Adtech & Martech Lingo

The Umbrex Media & Entertainment Industry Practice has prepared this guide to terminology, acronyms, shorthand, and insider language to help a newcomer to the Adtech & Martech sector get up to speed rapidly.

Adtech Market Structure

Demand-Side Platform (DSP)

A demand-side platform lets advertisers, agencies, and their algorithms buy digital advertising inventory across multiple publishers and exchanges. The DSP evaluates an ad opportunity, decides whether to bid, selects a creative, and sets the bid price, often within milliseconds.

Practitioners use DSP to mean both the technology and the buying relationship. When someone asks, “Which DSP is holding the seat?”, they are asking which platform account is submitting the bid and appearing in the transaction record. A DSP is not simply an ad-buying dashboard; its bidding models, identity integrations, supply access, fee structure, and optimization logic can materially change campaign outcomes.

Supply-Side Platform (SSP)

A supply-side platform helps publishers package, price, and sell advertising opportunities to buyers. It receives ad requests from publisher systems, exposes eligible opportunities to DSPs or exchanges, runs or participates in auctions, and returns the winning ad.

SSPs are sometimes called sell-side platforms. They overlap with exchanges and publisher ad servers, but they are not identical. An SSP primarily represents the seller’s inventory and yield interests. In practice, publishers often use several SSPs, which is why duplicate supply paths, auction mechanics, and seller transparency receive so much attention.

Ad Exchange

An ad exchange is a transactional marketplace where advertising opportunities are matched with bids, commonly through real-time bidding. The exchange applies eligibility rules, auction logic, floors, and transaction controls before identifying a winner.

The line between an exchange and an SSP has blurred because many platforms perform both functions. When practitioners say “the exchange,” they may mean the auction environment, the company operating it, or the specific supply path through which a bid arrived. Clarifying which meaning applies can save a surprisingly long argument.

Ad Network

An ad network aggregates inventory from publishers and resells it to advertisers, usually as audience, contextual, or placement packages. Unlike a pure exchange, a network may curate inventory, set package pricing, manage campaigns, or assume more direct commercial responsibility.

Networks remain relevant even in programmatic markets, particularly for specialized audiences, local media, mobile applications, video, and managed service. Newcomers often assume that “network” means obsolete or non-programmatic. It does not. A network can use programmatic infrastructure while still operating as an intermediary with bundled inventory and services.

Publisher Ad Server

The publisher ad server determines which eligible campaign, line item, or programmatic demand source receives an ad opportunity on a publisher’s property. It enforces priorities, pacing rules, targeting, frequency controls, and competitive exclusions before returning an ad or calling another selling system.

This is distinct from an SSP. The ad server usually acts as the publisher’s central decisioning and recordkeeping layer, while SSPs connect inventory to external demand. In operating reviews, the ad server is often treated as the source of truth for delivered impressions, although buyers may have a different source of truth and an entirely different number.

Advertiser Ad Server

An advertiser ad server</strong provides centralized creative delivery, tracking, attribution inputs, and reporting across publishers and buying platforms. It can record impressions and clicks independently of the publisher or DSP, making it useful for cross-channel measurement and billing reconciliation.

The familiar example is Campaign Manager 360, although the category is broader than one product. The advertiser ad server does not usually decide the auction winner. It serves or tracks the creative after a buying decision has been made. Practitioners may call its counts third-party numbers, meaning third-party relative to the publisher’s own ad server.

Walled Garden

A walled garden is a media platform that controls its audience data, inventory, buying tools, and measurement environment with limited external access. Large search, social, commerce, and connected television platforms commonly operate this way.

The important feature is not simply scale. It is the platform’s ability to define the identity, auction, reporting, and attribution rules inside its own boundary. A walled garden may provide strong deterministic data and closed-loop measurement, but comparison with open-web media can be difficult because definitions, windows, and methodologies differ.

Inventory and Ad Formats

Ad Opportunity, Ad Request, and Impression

An ad opportunity is a potential chance to show an ad. An ad request is a system call seeking an ad for that opportunity. An impression is counted only after the applicable serving or measurement condition has been met.

These are not interchangeable. One page load can generate multiple requests, retries, auctions, or bid solicitations without producing the same number of rendered impressions. When funnel numbers seem inconsistent, first determine whether each system is counting opportunities, requests, wins, served ads, rendered ads, or measured impressions.

Ad Unit and Placement

An ad unit defines a sellable ad container or format, including characteristics such as size, media type, and supported creative behavior. A placement usually identifies where that unit appears, such as a homepage position, article template, application screen, or video break.

Implementations vary, and platforms sometimes reverse or blur the terms. The operational distinction is useful: the unit describes what can render, while the placement describes the inventory location and context. Buyers care because two placements using the same nominal size can have very different visibility, audience, and performance.

Native Advertising

Native advertising uses creative components that are assembled to match the surrounding product or editorial experience. Instead of delivering one fixed display image, the buyer may provide a headline, image, description, logo, and call to action, which the publisher renders in its own template.

Native does not mean undisclosed. Legitimate native placements still require advertising disclosure. In programmatic systems, native inventory is often represented through structured assets in the bid request and response. Practitioners distinguish it from standard display because rendering control sits partly with the publisher rather than entirely inside the creative file.

VAST

Video Ad Serving Template (VAST) is an IAB specification for communicating video ad files, tracking events, click destinations, and related metadata between ad systems and video players. A VAST response can contain the media directly or point to another VAST response through a wrapper.

Wrapper depth, unsupported media files, tracking failures, and player compatibility can all prevent a nominally valid response from becoming a played impression. Legacy conversations may mention VPAID, an older interactive and measurement approach with security and performance problems. SIMID and OMID address parts of that functionality through more controlled standards.

Instream and Outstream Video

Instream video appears within video content, such as before, during, or after a program. Outstream video appears outside a conventional video stream, often inside an article or feed where a player opens as the unit enters view.

The distinction affects pricing, user expectations, measurement, and standards eligibility. A video-shaped ad is not automatically instream. Buyers may pay a premium for instream inventory because it is associated with intentional video consumption, sound behavior, and higher completion rates. Misclassification is therefore both a quality problem and an economic one.

Ad Pod

An ad pod is a sequence of ads delivered within one commercial break, especially in connected television and streaming video. Pod metadata may specify total duration, number of positions, individual ad length, and whether a position is first, middle, or last.

Pod position matters because the first slot may receive more attention and the last may sit next to program resumption. Systems must also manage competitive separation, repeated creatives, and total break length. A campaign can technically deliver its impressions while producing a poor viewer experience if pod deduplication is weak.

CTV, OTT, and FAST

Connected TV (CTV) refers to advertising delivered to an internet-connected television screen or device. Over-the-top (OTT) refers more broadly to video delivered over the internet rather than through a traditional distribution path. Free ad-supported streaming television (FAST) describes services offering linear-like channels without a subscription fee.

CTV is a device or screen context, OTT is a distribution concept, and FAST is a business and programming model. They overlap but are not synonyms. The distinctions matter for identity, household reach, podding, measurement, and whether the inventory resembles linear television, on-demand streaming, or digital video.

Ad Serving and Trafficking

Ad Tag and Tracking Pixel

An ad tag is code or a URL that requests, renders, or redirects to advertising content. A tracking pixel is a small request used to record an event, often an impression, page view, or conversion. Modern pixels do not need to involve a visible one-pixel image; the name survived the implementation.

Tags can execute scripts, call several vendors, set or read identifiers where permitted, and trigger measurement requests. A newcomer should ask whether a team means a serving tag, impression tracker, click tracker, conversion pixel, or container tag. “Add the tag” is often the beginning of the question, not the answer.

Line Item

A line item is a configurable delivery object inside an ad server or buying platform. It commonly contains targeting, dates, budget or impression goals, bid or rate, priority, pacing, creative assignments, and frequency rules.

Line items operationalize the media plan. A single campaign may contain many of them to separate audiences, geographies, formats, devices, deals, or optimization strategies. When someone says performance is “at the line-item level,” they usually mean that aggregation at campaign level is hiding materially different behavior underneath.

Creative Trafficking

Creative trafficking is the specialized process of configuring creatives, trackers, click URLs, placements, rotation rules, and technical validations in ad systems. It is more than uploading artwork. The trafficker must ensure that the correct asset serves in the correct environment and that all measurement calls fire as intended.

Common failures include mismatched dimensions, invalid redirects, expired landing pages, unsupported video codecs, missing macros, and trackers placed in the wrong event. A campaign can be commercially approved and fully bought yet remain unable to spend because trafficking is incomplete.

Flight and Pacing

A flight is the scheduled delivery period for an advertising line item or campaign. Pacing is the logic used to distribute the goal or budget across that period, such as evenly, ahead of schedule, or as quickly as possible.

Underdelivery does not necessarily mean insufficient inventory. It may reflect restrictive targeting, low bids, frequency caps, creative rejection, weak identity coverage, or conservative pacing. Conversely, aggressive front-loading can consume budget before higher-value opportunities arrive later in the flight.

Frequency Cap

A frequency cap limits how often an identified browser, device, user, or household can receive an ad within a specified period. A typical rule might be three impressions per user per seven days.

The phrase sounds more precise than the underlying identity often is. Cookie loss, device switching, platform boundaries, and household-level identifiers make cross-channel enforcement difficult. A cap of three inside each platform can become considerably more than three for the person watching the ads.

Waterfall

A waterfall, sometimes called a daisy chain, offers an ad opportunity to demand sources sequentially according to a configured priority or expected value. If the first source does not fill, the request passes to the next.

Waterfalls can be simple to operate but introduce latency and imperfect price competition. A lower-ranked source willing to pay more may never see the opportunity. Header bidding developed largely to let multiple demand sources compete before the publisher ad server makes its final decision.

Header Bidding and Wrapper

Header bidding solicits bids from multiple demand partners before the publisher ad server selects the winning line item. A wrapper coordinates those calls, applies timeouts and configuration rules, and sends bid values into the ad server.

The method can increase competition and price transparency, but every additional partner can add processing, network traffic, and operational complexity. When a team discusses wrapper performance, it is usually balancing bid density and revenue against latency, user experience, and browser limits.

Prebid.js and Prebid Server

Prebid.js is a widely used open-source header bidding framework that runs primarily in the browser. Prebid Server moves much of the auction communication to a server environment, which can reduce client-side calls and support environments where browser execution is limited.

Server-side execution is not automatically superior. It may have lower cookie match rates, different privacy obligations, and less direct client context. Many publishers use hybrid configurations, assigning selected partners to client-side or server-side paths according to latency, identity, and demand performance.

Timeout and Passback

A timeout stops waiting for a demand source after a defined interval. A passback redirects an unfilled or unusable opportunity to another demand source. Both protect delivery, but both can create lost value or latency when poorly configured.

A short timeout may exclude slow but valuable bids. A long timeout may delay the page or video. Passbacks can also create loops, duplicate requests, or measurement confusion. “The demand is there, but it is arriving after timeout” means the commercial relationship exists while the technical opportunity does not.

Ad-Serving Discrepancy

An ad-serving discrepancy is the difference between impression, click, or conversion counts reported by two systems involved in the same campaign. Publisher and advertiser ad servers commonly differ because they count at different points, operate in different time zones, filter traffic differently, or fail to receive the same calls.

Small differences are normal. Large differences can indicate broken tags, redirect latency, blocked trackers, invalid traffic filtering, cache behavior, or implementation errors. The contractual question is which system controls billing and what discrepancy threshold triggers investigation or reconciliation.

Programmatic Trading

Real-Time Bidding and OpenRTB

Real-time bidding (RTB) is the automated process of evaluating and bidding on individual ad opportunities as they occur. OpenRTB is the IAB Tech Lab protocol commonly used to structure bid requests and responses between supply and demand platforms.

OpenRTB standardizes the language, not every implementation. Platforms add extensions, interpret fields differently, and omit information for privacy or commercial reasons. A field existing in the specification does not guarantee that it is populated, accurate, or accepted by every trading partner.

Bidstream

The bidstream is the flow of structured information associated with programmatic auction opportunities. It may include the publisher or application, ad format, device type, coarse location, content context, identifiers, consent signals, floors, and supply-chain metadata.

For buyers, bidstream quality determines what their models can evaluate. For sellers, unnecessary fields can create privacy exposure or allow buyers to infer commercially sensitive information. “It is in the bidstream” means the field may be available to auction participants, not that the underlying data is necessarily trustworthy.

Bid Request and Bid Response

A bid request describes an available advertising opportunity and asks eligible buyers for offers. A bid response returns a bid price, creative reference or markup, buyer identity, campaign identifiers, and related metadata.

A request can produce no response because the DSP declines to bid, times out, lacks an eligible creative, cannot use the identity, or does not value the opportunity above the floor. A bid response can still lose the auction, fail creative review, or fail to render. “We sent the request” is therefore several steps away from “the ad appeared.”

Seat ID

A seat ID identifies a buyer account or buying entity within a DSP, exchange, or protocol. Agencies, advertisers, trading desks, and resellers may operate through different seats even when using the same platform.

Seat identity matters for deal eligibility, fee arrangements, creative approval, reporting, and supply-chain transparency. If a private deal is configured for the wrong seat, the buyer may see the inventory in the open auction but remain unable to transact under the negotiated terms.

First-Price Auction and Second-Price Auction

In a first-price auction, the winner generally pays its submitted bid, subject to platform rules and fees. In a traditional second-price auction, the winner pays an amount related to the next-highest eligible bid rather than its full bid.

Most major programmatic auctions moved toward first-price mechanics. Buyers therefore need to estimate the market-clearing price instead of bidding their full theoretical value every time. Auction language still requires care because exchanges may apply floors, deal priorities, hybrid logic, or other adjustments around the headline auction type.

Bid Shading

Bid shading estimates a bid below the buyer’s maximum value that may still win a first-price auction. DSPs typically use historical auction data, context, floor signals, and predicted competition to calculate the shaded bid.

Aggressive shading reduces media cost but can lower win rate and delivery. Weak shading overpays. When a campaign cannot scale, practitioners may ask whether the problem is audience scarcity, low base bids, or a shading model that is trimming too enthusiastically.

Floor Price

A floor price is the minimum acceptable bid for an opportunity or inventory segment. Floors can vary by placement, device, geography, format, buyer, deal, or predicted demand.

A hard floor rejects bids below the threshold. A soft floor influences auction pricing without necessarily rejecting every lower bid, although terminology and implementation vary by platform. Floors protect value only if they do not suppress competition or fill more than the higher price compensates for.

Deal ID

A deal ID is an identifier used to recognize negotiated programmatic terms between a seller and eligible buyer. It can activate special inventory access, pricing, priority, targeting, or auction rules.

The ID itself does not guarantee delivery. Both sides must configure compatible seats, formats, floors, dates, creatives, and targeting. When someone says “the deal is live,” the useful follow-up is whether bid requests bearing that deal ID are reaching the intended DSP seat and receiving eligible bids.

PMP, Preferred Deal, and Programmatic Guaranteed

These terms describe distinct forms of negotiated programmatic access:

Structure	Typical mechanics	Commitment
Private marketplace (PMP)	Restricted auction among selected buyers, usually using a deal ID	Generally no guaranteed spend or volume
Preferred deal	Buyer receives first-look or preferential access at a negotiated fixed price	Usually non-guaranteed
Programmatic guaranteed	Automated execution of reserved inventory at agreed terms	Buyer and seller commit to volume or spend

Platforms do not always use the labels consistently. The practical questions are whether inventory is reserved, whether price is fixed, whether the buyer must purchase, and where the deal sits in the publisher’s decision hierarchy.

Curated Marketplace

A curated marketplace is a packaged programmatic supply offering assembled around selected inventory, audience attributes, contextual criteria, quality controls, or performance objectives. The curator may use SSP tools to create a deal that buyers activate through their DSP.

Curation can simplify fragmented supply, but it also introduces another decision-maker and potentially another fee. Buyers should understand who selected the inventory, whether the curator has exclusive data or merely repackaged public signals, and how the curation fee appears in the supply chain.

Bid Rate and Win Rate

Bid rate is the share of eligible bid requests on which a buyer submits a bid. Win rate is the share of submitted bids, or sometimes eligible requests, that result in auction wins. Denominators vary, so the metric definition matters.

Low bid rate can indicate restrictive targeting, missing identity, creative incompatibility, model rejection, or budget controls. Low win rate often points to price, floors, competition, or latency. Neither metric proves that impressions rendered or achieved the desired business result.

Identity and Addressability

First-Party, Second-Party, and Third-Party Data

First-party data is collected by an organization through its direct relationship with users or customers. Second-party data usually means another organization’s first-party data accessed through a partnership. Third-party data is assembled by an entity without the same direct relationship to the individual.

These labels describe provenance, not automatic permission, quality, or regulatory status. First-party data can still be used unlawfully, and third-party data is not inherently inaccurate. Practitioners should ask how the data was collected, what notices and permissions apply, how current it is, and whether the planned activation is compatible with its original purpose.

A third-party cookie is set or read in a context where the cookie’s domain differs from the site the user is visiting. Adtech historically used these cookies for cross-site identity, audience targeting, frequency control, and attribution.

Browser restrictions and user controls have made third-party cookie availability uneven. “Cookieless” does not mean identifier-free, and continued cookie support in one browser does not restore universal addressability. Teams increasingly combine authenticated identifiers, first-party storage, contextual signals, platform APIs, and modeled measurement.

Mobile Advertising ID (MAID)

A mobile advertising ID is a resettable device identifier provided by a mobile operating system for advertising-related use. Examples include Apple’s Identifier for Advertisers and Google’s advertising ID.

MAIDs are not permanent device serial numbers, and access is constrained by operating-system policies and user choices. They historically supported mobile audience matching, suppression, frequency management, and attribution. Reduced availability shifts more measurement toward consented first-party signals, aggregated frameworks, and modeling.

Cookie syncing, also called ID syncing, maps the identifier used by one adtech platform to the identifier used by another. A redirect or server interaction lets both parties store a correspondence such as “DSP user 123 equals SSP user 789.”

Syncing does not reveal a universal human identity. It connects platform-specific pseudonymous identifiers, often with substantial loss. Browser restrictions, consent requirements, redirect blocking, and identifier expiration reduce coverage. Every additional sync can also add latency and data-sharing complexity.

Deterministic and Probabilistic Identity

Deterministic identity links records using a direct, stable signal such as a login, hashed email address, or verified account relationship. Probabilistic identity infers a connection from signals such as device characteristics, location patterns, network information, and behavior.

Deterministic does not mean infallible. Shared accounts, recycled email addresses, and household logins create ambiguity. Probabilistic does not mean random; it means confidence-based. The distinction matters because activation, measurement, privacy obligations, and acceptable error rates differ by use case.

Identity Graph and Household Graph

An identity graph stores relationships among identifiers, devices, accounts, and profiles believed to represent the same person or related people. A household graph links devices and accounts at household rather than individual level.

Graphs support cross-device reach, suppression, attribution, and frequency control. Their value depends on freshness, coverage, confidence rules, and the permissible uses of each input. A CTV campaign may have strong household identity while remaining unable to determine which person on the sofa saw the ad.

Universal ID and Publisher-Provided ID

A universal ID is an identifier intended to work across participating publishers and adtech platforms, often derived from an authenticated or pseudonymous signal. A publisher-provided ID (PPID) is generated or supplied within a publisher’s first-party relationship and made available under controlled conditions.

Neither is universal in the literal sense. Adoption, consent, technical integration, interoperability, and platform acceptance determine usable scale. Practitioners care less about the theoretical identifier population than about how much eligible inventory and buyer demand can actually transact on it.

Match Rate

Match rate measures the share of records or identifiers successfully linked between datasets or platforms. A simple expression is matched eligible records / records submitted, but exclusions and denominator rules vary.

High match rate does not prove accurate matching, and low match rate may reflect normalization problems rather than limited audience overlap. Email casing, whitespace, hashing sequence, consent filtering, stale identifiers, and country restrictions can all affect the result. Always ask whether the rate is based on uploaded records, eligible records, unique people, devices, or activated profiles.

Supply Quality and Transparency

ads.txt and app-ads.txt

ads.txt lets website publishers declare which companies are authorized to sell their digital inventory. app-ads.txt applies the same concept to mobile and connected applications. Buyers can compare the declared seller account with information in the bidstream.

Entries identify the advertising system, seller account, relationship type, and optional certification authority ID. DIRECT means the publisher controls the listed seller account; RESELLER means another party does. Neither label alone proves inventory quality, but undeclared or inconsistent paths deserve scrutiny.

sellers.json

sellers.json is published by an advertising system to identify the entities selling through its platform. It complements ads.txt by describing the seller account from the platform side, including whether the entity acts as a publisher, intermediary, or both.

Some seller records are confidential, which limits transparency. Buyers use sellers.json to inspect intermediaries, validate seller identities, and understand whether a supposedly direct path actually contains several commercial layers.

SupplyChain Object

The SupplyChain object, commonly called schain, is an OpenRTB structure that lists the sequence of entities involved in selling an impression. Each node can identify an advertising system, seller account, request identifier, and whether the node participates directly in payment.

Ads.txt shows who may sell, sellers.json shows who a platform says its sellers are, and schain describes the path for a particular request. The three are related but not substitutes. Missing or implausible nodes can indicate poor implementation, opaque reselling, or fraudulent representation.

Supply-Path Optimization (SPO)

Supply-path optimization is the buyer-side practice of selecting more efficient, transparent, and effective routes to publisher inventory. Buyers assess fees, duplication, auction behavior, identity quality, latency, seller relationships, and fraud controls across SSPs.

SPO is not simply reducing the number of SSPs. The cheapest path may lack scale or preferred access, while the most direct path may not always win. Good SPO seeks the best economic and operational route, not the shortest diagram for the board slide.

Domain Spoofing and App Spoofing

Domain spoofing falsely represents low-quality or fraudulent inventory as coming from a more valuable website. App spoofing does the same with application identifiers or bundle information.

Spoofing exploits the fact that buyers price inventory partly from declared context. Ads.txt, app-ads.txt, sellers.json, schain, application-store validation, and independent verification all help detect it. A plausible domain string in a bid request is not proof that an ad appeared there.

Invalid Traffic, GIVT, and SIVT

Invalid traffic (IVT) includes impressions or interactions that should not be treated as legitimate human advertising activity. General invalid traffic (GIVT) covers patterns detectable through routine methods, such as known data-center bots. Sophisticated invalid traffic (SIVT) requires more advanced analysis, coordination, or human review.

IVT can arise from malicious fraud, automated tools, measurement systems, or implementation defects. Different verification vendors may classify the same traffic differently because their detection data and methods differ. A low IVT rate is reassuring, but no vendor sees every part of the transaction.

Made-for-Advertising (MFA)

Made-for-advertising describes properties designed primarily to generate advertising impressions, often through inexpensive traffic acquisition, highly templated content, aggressive ad density, and rapid page consumption. MFA classification is based on a pattern of characteristics rather than one universal legal definition.

MFA inventory may be viewable and technically human while still providing weak attention or brand value. That makes it different from classic bot fraud. Buyers increasingly apply exclusion lists, page-level signals, attention measures, and supply-path controls rather than relying only on invalid-traffic filters.

Brand Safety and Brand Suitability

Brand safety seeks to avoid content broadly considered harmful, illegal, or inappropriate for advertising. Brand suitability applies an advertiser’s own tolerance and context rules, which may be stricter or more nuanced.

A news article about a tragedy may be legitimate journalism and safe under one policy but unsuitable for a particular brand. Overly blunt blocking can also defund reputable news and eliminate valuable reach. The distinction matters because “safe” is not the same as “right for this advertiser.”

Viewability and OMID

Viewability assesses whether an ad had the opportunity to be seen according to a defined threshold. Common Media Rating Council benchmarks include at least 50 percent of display pixels in view for one continuous second and 50 percent of video pixels for two continuous seconds, subject to applicable standards.

Open Measurement Interface Definition (OMID) provides a standardized way for verification vendors to measure ads in mobile application and video environments. A viewable impression does not prove attention, comprehension, or business effect. It confirms only that the placement met the specified opportunity-to-see condition.

Media Economics and Delivery Metrics

CPM and eCPM

Cost per mille (CPM) is the price paid per thousand impressions: spend / impressions x 1,000. Effective CPM (eCPM) converts revenue or cost from another pricing structure into an equivalent thousand-impression basis.

CPM is a pricing unit, not a quality measure. A lower CPM may reflect less valuable inventory, poorer viewability, or weaker audience precision. Publishers use eCPM to compare demand sources with different commercial models, while buyers use it to compare the cost of obtaining exposure across tactics.

vCPM

Viewable CPM (vCPM) prices or normalizes media against viewable impressions rather than all served impressions. If half of served impressions are viewable, the effective cost per thousand viewable impressions will be roughly twice the served-impression CPM, all else equal.

Some campaigns transact directly on viewable impressions, while others merely report vCPM as an efficiency metric. The measurement vendor, viewability standard, measurable rate, and billing definition must be clear before two vCPM figures can be compared.

CPC, CPA, and CPL

Cost per click (CPC), cost per acquisition or action (CPA), and cost per lead (CPL) price or evaluate media against downstream events. These measures move the commercial focus beyond impressions, but they also create disputes about attribution and event quality.

A low CPL can be produced by weak leads, duplicate submissions, or generous attribution. A CPA campaign still depends on impression-level auctions underneath, and the platform must predict which opportunities are likely to create the billable action.

Fill Rate and Render Rate

Fill rate measures how many eligible ad requests receive an ad response or winning creative. Render rate measures how many returned or won ads actually render and generate the relevant impression event.

Fill can look healthy while render rate is weak because users leave, pages load slowly, creatives fail, video players reject files, or verification blocks execution. Denominators vary widely, so “90 percent fill” means little until the team defines the request population.

Yield and eRPM

Yield is the value a publisher realizes from its available inventory after considering price, fill, demand competition, user experience, and opportunity cost. Effective revenue per mille (eRPM) expresses publisher revenue per thousand page views, sessions, requests, or impressions, depending on the implementation.

Yield optimization is not equivalent to maximizing CPM. A high-price campaign with low fill may earn less total revenue than broader demand at a lower price. Publishers also consider subscription effects, latency, ad load, and long-term audience value.

Take Rate

An adtech platform’s take rate is the share of transacted media value retained as fees or gross revenue. A simplified formula is platform revenue / gross media spend.

Comparisons are difficult because platforms report gross and net revenue differently and may bundle data, verification, or managed service fees. A low stated take rate can coexist with significant total deductions elsewhere in the path. Practitioners therefore examine the full fee stack, not just one percentage.

Ad Load

Ad load describes the amount of advertising relative to content or user activity. It may be expressed as ads per page, minutes of advertising per hour, impressions per session, or share of screen and time occupied by ads.

Increasing ad load can raise short-term inventory while reducing attention, page speed, retention, and advertiser value. In streaming, it also affects pod repetition and viewer abandonment. More ad slots do not necessarily produce proportionately more monetization.

Working Media

Working media generally means the portion of advertising expenditure directly associated with purchased media exposure, excluding some combination of technology, data, creative, agency, and measurement costs. There is no universally accepted calculation.

The term often appears in procurement and transparency discussions. A higher working-media percentage is not automatically better if excluded services improve targeting, creative quality, fraud prevention, or incrementality. The useful conversation is about which costs create value, not how many can be moved outside a denominator.

Measurement and Attribution

Conversion Pixel and Conversion API

A conversion pixel records an event through browser or application code, such as a purchase, registration, or qualified visit. A conversion API (CAPI) sends events from a server or controlled backend directly to an advertising or measurement platform.

Server-side transmission can improve resilience and data quality, but it does not bypass consent or purpose limitations. When browser and server events are both sent, they require a shared event ID or other deduplication logic. Otherwise, better tracking may produce the exciting discovery that every customer purchased twice.

Click ID

A click ID is a unique parameter attached to an advertising click so a later conversion can be associated with the originating platform, campaign, or ad. Different platforms use proprietary click-ID formats.

Redirects, link shorteners, application transitions, consent settings, and landing-page code can strip the identifier. Server-side systems often capture it at arrival and preserve it with the resulting first-party record. A click ID supports linkage; it does not by itself prove that the ad caused the conversion.

Attribution Window

An attribution window is the period during which an ad exposure or click remains eligible to receive credit for a conversion. Click windows and view windows are often configured separately.

Longer windows generally capture more conversions but also increase the chance of crediting outcomes that would have occurred anyway. Platform comparisons are meaningless unless their windows, event timestamps, time zones, and identity rules are aligned.

Click-Through and View-Through Attribution

Click-through attribution credits an ad after a user clicks and later converts within the allowed window. View-through attribution credits an ad after a recorded impression without a qualifying click.

View-through attribution is important for video, display, and television-like formats, but it is vulnerable to over-crediting because exposure is not causation. Practitioners typically apply shorter view windows, viewability requirements, or incrementality testing to interpret it responsibly.

Last-Touch Attribution

Last-touch attribution assigns conversion credit to the final eligible marketing interaction before the conversion. Last-click is the common version that considers only clicks.

The method is simple and operationally useful, but it favors channels close to conversion, such as branded search, retargeting, and affiliate traffic. It often undervalues channels that created awareness or consideration earlier. “Last touch won” may simply mean “last touch was standing nearest the cash register.”

Multi-Touch Attribution (MTA)

Multi-touch attribution distributes conversion credit across several marketing interactions using rules or statistical models. Models may assign equal credit, emphasize certain positions, apply time decay, or estimate contribution algorithmically.

MTA requires sufficiently complete person-level exposure and conversion data. Identity loss, walled gardens, offline activity, and privacy restrictions make that increasingly difficult. It can support tactical optimization, but its precision should not be confused with proof of causality.

Marketing Mix Modeling (MMM)

Marketing mix modeling uses aggregated time-series data to estimate how media, promotions, pricing, seasonality, economic conditions, and other factors contribute to business outcomes. Modern MMM often uses Bayesian methods and incorporates saturation and carryover effects.

MMM works without person-level tracking and can compare online and offline channels, but it is less granular than event-level attribution. Results depend heavily on data variation, model specification, priors, and control variables. It is most useful for allocation and scenario planning, not deciding which banner creative should run tomorrow morning.

Incrementality and Holdout

Incrementality is the additional outcome caused by advertising compared with what would have happened without it. A holdout is an eligible group intentionally withheld from treatment so its outcomes can provide a counterfactual comparison.

Incremental conversions equal the treatment group’s observed outcomes minus the estimated baseline. The hard part is constructing comparable groups and avoiding contamination. Reported conversions can rise while incrementality remains low if advertising mainly reaches people already likely to buy.

Ghost Ads and PSA Controls

Ghost ads identify control-group users who would have been served an ad, then suppress the actual creative while preserving the auction-selection logic. A public service announcement (PSA) control serves a neutral alternative ad to the control group.

Both approaches aim to create a more comparable control than simply withholding all media. Ghost ads reduce the cost and behavioral effect of a control creative, while PSA controls can better equalize ad exposure mechanics. Platform capabilities and auction interference determine which method is feasible.

Reach, Frequency, and Deduplicated Reach

Reach estimates the number of distinct people, devices, or households exposed. Frequency measures average exposures among those reached. Deduplicated reach attempts to count an audience once across publishers, devices, or channels.

Every figure depends on the identity unit. Device reach is not person reach, and household reach is not individual reach. Cross-platform deduplication often relies on panels, graphs, clean rooms, or modeled overlap rather than direct observation of every exposure.

ROAS and Incremental ROAS

Return on ad spend (ROAS) is typically attributed revenue / advertising spend. Incremental ROAS (iROAS) uses revenue causally generated by the advertising rather than all revenue assigned by an attribution rule.

ROAS can look strong for campaigns targeting existing loyal customers. iROAS may be lower but answers the more economically important question. Both require clarity about revenue versus contribution, returns, cancellations, attribution windows, and whether platform-reported sales include customers who would have purchased anyway.

Privacy and Platform Policy

A consent management platform collects, stores, and communicates user choices about data processing and advertising. It can display notices, manage vendor lists, encode consent signals, and block or permit tags according to configured rules.

A CMP does not make an implementation compliant merely by being present. The notice language, legal basis, vendor configuration, tag behavior, regional logic, and recordkeeping must all match actual processing. A beautifully designed banner that fires every adtech tag before consent is mainly decorative.

The IAB Europe Transparency and Consent Framework (TCF) standardizes how participating publishers, CMPs, and vendors communicate privacy choices in relevant European advertising contexts. The encoded Transparency and Consent string, or TC string, records information such as purposes, vendors, legal bases, and user choices.

The string is a communication mechanism, not proof that every downstream action is lawful. Vendors must interpret it correctly, honor restrictions, and maintain their own legal responsibilities. Missing, malformed, or stale strings can cause demand loss as well as regulatory exposure.

Global Privacy Platform (GPP)

The Global Privacy Platform is an IAB Tech Lab specification for communicating privacy, consent, and consumer-choice signals across multiple regulatory jurisdictions. It uses sections corresponding to different legal frameworks or market requirements.

GPP reduces the need for unrelated signal formats, but it does not eliminate jurisdiction-specific logic. A receiving platform must know which sections apply, what each field means, and how to translate the signal into permitted processing.

Global Privacy Control (GPC)

Global Privacy Control is a browser or device signal expressing a user’s request to opt out of certain sale or sharing of personal information. It is particularly significant under laws that require businesses to recognize qualifying preference signals.

GPC is not the same as a generic “do not track” preference, and it should not be treated as optional decoration. Adtech implementations must determine how the signal affects audience creation, cross-context behavioral advertising, data transfers, and downstream partners.

In some privacy laws, particularly California’s framework, sale and sharing are defined terms broader than an ordinary cash transaction. Sharing can include disclosing personal information for cross-context behavioral advertising, even without direct monetary payment.

This is why a link may say “Do Not Sell or Share My Personal Information.” Commercial teams sometimes hear “sale” and assume the company does not sell lists, so the rule is irrelevant. The legal definition is interested in the data flow, not the sales team’s vocabulary.

App Tracking Transparency and Limited Ad Tracking

Apple’s App Tracking Transparency (ATT) framework requires applications to obtain permission before engaging in defined tracking across other companies’ apps and websites or sharing data with data brokers. A denied request limits access to the Identifier for Advertisers and constrains related practices.

Limited ad tracking is a broader historical phrase for operating-system choices that restrict advertising identifier use. ATT status should not be inferred solely from identifier presence. Developers and adtech partners must follow platform policy in addition to applicable privacy law.

SKAdNetwork and AdAttributionKit

SKAdNetwork (SKAN) is Apple’s privacy-preserving framework for mobile application advertising attribution. It returns delayed, constrained postbacks rather than unrestricted user-level conversion records. AdAttributionKit extends and modernizes attribution capabilities across supported Apple environments.

Practitioners discuss conversion values, postback tiers, crowd-anonymity thresholds, lock windows, and modeled results because the framework deliberately limits granularity. A campaign may be measurable in aggregate while remaining impossible to trace user by user.

Protected Audience API and Topics API

The Protected Audience API supports on-device interest-group advertising and remarketing-style auctions with reduced cross-site data exposure. The Topics API provides coarse interest topics derived through browser-controlled logic rather than traditional third-party tracking.

Both originated within Google’s Privacy Sandbox initiative. They are technical mechanisms, not universal replacements for every cookie use case. Adoption, browser support, auction integration, policy decisions, and economic performance determine their practical importance.

Data Clean Room

A data clean room is a controlled environment where parties can compare or analyze data without routinely exposing raw user-level records to one another. Controls may include permissioned queries, aggregation thresholds, row-level restrictions, auditing, and privacy-enhancing computation.

Clean rooms support audience overlap, campaign measurement, reach analysis, and retailer or platform collaboration. They do not create consent, fix poor identity, or make an impermissible use permissible. If both parties bring incomplete data into a clean room, the room remains clean and the answer remains incomplete.

Martech Data Architecture

Customer Data Platform (CDP)

A customer data platform ingests customer data from multiple sources, resolves or unifies profiles, creates audiences and attributes, and makes them available to downstream marketing tools. Persistent profiles and marketer-accessible activation are central expectations, although product boundaries vary.

Some CDPs store the primary customer record; others rely on a cloud warehouse. Some emphasize identity and audience management, while others add analytics, personalization, or journey orchestration. Buying a CDP does not automatically produce a usable customer view. Source quality and identity rules still have to do the less glamorous work.

Data Management Platform (DMP)

A data management platform historically manages pseudonymous audience data for advertising, often using cookies, device IDs, and third-party segments. It creates segments and distributes them to buying and serving platforms.

A DMP generally focuses on media audiences with shorter-lived identifiers, while a CDP focuses on persistent first-party customer profiles. Products can overlap, but using the labels interchangeably obscures differences in identity, retention, data types, permissions, and use cases.

Composable or Warehouse-Native CDP

A composable CDP, often called a warehouse-native CDP, performs segmentation and activation using data already stored in the organization’s cloud data warehouse rather than copying everything into a separate packaged data store.

The model can reduce duplication and give data teams more control, but it transfers responsibility for schemas, identity, data quality, and compute economics to the organization. “Composable” means the components can be assembled, not that they assemble themselves.

Customer 360 and Golden Record

Customer 360 describes a unified view of customer interactions and attributes across systems. A golden record is the selected or reconciled representation treated as authoritative for a person, account, or entity.

Neither requires every source to be physically merged into one database. The difficult questions are which identifiers link records, which source wins when fields conflict, how quickly updates propagate, and whether each attribute is permitted for the intended marketing use.

Profile Unification and Computed Traits

Profile unification combines events and attributes believed to belong to the same person, account, or household. Computed traits are derived attributes such as predicted lifetime value, days since last purchase, product affinity, or engagement score.

Unification errors can merge different people or fragment one person into several profiles. Computed traits inherit the quality and timing of their inputs. A highly sophisticated propensity score attached to the wrong customer is still wrong, only with more decimal places.

Event Taxonomy

An event taxonomy defines the names, properties, triggers, and semantic meaning of behavioral events such as product_viewed, checkout_started, or subscription_cancelled. It allows analytics and activation systems to interpret events consistently.

Good taxonomies distinguish an event from its properties and specify when the event fires. Without that discipline, one platform’s “purchase” may mean payment initiated while another means payment settled. Downstream audiences and attribution will faithfully reproduce whichever ambiguity was instrumented upstream.

Data Layer

A data layer is a structured interface through which a website or application exposes contextual and behavioral information to tags and marketing systems. It may contain page type, product identifiers, transaction values, consent state, or user status.

The data layer decouples marketing tags from unstable page markup. It should be treated as a governed interface, not an informal bucket. Renaming a field or changing its timing can silently break analytics, personalization, and campaign measurement across many tools.

Tag Management System and Container

A tag management system (TMS) controls the deployment and firing rules for analytics, advertising, and personalization code. A container is the implementation bundle loaded on a site or application to execute those configured tags.

Tag managers accelerate deployment but can also bypass normal software controls if permissions are loose. Teams need versioning, consent integration, environment separation, and testing. The fact that a marketer can publish a tag without a release cycle does not always mean they should.

Client-Side and Server-Side Tagging

Client-side tagging sends marketing and analytics requests directly from the user’s browser or application. Server-side tagging routes events through infrastructure controlled by the organization before forwarding approved data to vendors.

Server-side tagging can improve control, resilience, and payload consistency. It can also obscure data flows if poorly governed. It does not erase browser restrictions, manufacture missing consent, or convert third-party use into first-party use merely because a company-owned server touched the request.

Source, Destination, and Reverse ETL

In customer-data tooling, a source produces events or records, while a destination receives them for analysis or activation. Reverse extract, transform, load (reverse ETL) moves modeled or curated data from a warehouse into operational tools such as email platforms, DSPs, and customer-engagement systems.

Activation requires more than moving columns. Data must be mapped to the destination’s identity keys, consent rules, update behavior, deletion process, and audience semantics. A successful sync message confirms technical delivery, not that the receiving platform matched or used every record.

Marketing Automation and Lifecycle Messaging

Marketing Automation Platform (MAP)

A marketing automation platform manages campaign workflows, contact segmentation, lead nurturing, scoring, and triggered communications, especially in business-to-business and lifecycle marketing. It often integrates with customer relationship management systems and web forms.

A MAP differs from an ad-buying platform. It primarily acts on known or permissioned contacts through channels such as email and messaging, while DSPs purchase media opportunities. Modern suites overlap, so teams should focus on the system performing orchestration, identity storage, and delivery rather than the vendor’s preferred category label.

Journey Orchestration

Journey orchestration selects and sequences customer communications across channels based on events, attributes, eligibility rules, and prior interactions. A journey may branch when a customer opens an email, abandons a cart, reaches a score threshold, or completes a purchase.

Orchestration requires arbitration when several journeys want to contact the same person. Without priority, suppression, and frequency rules, each workflow can behave correctly while the customer receives a small avalanche of individually reasonable messages.

Triggered Campaign

A triggered campaign begins when a defined event or state change occurs rather than solely on a scheduled batch date. Typical triggers include product abandonment, account activation, replenishment timing, contract renewal, or lapse behavior.

The important implementation details are event latency, eligibility, deduplication, cooldown periods, and cancellation logic. A delayed purchase event can cause an abandonment message to arrive after the customer has already bought the product.

Nurture Stream and Drip Campaign

A nurture stream is a staged sequence of communications intended to advance a contact through education, evaluation, or readiness. A drip campaign often refers to a simpler time-based series, although practitioners do not always maintain the distinction.

More sophisticated nurture programs branch according to behavior, score, role, product interest, or account status. Entry and exit rules matter as much as message content. A prospect who becomes a customer should not remain trapped in a six-email sequence explaining why they should become one.

Lead Scoring

Lead scoring assigns points or predicted probabilities based on fit and behavior to estimate sales readiness or conversion likelihood. Explicit signals may include company size or role; implicit signals may include site visits, content engagement, and event attendance.

Rules-based scores are easy to explain but can reward noisy behavior. Predictive scores can capture complex patterns but require adequate outcomes and monitoring. The score should be validated against progression and revenue, not admired because it goes to 100.

MQL, SAL, and SQL

A marketing-qualified lead (MQL) meets marketing’s agreed threshold for fit or engagement. A sales-accepted lead (SAL) has been reviewed and accepted for follow-up. A sales-qualified lead (SQL) has met a stronger sales qualification standard, often involving confirmed need, authority, or buying intent.

Definitions vary sharply between organizations. The useful artifact is the service-level agreement specifying criteria, routing, rejection reasons, and recycling rules. If MQL volume rises while acceptance falls, the scoring system may be optimizing the handoff metric rather than actual demand.

Deliverability and Inbox Placement

Delivery means a receiving mail server accepted a message. Deliverability concerns whether messages can reliably reach intended recipients, while inbox placement distinguishes the primary inbox from spam or other filtered locations.

A platform can report high delivery even when many messages land in spam. Sender reputation, authentication, complaint rates, engagement, list quality, content, and sending patterns all influence placement. “Sent” is an operational event, not evidence that a human saw the message.

SPF, DKIM, and DMARC

Sender Policy Framework (SPF) identifies servers authorized to send mail for a domain. DomainKeys Identified Mail (DKIM) applies a cryptographic signature to validate message integrity and domain responsibility. Domain-based Message Authentication, Reporting and Conformance (DMARC) uses SPF or DKIM alignment and specifies how receivers should handle failures.

These controls support authentication and reputation but do not guarantee inbox placement. Misalignment between visible sender domains, return paths, and signing domains is a common failure. DMARC reporting also helps detect unauthorized use of the brand’s domain.

Retail Media and Commerce

Retail Media Network (RMN)

A retail media network lets a retailer or commerce platform monetize advertising using its shopping environments, customer relationships, transaction data, and media inventory. It may offer sponsored search, display, video, offsite media, connected television, and in-store formats.

The strategic attraction is closed-loop data connecting exposure to purchase. Capabilities vary widely, however. Some RMNs operate sophisticated buying and measurement platforms; others are collections of inventory and reports sharing a logo and a revenue target.

Onsite, Offsite, and In-Store Media

Onsite media appears on the retailer’s owned digital properties. Offsite media uses retailer audiences or data on external publishers and platforms. In-store media includes screens, audio, digital fixtures, and other advertising within physical locations.

The channels differ in identity, creative context, inventory ownership, measurement, and fee structure. Offsite activation often requires a DSP, clean room, or identity partner, while onsite sponsored products are closely tied to search and product availability.

Closed-Loop Measurement

Closed-loop measurement connects advertising exposure or interaction with transaction outcomes observed inside the same retailer or platform ecosystem. It can report attributed sales, units, customers, or basket behavior using the platform’s first-party data.

The loop may be closed technically while remaining narrow economically. It may exclude purchases at other retailers, use platform-specific attribution windows, or credit customers who already intended to buy. Buyers should ask whether outcomes are online only, store only, omnichannel, matched, modeled, or incremental.

Endemic and Non-Endemic Advertisers

An endemic advertiser sells products through the retailer operating the media network. A non-endemic advertiser does not sell there directly but wants access to the retailer’s audience or media, such as a bank targeting home-improvement shoppers.

Endemic campaigns can be measured against product sales with relative ease. Non-endemic campaigns need different outcomes, such as site visits, applications, brand lift, or offline conversion. The distinction changes inventory, measurement, sales organization, and the credibility of closed-loop claims.

New-to-Brand (NTB)

New-to-brand classifies purchases from customers who have not bought the advertiser’s brand within a defined lookback period on that retail platform. Common outputs include NTB orders, sales, customer rate, and acquisition cost.

NTB does not necessarily mean new to the brand everywhere. A customer may have purchased through another retailer or outside the lookback window. The measure is still useful, but its platform boundary and historical period should be explicit.

Share of shelf measures a brand’s visibility within a digital or physical commerce environment. Share of search commonly measures the brand’s portion of search visibility, query volume, impressions, or clicks, depending on the methodology.

Paid and organic visibility should be separated. A brand can gain sponsored placement while losing organic relevance, or dominate impressions while remaining out of stock. The metric is most useful when tied to specific queries, categories, positions, and availability conditions.

Commercial and Buying Mechanics

Insertion Order (IO)

An insertion order is the campaign-level commercial document authorizing media placement under specified dates, units, volumes, rates, targeting, and billing terms. It often sits beneath a broader master agreement or standard industry terms.

Despite the name, an IO is not merely an administrative form. It can control cancellation, data use, measurement source, makegoods, and liability. Programmatic self-service buying may not use a traditional IO for every campaign, but negotiated and managed-service media commonly does.

Managed Service and Self-Service

In self-service, the advertiser or agency operates the buying platform directly. In managed service, the platform or media partner performs campaign setup, optimization, and reporting on the buyer’s behalf.

The distinction affects fees, transparency, data access, control, and accountability. A managed-service campaign may combine media, technology, and labor into one rate, making direct comparison with self-service platform fees difficult.

Principal and Agent Buying

In agent buying, an agency purchases media on the advertiser’s behalf and typically passes through the underlying media terms subject to agreed compensation. In principal buying, the agency or intermediary purchases inventory on its own account and resells it to the advertiser.

Principal models can provide packaged pricing, guarantees, or proprietary inventory access, but they change transparency and conflict considerations. Commercial documents should make clear whether the intermediary acts for the advertiser or as the advertiser’s counterparty.

Gross Media and Net Media

Gross media and net media distinguish spend before and after specified commissions, discounts, rebates, or platform deductions. The exact convention varies by market, channel, and contract.

A quoted CPM may be gross to the advertiser, net to the publisher, or exclusive of data and technology fees. Practitioners should identify the reference point before comparing prices. Otherwise, two people can agree on the number while discussing different amounts of money.

Makegood

A makegood is replacement inventory, credit, or another remedy provided when contracted media fails to meet agreed delivery or quality terms. Triggers can include underdelivery, incorrect placement, technical failure, or failure to achieve a guaranteed audience condition.

A makegood does not necessarily reproduce the original value. Inventory in a later flight may have different seasonality, audience, or market price. The parties therefore negotiate timing, equivalence, measurement, and whether the remedy is additional media or a financial credit.

Sequential Liability

Sequential liability is a media-contract concept under which an agency’s obligation to pay a publisher depends on receiving the corresponding funds from the advertiser, while the advertiser’s liability may be discharged once it pays the agency.

The clause allocates credit risk across advertiser, agency, and media owner. Its operation depends on the governing agreement and jurisdiction. Finance teams should understand it before assuming the party that placed the order necessarily bears unconditional payment responsibility.

The Phrase Translator

“The deal ID is live, but we are not seeing bid requests on the seat.”

It may mean: The seller configured the private deal, but the technical or account setup is wrong. Check seat eligibility, dates, formats, targeting, publisher routing, and whether requests actually contain the negotiated deal ID.

“The wrapper is timing out the long tail.”

It may mean: Slower header-bidding partners are returning bids after the publisher’s deadline. They may claim demand, but bids that arrive after timeout have roughly the economic value of a train arriving after you left the station.

“We are over the floor after shading, but win rate is still soft.”

It may mean: The DSP believes its bid clears the stated minimum, yet competitors, hidden auction dynamics, deal priority, latency, or supply-path differences are preventing wins.

“The SSP says the path is direct, but sellers.json shows an intermediary.”

It may mean: The commercial description and technical transparency files do not align. The team needs to inspect ads.txt, sellers.json, and schain before treating the inventory as a direct publisher relationship.

“Fill is healthy; render rate is not.”

It may mean: Auctions are returning ads, but users are not actually receiving countable rendered impressions. Investigate latency, creative errors, page abandonment, player compatibility, blocked requests, and impression-counting points.

“Frequency is capped in-platform, not deduplicated across the household.”

It may mean: Each buying environment is enforcing its own cap against its own identifiers. The same household can still receive many more impressions across devices, publishers, and walled gardens.

“The placement is brand-safe, but not brand-suitable for this brief.”

It may mean: The content is not broadly harmful or prohibited, but its subject matter, tone, or audience context conflicts with this advertiser’s specific standards.

“We need an incrementality read, not another last-click ROAS chart.”

It may mean: Attributed sales are not answering whether the campaign caused additional sales. The team wants an experiment, holdout, or causal model rather than another allocation of credit.

It may mean: Improve tracking resilience through a conversion API or server container without sending unpermitted data or counting the browser and server copies as separate conversions.

“The CDP has profiles, but activation coverage is low.”

It may mean: Customer records exist in the platform, yet too few can be matched or transmitted to useful destinations. Identity keys, permissions, destination integrations, and audience eligibility may be the real constraints.

“The TC string is missing on part of EEA traffic.”

It may mean: Some European requests are reaching adtech vendors without the expected TCF signal. Buyers may suppress bidding, and the publisher may face both monetization loss and privacy concerns.

“SKAN is returning postbacks, but the conversion values are too coarse.”

It may mean: Apple’s framework is technically functioning, but privacy thresholds, delayed reporting, or the conversion schema provide insufficient detail for the optimization the team wants.

“The RMN has closed-loop sales, but the result is not portable.”

It may mean: The retailer can connect ads to transactions inside its own environment, but its definitions, identity, and attribution methodology do not translate cleanly to other retailers or open-web channels.

“NTB is up, but we have not proved those customers are incremental.”

It may mean: More attributed buyers lacked recent purchase history on that platform. They may still have purchased elsewhere, returned after the lookback period, or converted without the advertising.

“We will makegood the underdelivery in the next flight.”

It may mean: The media owner did not satisfy the original delivery commitment and proposes replacement inventory later. The buyer should confirm that timing, audience, format, and value remain comparable.

“The clean-room match rate is high because the denominator is eligible records.”

It may mean: The headline percentage excludes records that lacked consent, valid identifiers, or required geography. The result may be correct, but it does not represent the full uploaded customer population.

Net Net

Adtech and martech language is difficult because media auctions, identity systems, customer-data architecture, privacy law, measurement science, and commercial terms intersect in the same workflow. A familiar word such as impression, match, conversion, or direct may have a precise system-dependent meaning, and several platforms may calculate it differently.

Which event is being counted here: opportunity, request, bid, win, served ad, render, viewable impression, click, or conversion?
What is the identity unit: cookie, device, authenticated person, account, or household?
Which platform or log is the source of truth, and what counting rule does it apply?
Is this inventory path direct, reseller-mediated, curated, or represented through a private deal?
Which auction mechanics, floor, fee stack, and bid-shading rules affect the price?
Is the result attributed, modeled, or experimentally incremental?
What attribution window, deduplication rule, and conversion definition produced the reported outcome?
Which consent, platform-policy, or jurisdictional signal controls whether this data can be used?
Where is the campaign in the request-to-render or event-to-activation workflow?
Which specialist function has authority over the disputed definition: ad operations, data engineering, measurement, privacy, media buying, or publisher yield?
What technical evidence supports the conclusion: bidstream samples, ad-server logs, schain nodes, event payloads, match files, or experiment results?
Which assumption, threshold, or identity loss would materially change the recommendation?

Real fluency does not come from memorizing every acronym. It comes from recognizing which system, identity, counting rule, permission, and commercial mechanism sit behind the term, then asking the question that makes those assumptions visible.

Media & Entertainment Industry Lingo >