uncurated buy delivers 10M impressions at a €4.00 CPM: €40,000. If 13% of that delivery fails your suitability standard, you bought 8.7M suitable impressions — a quality CPM of €4.60. A curated buy delivers 8.5M impressions at €4.25 (raw CPM up 6%, the objection stands): €36,125. But 97.5% of it is suitable — 8.29M suitable impressions at a quality CPM of €4.36. The "more expensive" buy is cheaper, in the only units that matter.
Introducing it without a turf war
Don't replace raw CPM; report beside it. Teams defend metrics they're bonused on, and raw CPM is often contractually embedded. Quality CPM works best introduced as a second line on the same dashboard, with the suitability standard behind it written down and signed off. Once both numbers sit side by side for a quarter, the conversation shifts on its own — because nobody wants to be the person arguing for the cheaper unsuitable impression out loud.
Inclusion or exclusion: which list, when
Curated YouTube buying has two instruments, and they are not interchangeable. Exclusion lists remove known-bad placements from an open buy; inclusion lists restrict delivery to a known-good universe. The choice between them is a choice about how much you trust the open exchange versus how much reach you're willing to trade for certainty.
When exclusion is the right tool
Exclusion suits scale objectives: broad awareness campaigns where reach is the point and the risk profile is moderate. It preserves the algorithm's freedom to optimize while fencing off the placements you've affirmatively rejected. Its weakness is structural — it can only exclude what has already been classified, so new low-quality channels enter delivery before they're caught. That's why refresh cadence matters more than list size: a stale exclusion list is a comfort blanket, not a control.
When inclusion is the right tool
Inclusion suits precision objectives: launches, sensitive categories, language-critical multilingual buys, and any campaign where a single bad placement costs more than the marginal reach is worth. Delivery only ever touches classified, approved placements. The trade is reach and, usually, a higher raw CPM — which is exactly why inclusion campaigns should be measured on quality CPM from day one.
In practice: both, layered
Mature setups run both. A shared exclusion list as the account-wide floor — the standard nothing may violate — with inclusion targeting layered onto the campaigns where precision pays. The floor protects everything by default; the inclusion layer is a per-campaign decision made against explicit reach and cost trade-offs. When the two are versioned and refreshed on the same cadence, the whole structure stays legible to everyone who has to defend it later.
We make every media dollar defensible.
PeakSpots was founded by adtech infrastructure engineers and media practitioners who kept watching the same meeting: a brand asks where its YouTube money actually went, and nobody in the room can answer at the placement level. We built the company that can.
Three rules we don't bend.
Every verdict shows its work
Classification without rationale is just opinion at scale. Every INCLUDE and EXCLUDE we produce carries a written reason your team can read, challenge, and overrule.
Your accounts, your assets
Lists, logs, and change history live in your accounts and stay with you. Access is scoped, logged, and revocable at any time.
Pilots, not pitch decks
Every engagement starts on live campaigns with results reported side by side. We'd rather lose to good numbers than win on a deck.
Small, senior, and unusually close to the metal.
We're a team of engineers and media operators who have shipped campaign infrastructure across Google Ads, DV360, and a dozen other platforms — the people who know what an API can actually do, not just what a sales deck says it can. We publish our methodology to clients, we take the hard platform questions head-on, and we hire people who find that fun.
Open roles: classification engineer, media operations lead →
| Founded | 2024 |
| Focus | Brand suitability for social video |
| Coverage | YouTube (Google Ads · DV360) · Meta & TikTok expanding |
| Languages classified | 38 |
| Operating regions | Europe & North America |
Judge us on your own campaigns.
A measured pilot beats a promise. We'll bring the numbers.
Talk to sales.
Tell us a little about your social video activity and we'll come back within one business day.
Thanks — we'll come back to you within one business day.
A 30-minute intro call
Your markets, formats, and standard. We'll tell you plainly whether we're a fit.
Define the standard
Your suitability criteria, drafted with your team and signed off in writing.
Pilot on live campaigns
Curated lists on a controlled slice of spend. Commit when the numbers land.
Include your markets and formats and we'll come to the first call prepared.
Privacy policy
This page is a structural placeholder. Before launch it should describe: what data PeakSpots collects through this website and through client engagements; the legal bases for processing under GDPR; how client campaign data is used (classification of the client's own placements and nothing else), retained, and deleted; sub-processors; international transfer mechanisms; and how to contact the data protection lead.
Terms of service
This page is a structural placeholder. Before launch it should cover: scope of services; account access terms (scoped, logged, revocable); ownership of lists and change history (the client's); confidentiality and non-pooling of client data; the disclosed-economics commitment as a contractual term; liability; and governing law.