Push-format terms, defined the way the math treats them

Technology that delivers ad creative through paths an ad blocker doesn’t recognise — typically by serving the ad payload from a first-party domain or rendering it through a script chain the blocker’s filter list hasn’t catalogued. Reduces blocked-impression rate but doesn’t eliminate it. Measure the lift on your own panel before believing the marketing number.

Systems that filter bot, proxy, datacenter, and incentivised traffic before the impression is billed. Quality varies — most panel anti-fraud scores are posterior probabilities, not verdicts. Overlay your own sub-source-level fingerprint check; the long-tail bad actors learn to defeat aggregated filters faster than the filters update.

Cost per Action — what the advertiser pays only when a defined conversion fires (signup, deposit, install). The action definition is the whole game. Same offer, different action threshold, different CPA. Never compare CPAs across networks without normalising the action definition and the attribution window.

Smart-bidding mode where the network’s bidder adjusts CPM bids to hit a target cost-per-acquisition. Needs ≥200 conversions per sub-source cohort before the model outperforms rule-based bidding — below that threshold the bidder overfits on noise. Specify the target, supply the volume, then wait.

Cost per Click — what the advertiser pays each time a user clicks an ad. Easy to game on the publisher side; treat raw CPC as a noisy signal until you’ve filtered out invalid clicks at your end. The honest metric is CPC weighted by day-7 CR.

Cost per Mille — what the advertiser pays per 1,000 ad impressions. The rate-card CPM and the actual clearing CPM are different numbers — the gap is where rate-card-vs-actuals leaks live. Trust the panel-reported clearing CPM, but triangulate it against your own conversion-per-dollar across the same period.

Click-Through Rate — clicks divided by impressions, within a defined window. Direct cross-network comparison requires the same window, same attribution method, same anti-fraud filter. Mobidea’s panel CTRs and Adsterra’s CTRs are not directly comparable without normalising for those three. CTR is a lagging indicator of nothing useful on its own; pair with day-7 CR.

Demand-Side Platform — buyer-side software that places programmatic bids on inventory across multiple SSPs. The DSP’s bidding model is only as good as the conversion signal you feed it. Garbage S2S postback in, garbage CPA-goal out. Wire the conversion API before you trust the bidder.

Effective CPM — revenue per 1,000 impressions normalised across pricing models (CPM, CPC, CPA all rolled into one comparable number). The publisher-side optimisation target. The distribution across sub-sources is usually bimodal, not normal — top decile delivers 55–65% of the revenue. Aggregated averages hide both ends.

Share of ad requests that get an ad served back, vs ad requests that come back empty. Networks publish ceiling fill rates; actual fill varies sharply by GEO, vertical, and time-of-day. A 96% fill rate in Tier-1 EU and a 42% fill rate in a Tier-3 GEO can live on the same panel — the average is meaningless.

Geographic targeting parameter. Country-level by default; city- or region-level on networks that expose the granularity. Every meaningful claim about CTR, CR, or CPM is conditional on GEO. “Push CTR is 3%” without a GEO is a meaningless number.

Push-shaped creative rendered inside the publisher page (no browser subscription required), iOS-eligible. Captures the iOS Safari audience classic web push can’t reach — in Tier-1 push, iOS is 30–41% of impressions depending on vertical. Day-7 CR decay is flatter than classic push because the audience isn’t a fixed subscriber base.

Full-screen creative displayed during a content transition. Highest CTR of any format (3.8–6.1% on Tier-1 iGaming in my Q2 2024 sample, n=1.8M) and the steepest day-7 CR decay. Frequency caps slow audience fatigue, they don’t fix it.

Ads styled to match the surrounding publisher content — typography, colour, layout inherited from the page. Higher CTR than display (1.9–2.7× in my Q1 2024 parallel-buy, n=1.4M), lower CR-given-click (roughly half of display). End-to-end CR per impression wins narrowly when publisher quality is high; collapses fast when it isn’t.

Payout cycle: the publisher is paid 14 days after the close of the earnings period. The cash-flow gap matters more for publishers and small operators than for mid-to-large advertisers. Pair the payout cycle with the minimum-payout threshold; a Net-14 with a $5 minimum is operationally faster than a Net-7 with a $100 minimum.

Payout cycle: the publisher is paid 7 days after the close of the earnings period. The market standard for the Cyprus-cluster networks and the operator-friendly default. Net-7 with a low minimum ($5–$20) is the publisher-side cash-flow target; advertisers don’t feel the difference except at small spend.

Full-page ad opened in a new tab behind the active window, triggered by a user click on the publisher page. Highest eCPMs in display advertising (Tier-1 iGaming clears $1.40–$2.80 median per my Q3 2023 sample, n=2.1M). Roughly half the day-1 conversions don’t survive a 7-day attribution window.

Return on Ad Spend — revenue generated per dollar spent on ads, in a defined attribution window. Day-7 ROAS and day-30 ROAS for the same campaign are different numbers; the gap is the LTV tail. Stating a ROAS without naming the window is uninterpretable.

Real-Time Bidding — programmatic auction where bids are placed per impression in tens of milliseconds. The clearing price is set by the second-highest bid (second-price auction in most implementations). The bidder’s signal quality dominates the outcome; without clean conversion data feeding back, RTB optimisation degrades to noise.

Server-to-server conversion tracking. The tracker fires a URL from your backend to the network when a conversion happens, mapping it back to the original click via a click ID. The honest way to attribute — pixel-based tracking loses ~12–24% of conversions to ad blockers and Safari ITP per my measurements; S2S doesn’t.

Automated bid adjustment that lowers your CPM bid when competition is weak (so you don’t overpay) and raises it when competition is strong (so you don’t get outbid). Convergence to a stable per-sub-source bid typically needs 180–240 cumulative conversions. Below that threshold the optimiser overfits on noise.

A single URL that routes traffic to the best-matching offer based on GEO, device, and other signals. Useful for beginners — the routing layer handles offer selection. The trade-off: smartlink abstracts away sub-source-level data, so sophisticated buyers wanting publisher-quality scoring at their end lose the granularity they need.

Supply-Side Platform — seller-side software publishers use to sell inventory to multiple DSPs through programmatic auctions. The SSP’s job is to maximise eCPM for the publisher; the bidder logic on the demand side is the buyer’s problem. Conflicts of interest between SSP and DSP behaviour are well-documented — don’t assume the auction is neutral.

High-value English-speaking markets — US, UK, CA, AU. Highest CPMs, deepest competition, most demanding fraud profile. PropellerAds and Adsterra have the deepest Tier-1 push inventory in my measurement; pricing at the top of the rate card matches the LTV math for iGaming, finance, and Tier-1 SaaS.

Mid-tier markets — most of continental EU, the larger LATAM markets (Brazil, Mexico, Argentina), parts of MENA. Mid CPMs, mid competition. Tier-2 popunder on Adsterra cleared roughly 28% cheaper than PropellerAds in my Q3 2023 parallel-buy, with comparable day-7 CR. The cost-per-conversion arithmetic often favours Tier-2 for mid-budget scaling.

Emerging markets — SEA (Indonesia, Vietnam, Thailand), parts of Africa, smaller LATAM. Lower CPMs, higher impression volume, easier scaling on raw reach. The fraud profile is the catch — Tier-3 sub-source quality is more bimodal than Tier-1, and the bot floor is higher. HilltopAds has the largest SEA publisher concentration in my testing.

Video Ad Serving Template — IAB specification that lets video players request, display, and report on ads. Versions 2.0, 3.0, and 4.0 are in production use. The completion event (and the q1/midpoint/q3 quartile events) are the cleanest fraud signal in the video format — wire all of them into your reporting, not just the impression event.

Category of offer — iGaming, VPN, dating, mobile-CPI, sweepstakes, utility, finance. The vertical determines the LTV math, the attribution window, the publisher-network fit, and which networks are worth testing first. CR and ROAS are uninterpretable without the vertical attached.