Glossary

Disapproval means the platform refused to serve an ad or asset because it violates policy or technical requirements. Fix the cited reason, resubmit, and check whether the issue is account-level (billing, verification) or ad-level (creative, destination, category).

Note: Policy Watch tracks platform rule changes; disapprovals often follow updates you missed.

Ad fatigue occurs when the same audience sees the same creative too often: CTR drops, CPM may rise, frequency climbs. Refresh creative hooks, expand audience, or rotate formats before killing a “winning” ad set. Monitor frequency and thumb-stop / hook rate on video.

Note: Frequency cap is a lever; creative refresh is the durable fix.

Ad format is the creative unit type — single image, video, carousel, collection, responsive search ad, etc. Each format has spec requirements (dimensions, duration, text limits). Format choice should follow placement and funnel stage: video for awareness, static + strong CTA for retargeting.

Note: Validator checks uploaded files against live spec rows.

Ad Library is a public archive of active and inactive ads (Meta Ad Library, Google Ads Transparency Center, TikTok Commercial Content Library). Operators use it for competitive research, policy compliance checks, and verifying competitor messaging — not for precise spend or performance data.

Note: Political and issue ads have richer disclosure in Meta Ad Library.

Ad placement is where an ad can appear — Feed, Stories, Reels, Search results, YouTube in-stream, etc. Placement choice affects specs, CPM, and user intent. Manual placement selection vs. Advantage+ / automatic placements is a tradeoff between control and delivery volume.

Note: See PaidScope Placements map for Google surface ↔ campaign type relationships.

Ad Rank determines whether your ad shows and in which position in auction-based channels (especially Google Search). It combines bid, expected CTR, ad relevance, and landing page experience (Quality Score components). Higher Ad Rank can win a higher position at lower CPC than a competitor with higher bid but weaker relevance.

Note: Display and social auctions use similar relevance signals under different names.

An ad set (Meta/TikTok terminology) is the targeting, placement, budget, and schedule layer below campaign and above individual ads. Google Ads uses ad groups instead. Changes at ad set level (audience, geo, optimization event) affect all ads inside it — duplicate ad sets to test audiences cleanly.

Note: LinkedIn uses Campaign Group → Campaign → Ad; Amazon uses campaign → ad group.

Advantage+ creative (Meta) automatically tests combinations of images, videos, headlines, and primary text across placements. It can improve efficiency when you supply diverse assets — but weak inputs get recombined at scale. Monitor placement and creative breakdowns; turn off when brand control or legal review requires fixed copy.

Note: Formerly “Dynamic Creative” on some Meta surfaces — naming varies by ad account version.

Aspect ratio is width ÷ height of the creative frame (e.g. 1:1 square, 4:5 vertical feed, 9:16 Reels/Stories, 16:9 YouTube). Wrong ratio triggers cropping, letterboxing, or disapproval — the Validator checks uploaded files against live spec tables.

Note: Safe zones matter on 9:16: keep logos and CTAs inside platform “safe area” guides.

An attribution window is the lookback period after an ad interaction during which a conversion can be credited to that ad. Click windows (7-day click is common on Meta) and view windows (1-day view) can both count — inflating reported ROAS if finance only recognizes click-through. Align platform defaults with your MMM or incrementality study before scaling.

Note: Shortening windows drops reported conversions without changing actual business outcomes.

Bid strategy is how a platform sets auction bids to hit your goal — manual CPC/CPM, target CPA/ROAS, maximize conversions, or lowest cost. Automated strategies need conversion volume to learn; switching strategies or budgets mid-flight resets learning on Meta and TikTok. Match strategy to funnel stage: Maximize Conversions for bottom-funnel, Target Impression Share for brand defense on Search.

Note: “Lowest cost” is not the same as “lowest CPA” — it minimizes cost per result within delivery constraints, not absolute efficiency.

Brand lift measures upper-funnel impact (awareness, consideration, intent) via survey holdouts exposed vs. unexposed to ads. Platforms (Meta, YouTube, TikTok) and vendors run lift studies when impression volume supports statistical significance. Use when CTR/CVR miss the story on awareness campaigns.

Note: Minimum impression thresholds apply — small tests may be inconclusive.

Call to action is the explicit next step you ask the user to take — Shop Now, Learn More, Sign Up. Platform CTA buttons and ad copy should align with landing page promise. Mismatch (ad says 50% off, page shows full price) kills CVR and can trigger policy disapprovals.

Note: Some placements allow custom CTA text; others use fixed button labels.

Campaign objective tells the platform what to optimize for — awareness, traffic, leads, app installs, sales, etc. Objective locks available placements, bid strategies, and optimization events. Picking the wrong objective (e.g. Traffic when you need Purchases) is a common ROAS leak.

Note: Google uses campaign type + conversion goal; naming differs but the principle is the same.

A click-through conversion is credited when a user clicks an ad and converts within the click attribution window. It is the closest analog to last-click in platform reporting and usually the baseline for Search efficiency — but still platform-defined, not your ERP truth.

Note: Do not confuse CTA (attribution) with CTR (click-through rate) — different acronyms, same letters.

Click-through rate (CTR) is the share of impressions that earn a click: CTR = clicks ÷ impressions (often shown as a percentage). It signals creative/message fit and auction competitiveness on click-based inventory (Search, many social feed units). Low CTR can raise effective CPC in auctions that weight expected CTR; high CTR with weak downstream conversion may mean the ad over-promises.

Note: Industry CTR varies wildly by channel — compare against your own historical baselines and placement mix, not generic blog benchmarks.

Conversion API sends server-side conversion events from your backend to ad platforms (Meta CAPI, TikTok Events API, LinkedIn CAPI), complementing browser pixels blocked by ATT or ad blockers. It improves match rates and deduplicates with pixel when both fire the same event_id. Required for reliable post-iOS14 Meta reporting.

Note: Implement deduplication — double-counting inflates reported ROAS.

Conversion rate is conversions divided by clicks (or sometimes sessions, depending on the report): CVR = conversions ÷ clicks. It measures post-click efficiency — a strong CTR with weak CVR often points to landing page, offer, or audience mismatch rather than ad copy alone.

Note: Platform “conversions” follow each network’s attribution window and counting rules.

Cost per action (often used interchangeably with cost per acquisition) is spend divided by counted conversions: CPA = spend ÷ conversions. It is the primary efficiency metric for lead-gen and purchase campaigns when the conversion event is well-defined and stable.

Note: Define the conversion event explicitly (purchase vs. lead vs. signup) before comparing CPA across campaigns.

Cost per click (CPC) is average spend divided by clicks: CPC = spend ÷ clicks. It is the primary efficiency metric for traffic and many lead-gen campaigns on click-priced auctions (Search, some social objectives). CPC rises when competition, audience narrowness, or low Quality Score / relevance increase your auction cost.

Note: “CPC cap” bidding sets a ceiling; actual CPC can still vary click-by-click within the cap.

Cost per impression is spend divided by impressions — mathematically equivalent to CPM ÷ 1,000 when both use the same impression definition. Operators usually say CPM for awareness buys and “cost per impression” when reconciling billing lines or non-standard inventory.

Note: Synonym confusion with CPM is common; always check whether the report uses 1,000-impression units.

Cost per mille (CPM) is what you pay for one thousand ad impressions. Formula: CPM = (spend ÷ impressions) × 1,000. Use CPM when buying or comparing awareness inventory (display, video, some social reach buys) — not when your goal is clicks or conversions, where CPC or CPA is the better lens. A $12 CPM on one platform is not automatically “more expensive” than $8 elsewhere if viewability, audience quality, or placement differ.

Note: Platforms report CPM in the ads manager UI; auction CPM and billed CPM can diverge when pacing, frequency caps, or learning phase apply.

Cost per view is spend divided by counted video views (platform-defined view threshold — often 30 seconds or complete). CPV is common on YouTube awareness and some social video buys. Compare CPV only within the same platform and view definition.

Note: YouTube TrueView CPV differs from CPM-priced bumper or non-skippable.

Creative testing compares variants (hooks, formats, offers) under controlled conditions to find winners. Structure tests with one variable changed per cell; avoid Advantage+ blending variants before you have readouts. Document learnings — “UGC hook B beat studio A on TikTok” — for the next flight.

Note: Statistical significance needs enough spend per variant — don’t call winners on 48h / $50.

A custom audience is a targeting segment you define from your own data: website visitors, app users, customer lists, or engagement (video viewers, lead forms). Platforms match hashed identifiers to users. Quality beats size — a 5k purchaser seed often outperforms a 500k all-visitors list for prospecting lookalikes.

Note: Refresh lists after major sales; stale seeds degrade lookalike quality.

Customer acquisition cost is total sales & marketing cost to acquire a customer — often broader than CPA (includes non-platform spend, promos, agency fees). Paid media CPA is a subset. Compare CAC to LTV before scaling prospecting; platform CPA alone can look healthy while blended CAC fails payback.

Note: Define “customer” consistently (first order vs. subscriber vs. trial activation).

Daily budget is the average amount a campaign or ad set may spend per calendar day on a given platform. Spend can exceed the daily budget on high-traffic days because networks pace delivery across the week — it is a pacing target, not a hard invoice cap. Raising daily budget resets learning on some Meta and TikTok objectives; lowering it can throttle delivery mid-day.

Note: Use lifetime budget when flight dates are fixed (events, promos); use daily budget for always-on performance.

Dayparting restricts ads to specific days or hours (ad schedule). Useful for B2B weekday windows, restaurant lunch hours, or live-event promos. Over-aggressive dayparting can starve learning phase volume; check hourly conversion reports before cutting nights/weekends.

Note: Google Ads ad schedule bid adjustments; Meta lifetime budget + manual pause is cruder.

A demand-side platform is software advertisers use to buy programmatic inventory across many SSPs and exchanges from one interface. DSPs handle bidding, frequency caps, creative rotation, and audience segments. Examples: Google DV360, The Trade Desk, Amazon DSP. Choosing a DSP affects data integrations, fees, and inventory access.

Note: Self-serve social platforms (Meta, TikTok) are not classic DSPs but compete for the same budget.

Engagement rate is interactions (likes, comments, shares, saves, clicks) divided by impressions or reach — varies by platform definition. Useful for organic and paid social creative diagnostics; not a substitute for conversion metrics on performance campaigns.

Note: Do not compare engagement rate across platforms — denominators differ.

First-party data is information collected directly from your customers or site/app users (CRM emails, purchase history, logged-in behavior). It is the most durable targeting signal as third-party cookies deprecate. Upload hashed lists to platforms (Customer Match, Custom Audiences) and pair with Conversion API for measurement.

Note: Consent and privacy policy disclosures are required before collection and ad use.

Frequency is the average number of times each reached user saw your ad in a period: frequency = impressions ÷ reach. High frequency on a small audience can signal audience exhaustion, bid cap issues, or over-narrow targeting — especially on social and video.

Note: Reach and frequency caps are the primary levers when frequency climbs without incremental conversions.

Frequency cap limits how many times one user sees your ad in a time window (e.g. 3 impressions per 7 days). Caps reduce ad fatigue and wasted reach in remarketing. Too low a cap on small audiences can under-deliver budget; too high allows banner blindness.

Note: Meta campaign-level frequency controls differ from reach-and-frequency buying.

Geotargeting restricts ad delivery to geographic areas — country, region, metro, radius around a point. Local businesses need tight radius; national brands layer exclusions (non-service states). Geo bid modifiers on Search adjust CPC by location performance.

Note: Verify compliance with local ad laws (gambling, alcohol, healthcare) per geo.

Header bidding (or prebid) lets publishers offer inventory to multiple SSPs simultaneously before calling their ad server, increasing competition and CPM. For advertisers, it can mean more auction duplication and frequency — use reach/frequency tools and deal IDs when buying premium publisher inventory.

Note: Server-side header bidding reduces page-weight vs. client-side wrappers.

An impression is counted when an ad is served and enters the measurable viewport (platform rules differ on viewability thresholds). Impressions are the denominator for CPM and CTR; rising impressions with flat clicks usually means creative or audience fatigue, not “more reach for free.”

Note: Viewable impressions (Active View, etc.) may be lower than served impressions — compare metrics apples-to-apples in reporting.

Incrementality measures the causal lift from ads — conversions that would not have happened without the spend — via holdout tests, geo experiments, or MMM. Platform-reported ROAS often overstates incrementality because of view-through and organic brand search. Use incrementality before cutting “expensive” upper-funnel channels.

Note: Holdout tests require statistical power — small budgets need longer test windows.

Landing page is the destination URL after the click — must match ad message, load fast on mobile, and track conversions. Quality Score and Meta landing page experience diagnostics penalize slow, irrelevant, or thin pages. One campaign → one clear LP per offer reduces attribution noise.

Note: Post-click experience is half the auction — don’t optimize ads in isolation.

Learning phase is the period after significant campaign edits when the platform’s delivery system recalibrates (typically ~50 optimization events per ad set on Meta). Performance fluctuates; avoid unnecessary budget or audience changes during learning. “Learning limited” means insufficient volume — widen audience or raise budget.

Note: Google Smart Bidding has similar calibration after major changes.

Lifetime budget is a fixed spend cap for a scheduled flight (start/end dates). Platforms pace delivery to exhaust the budget by the end date — unlike daily budget, you are committing to total spend over the flight. Common for promos, product launches, and TikTok/Meta burst tests.

Note: Underspend at period end means delivery could not win auctions or audiences were too narrow — not always “unused budget to refund.”

A lookalike audience is a prospecting segment modeled from a seed list (purchasers, leads, site visitors) so the platform finds new users with similar attributes. Quality depends entirely on the seed: polluted seeds (all visitors, one-SKU buyers) produce broad, expensive audiences. Refresh seeds after major promo periods; exclude existing customers when the goal is new-customer acquisition.

Note: Meta % lookalike sizing trades reach for similarity; smaller % is not always “better” if CPM becomes prohibitive.

Marketing efficiency ratio is total revenue divided by total marketing spend across channels (often blended, not platform-attributed): MER = revenue ÷ ad spend. It is a sanity check when platform ROAS disagrees — useful for founders, not for intraday bid tweaks.

Note: MER includes non-platform costs only if you define it that way — document the formula for your team.

Media mix modeling uses regression on historical spend and outcome data (often weekly) to estimate each channel’s contribution, including offline and organic effects. MMM complements platform attribution when cookies break and for budget allocation across Search, social, TV, and retail media. Not for daily bid changes.

Note: Requires 2+ years of data and expert setup — not a DIY spreadsheet for most teams.

Performance Max is a Google Ads campaign type that serves across Search, Shopping, Display, Discover, Gmail, YouTube, and Maps from one asset group and conversion goal. You do not pick individual placements — you supply assets and signals; Google maps them to eligible surfaces.

Note: Use placement reporting and brand exclusions when PMax delivery skews to low-intent inventory.

Programmatic advertising is the automated buying and selling of ad inventory via software (DSPs, SSPs, ad exchanges) rather than manual insertion orders. Real-time bidding (RTB) sets price per impression based on audience, context, and bid. Operators use programmatic for scale, data targeting, and frequency control — but must watch viewability, brand safety, and supply-path transparency.

Note: PMP (private marketplace) deals sit between open exchange and direct IO buys.

Quality Score is Google Ads’ composite relevance signal (expected CTR, ad relevance, landing page experience) expressed as 1–10 at keyword level in Search. Higher QS lowers auction CPC for the same position — it is not a “grade” on creative beauty, but on match between query, ad, and page.

Note: Display and Video use different diagnostic labels (e.g. ad strength) — do not expect QS on every Google campaign type.

Reach is the number of unique users who saw your ad at least once in a reporting period. Reach = impressions ÷ frequency (approximately). Low reach with high spend signals narrow targeting or high frequency; reach buying (CPM awareness objectives) optimizes for unique users over clicks.

Note: Platform reach definitions differ on viewability thresholds and deduplication windows.

Real-time bidding is an auction that completes in milliseconds when an ad impression becomes available: the SSP sends a bid request, DSPs evaluate the user and context, and the highest eligible bid wins. Most display and video open-exchange inventory is RTB-priced. CPM varies by audience quality, viewability, and competition.

Note: Not all programmatic is RTB — guaranteed and fixed-price PMP deals skip the open auction.

Remarketing (Google’s term; Meta says retargeting) shows ads to people who previously visited your site or used your app. Lists are built from pixel/tag events or uploaded customer data. Remarketing usually converts at higher CVR but higher CPM — cap frequency and exclude recent purchasers when prospecting budget is tight.

Note: Consent required in EU/EEA before personalized remarketing.

Retargeting shows ads to people who already interacted with your site, app, or customer list. It typically lifts conversion rate at higher CPC/CPM because the audience is warm — but list freshness, frequency caps, and consent (GDPR/CPRA) govern whether you can legally and effectively reach them.

Note: Platform naming: Google “remarketing,” Meta “custom audiences,” TikTok “retargeting.”

Return on ad spend (ROAS) is revenue attributed to ads divided by ad spend: ROAS = revenue ÷ spend (sometimes expressed as a ratio like 3.2× or 320%). It answers “did this spend pay for itself?” for ecommerce and revenue-tracked funnels. Break-even ROAS depends on margin — a 4× ROAS target for a 25% margin product is not the same as for a 60% margin SKU.

Note: Platform ROAS uses each network’s attribution window and model; cross-check with your source-of-truth (shopify, GA4, MMM) before reallocation decisions.

SKAdNetwork is Apple’s privacy-preserving framework for attributing iOS app install campaigns without IDFA. Advertisers receive delayed, aggregated postbacks with limited granularity (campaign ID buckets, no user-level data). Meta, Google, TikTok, and MMPs map SKAN IDs to campaigns — expect modeled gaps vs. pre-ATT reporting.

Note: SKAN 4.0+ adds more conversion fine/coarse values but still limits optimization signals.

Skippable in-stream ads play before, during, or after YouTube content; viewers can skip after five seconds. Billing models vary (CPV after 30s watch or skip threshold depending on campaign type). Non-skippable and bumper units have hard duration caps — do not traffic 30s masters into 6s bumper slots.

Note: Companion banners and CTAs differ between YouTube campaigns and Demand Gen — check the active campaign type.

Split testing runs two or more variants with equal conditions to isolate what drives lift. Platform native tools (Meta Experiments, Google Experiments) handle traffic split; for clean tests change one variable, run long enough for conversions, and pre-register success metrics.

Note: Avoid editing both variants mid-test — invalidates results.

A supply-side platform helps publishers sell ad inventory programmatically by connecting to multiple DSPs and maximizing yield per impression. SSPs manage floor prices, ad quality filters, and auction mechanics. Publishers and app developers use SSPs; media buyers encounter them indirectly through DSP bid streams.

Note: Header bidding lets publishers call multiple SSPs before the ad server auction.

Target CPA is automated bidding to achieve an average cost per acquisition at your target (e.g. $40 lead). Google, Meta, TikTok, and LinkedIn offer CPA-target variants with different minimum conversion thresholds. Sudden target CPA cuts often cause learning resets and volume drops.

Note: Define “acquisition” consistently — lead vs. qualified lead vs. purchase.

Target ROAS is a Smart Bidding strategy where Google (or other platforms) sets bids to hit your target return on ad spend (e.g. 400% = 4×). Requires conversion value tracking and enough conversion volume. Setting tROAS above historical reality throttles delivery; below reality wastes margin.

Note: Break-even ROAS = 1 ÷ profit margin — know yours before setting tROAS.

A third-party cookie is set by a domain other than the site the user is visiting — historically used for cross-site tracking, retargeting, and attribution. Browsers and platforms are phasing them out; operators shift to first-party data, server-side tagging, modeled conversions, and platform-native audiences.

Note: Chrome Privacy Sandbox and Safari ITP already limit most third-party cookie use.

Thumb-stop ratio is the share of impressions where a user paused on your video ad (Meta and some video platforms expose a related metric). It is a creative hook diagnostic — low thumb-stop with acceptable CPM means the first frame is not earning attention.

Note: Not available uniformly across Google Search/Display — primarily social/video tooling.

A tracking pixel is a snippet (JavaScript or image beacon) placed on a site or app that fires when a user takes an action, sending data to an ad platform for attribution and audience building. Pixels depend on browser storage — pair with Conversion API and consent mode for resilient measurement.

Note: Platform naming: Meta Pixel, Google tag (gtag/GTM), LinkedIn Insight Tag, TikTok Pixel.

A view-through conversion is counted when a user sees an ad (impression) but does not click, then converts within the platform’s view attribution window. View-through credit inflates reported ROAS on video and display-heavy mixes — compare click-through and view-through columns before reallocating budget. Finance teams often exclude view-through from payback models.

Note: Default windows differ: Meta 1-day view vs. Google Ads view-through settings per conversion action.

Viewability measures whether an ad had a chance to be seen — typically 50% of pixels in view for at least one second (display) or two seconds (video), per MRC/IAB standards. Served impressions exceed viewable impressions on long-tail Display and Audience Network; compare CPM on viewable basis when auditing partner inventory.

Note: Platforms report “viewable impressions” differently — export raw logs for audits, not just dashboard totals.