L'ad-tech, défini simplement
Définitions courtes et sans opinion des termes que vous rencontrerez dans nos bancs d'essai. Lorsqu'un terme a une nuance régionale, propre au format ou liée au cas d'usage, nous la signalons en ligne — moyenner une définition entre les marchés, c'est ainsi que les expressions « standard de l'industrie » perdent leur sens.

Ce glossaire couvre le vocabulaire ad-tech que nous utilisons dans nos bancs d'essai, nos annexes méthodologiques et nos comparatifs en face à face. Mis à jour en continu lorsqu'un fournisseur redéfinit un terme d'une manière qui change matériellement la façon dont il doit être lu sur une grille tarifaire.
A
- Anti-AdBlock
Technology that routes around browser-level and OS-level ad blocking so publishers retain monetisable impressions on blocked inventory. The catch is that the disclosure norms on anti-AdBlock are looser than on the rest of ad tech; reputable networks differ on how aggressive their implementations are, and the trade-press coverage has historically been quieter on this category than the structural importance warrants.
- Anti-fraud
Systems that block bot, proxy, datacenter and incentivised traffic before billing. The category where every network claims multi-layer ML scoring and where the real test is whether the network credits back fraud caught by the advertiser’s tracker after the campaign ships. In my Q3 2024 testing across six networks, two credited without negotiation, three negotiated, and one declined.
C
- CPA
Cost per Action — the price advertisers pay only when a defined conversion (signup, deposit, install) occurs. The pricing model that aligns network and advertiser incentives most cleanly, with the caveat that the “action” definition has to match the postback the advertiser fires. Mismatched definitions are where reconciliation disputes start.
- CPA Goal
A smart-bidding mode where the panel optimises CPM bids automatically to hit a target cost per acquisition. PropellerAds SmartCPM, Adsy in-house RTB and RichAds calendar push each implement the concept differently; the relevant question is how much conversion volume the algorithm needs before the optimisation stabilises. Cold-start matters more than the marketing materials concede.
- CPC
Cost per Click — the price advertisers pay each time a user clicks an ad. The format-comparison metric most reviewers default to and the one that hides the most. CPC at parity between two networks rarely means parity in cost per acquisition; the click-to-conversion ratio is where the comparison actually lives.
- CPM
Cost per Mille — the price advertisers pay per 1,000 ad impressions. Same metric, different measurement methodologies across networks: PropellerAds reports CPM against gross impressions, Adsterra against rendered impressions, Adsy against in-house RTB clearing impressions. The gap is small in aggregate and material when reconciling spend against advertiser-side trackers.
- CTR
Click-Through Rate — clicks divided by impressions. Same metric, different methodologies across networks. PropellerAds reports CTR against gross impressions; Adsterra reports against rendered impressions; the gap matters when reconciling format performance across two panels. Use it as a creative-quality indicator rather than a between-network comparator without the methodology footnote.
D
- DSP
Demand-Side Platform — the buyer-side software advertisers use to purchase ad inventory programmatically. The category that consolidated through the 2018-to-2022 cycle into the holding-company structure trade-press readers will recognise from the Magnite, Equativ-Sharethrough and Outbrain-Teads coverage. For ad-network buyers, the relevant distinction is whether the network operates its own DSP layer (Adsy in-house RTB) or aggregates demand from external DSPs (most of the cohort).
E
- eCPM
Effective CPM — revenue per 1,000 impressions normalised across pricing models. The publisher-side metric that matters more than any other and the one that varies the most across networks running the same inventory. Monetag, Adsterra and PropellerAds will each quote different eCPMs on matched inventory because the auction stack, demand mix and revenue share differ — comparing requires running the same site across two networks for a full test cycle.
F
- Fill rate
Percentage of ad requests that receive a served ad versus returning empty. The publisher-side metric where headline figures and reconciled figures diverge most often. Tier-2 LATAM and Tier-3 SEA fill rates vary materially by network — Adsterra and HilltopAds typically run higher than the incumbent Cyprus-cluster competitors on those GEOs, but only against matched ad-format requests.
G
- GEO
The geographic-targeting parameter. Country-level by default, with city- or region-level granularity available on most panels. The single biggest non-creative variable in performance testing; the same offer on the same network reads as a different product across Tier-1, Tier-2 and Tier-3 GEOs, and any comparison that does not hold GEO fixed is comparing apples and weather.
I
- In-Page Push
A push-style notification rendered inside the publisher page with no subscription event required. iOS-eligible, which classic web push is not. The format that has eaten classic-push share on every panel I have benchmarked since 2023, with RichAds and PropellerAds carrying the deepest Tier-1 inventory.
- Interstitial
A full-screen advertisement displayed between content transitions, frequency-capped to protect the surrounding user experience. The unit that wins on app-install and 5-second creative sequences; the unit that loses to popunder on iGaming player-acquisition by roughly 2.1x at matched spend in the Q3 2024 parallel buys I ran across PropellerAds and Adsterra.
N
- Native Ads
Advertisements styled to match the visual register of the surrounding editorial content. Higher CTRs than display, lower banner blindness, longer effective creative cycle. The format where Adsy’s UTM-tagged source-publisher attribution beats the incumbent native networks on the basic product decision, and where Mondiad runs as the runner-up at small-to-mid spend.
- Net-14
The payout cycle where the publisher receives payment 14 days after the end of the earnings period. The middle tier — slower than Net-7 (HilltopAds, PropellerAds, Mondiad), faster than Net-30 (Adcash, ExoClick, TwinRed). Where it sits in the network’s value proposition is usually downstream of when the network last revised its publisher contracts.
- Net-7
The payout cycle where the publisher receives payment 7 days after the end of the earnings period. The publisher-friendly default at HilltopAds, Mondiad, PropellerAds and Adsy; the standard against which the slower Net-14 and Net-30 cycles are evaluated. Cash-conversion cycle matters more on the publisher side than the trade press tends to acknowledge.
P
- Popunder
The full-page advertisement that opens in a new browser tab behind the active window. The highest-paying display format on most networks and the one most likely to be dismissed by trade-press writers who have never run a parallel buy across PropellerAds, Adsterra and Adsy on matched inventory.
R
- ROAS
Return on Ad Spend — revenue generated per dollar spent on ads. The advertiser-side metric the trade press uses interchangeably with ROI even though the two diverge once gross-versus-net revenue and lifetime value enter the conversation. Network-reported ROAS without the attribution-window disclosure is theatre; the methodology footnote does the work.
- RTB
Real-Time Bidding — the programmatic auction where advertisers bid per impression in milliseconds. The auction layer most ad networks aggregate to the daily roll-up in the panel; Adsy in-house RTB is the exception in the cohort that exposes clearing-CPM per impression. The transparency difference is small in aggregate and material on reconciliation.
S
- S2S Postback
Server-to-server conversion tracking. The advertiser’s tracker fires a URL when a conversion occurs, which maps the conversion back to the originating click. The single biggest tracking-reliability upgrade over pixel-based attribution and the one feature whose absence on a network panel should change the buy decision. Voluum, Bemob and RedTrack are the trade-standard trackers I see in the wild.
- Smart CPM
Automated bid adjustment that lowers the advertiser’s CPM bid when auction competition is weak, preserving spend efficiency. PropellerAds SmartCPM is the implementation the trade-press refers to most often; the marketed behaviour matched my parallel-buy testing in Q1 2025 across Tier-1 iGaming campaigns. Adsy and RichAds run conceptually similar optimisations on different naming.
- Smartlink
A single URL that routes traffic to the best-matching offer based on GEO, device and other signals. The Mobidea-popularised technology that abstracts the offer-selection layer away from the affiliate, with the trade-off that sophisticated buyers lose direct control over which advertiser their click reaches. Right tool for beginners; wrong tool for buyers who optimise at the offer level.
- SSP
Supply-Side Platform — the seller-side software publishers use to sell inventory to multiple DSPs. The mirror of the DSP category, with the same consolidation history and the same holding-company logic. Publisher-network composition across Monetag, PropellerAds and Adsterra ultimately resolves to which SSPs each network integrates with and how the auction stack ranks bids; the headline numbers downstream of those decisions.
T
- Tier-1 GEO
High-value English-speaking markets — US, UK, CA, AU — plus Western Europe in most network definitions, where the definition matters. Highest CPMs, deepest competition. PropellerAds and Adsterra concentrate publisher relationships here; the auction is the most efficient and the AM allocation reflects that.
- Tier-2 GEO
The mid-tier markets — most of EU, LATAM, and depending on the network’s definition parts of MENA. Mid CPMs, mid competition, the segment where Adsterra’s roughly 30 percent Tier-2 popunder cost-advantage over PropellerAds in Q3 2023 testing has held within a few percentage points since.
- Tier-3 GEO
Emerging markets — SEA (Indonesia, Vietnam, Thailand), parts of Africa, smaller LATAM markets in most network definitions. Lower CPMs and higher volume; the segment where HilltopAds outperforms the Tier-1-focused incumbents on inventory depth, and where Clickadu carries inventory the larger networks structurally do not compete for.
V
- VAST
The IAB Video Ad Serving Template — the specification that lets video players request, display and report on advertisements. VAST 2.0, 3.0 and 4.0 are the relevant versions, with backwards compatibility uneven across the major DSPs. The reconciliation between server-reported and verified-viewable impressions is closer for VAST video than for any other format, which is one of the format’s structural advantages over interstitial and in-page push on brand-side reporting.
- Vertical
The category of offer or content — iGaming, VPN, dating, mobile-CPI, eCommerce and so on. The variable that decides which network is the right call more often than spend tier or GEO. Adsterra and PropellerAds for iGaming Tier-1, ExoClick and TwinRed for adult, Mobidea for mobile-CPI smartlink, RichAds for push-format-first dating and nutra — vertical specialisation runs deeper than the marketing materials concede.