Advertising glossary
Every metric on this page gets the same treatment: what it measures, the formula written out where there is arithmetic to write, a worked example with numbers you can check, benchmarks with the page they came from wherever a published source exists, and an honest account of where the number misleads. No figure appears here without a source you can open, and where we could not find one, the figure is absent rather than invented.
Every term
Account structure
Account structure is how an advertising account is arranged into campaigns, ad groups or ad sets, and ads. Under automated bidding it stopped being a filing decision and became an allocation one: structure decides what the platform chooses between.
Ad fatigue
Ad fatigue is the decline in response that follows repeated exposure of the same audience to the same creative. It shows as a falling click-through rate against rising frequency, and the cost per result usually moves last.
Ad format
An ad format is the structural type of the unit — single image, video, carousel, collection, responsive text — with its own dimensions, limits and mechanics. The format decides what an ad is able to say before any creative decision is made.
Ad group
An ad group is the container between a campaign and its ads in Google Ads and Microsoft Advertising. It holds keywords and the ads that serve against them, which makes it the unit at which a search account's relevance is decided.
Ad Rank
Ad Rank is the value Google Ads computes for each ad in each auction to decide whether it shows and in what position. It multiplies the bid by quality estimates, which is why a more relevant ad can outrank a higher bid and pay less for the position.
Ad set
An ad set is the container between a campaign and its ads on Meta, holding the audience, the placement selection, the schedule and — depending on the campaign's budget arrangement — its own budget. It is a unit of delivery rather than of meaning.
AOV
Average order value is revenue divided by the number of orders over the same period. It sits inside almost every other metric in the account — it is the revenue half of ROAS and the ceiling behind every acquisition target — and it is skewed enough that the average rarely describes a typical order.
Approval workflow
An approval workflow is the process a proposed change passes through before it takes effect. The check can be a person or a rule written in advance, and the useful versions record which one cleared each change.
Asset group
An asset group is the creative container in a Google Ads Performance Max campaign: text, images, video, logos and a signal about who the customer is, assembled by the platform across every surface the campaign reaches.
Attribution window
An attribution window is the period after an ad interaction during which a conversion is still credited to it. Its length is a setting rather than a fact, and changing it changes every attributed figure in the account without anything about the advertising changing.
Attribution
Attribution is the rule an account uses to decide which advertising gets credit for a conversion. It observes what happened and distributes credit by a convention somebody chose, which is why different models produce different numbers from the same week.
Ad auction
An ad auction is the process that decides which advertiser wins an impression and what they pay. It ranks on expected value rather than on bid alone, which is why a more relevant ad can beat a higher bid and pay less for the privilege.
Audience overlap
Audience overlap is the share of one audience's members who also belong to another. It matters less for auction competition than folklore suggests and more for reporting, because two overlapping audiences cannot be compared against each other.
Audience signal
An audience signal tells a Google Ads campaign who the advertiser believes the customer is, as a starting point for the model rather than a restriction. The campaign can and will deliver outside it, which is the property that distinguishes a signal from targeting.
Audience
An audience is the set of people a campaign is allowed to reach, defined by who they are, what they have done, or what a platform infers about them. It sets eligibility rather than delivery: the bidder still chooses who inside it actually sees the ad.
Automated bidding
Automated bidding hands the per-auction bid to the platform, which predicts the value of each impression and bids accordingly. The advertiser still sets the objective and the target, and those two inputs decide almost everything about the result.
Bid cap
A bid cap tells the platform never to bid above a stated figure in any single auction. Meta and TikTok offer it under that name; Google Ads reaches it through a maximum cost-per-click. It binds every auction rather than an average.
Bid strategy
A bid strategy is the rule a campaign uses to decide what to bid in each auction. The choice is really a choice of objective: every strategy optimises something faithfully, and picking the wrong one produces exactly what it was asked for.
Bid
A bid is the maximum an advertiser is willing to pay for the outcome an auction is selling — a click, a thousand impressions, a conversion. Under automated strategies the advertiser sets a target instead, and the per-auction bid is computed from it.
Blended CAC
Blended customer acquisition cost divides everything spent on acquisition in a period by every new customer won in it, crediting no channel for anything. It is immune to the double-counting that per-channel attribution produces, and it has no channel dimension at all.
Break-even ROAS
Break-even ROAS is the return on ad spend at which a campaign covers the goods it sold and the advertising that sold them, and contributes nothing beyond that. It is the reciprocal of contribution margin, so it comes from your own accounts rather than somebody else's.
Broad match
Broad match makes a keyword eligible for searches the platform judges related to it, including searches that share no words with the keyword. It is the widest reach available and it depends entirely on what else in the account is doing the filtering.
Budget allocation
Budget allocation is the division of a total across channels, campaigns and periods. The reported figures pull it toward whatever captures existing demand, so the decision is partly about resisting the numbers that are easiest to read.
CAC
Customer acquisition cost is everything spent to win a customer divided by the number of customers won. Unlike a platform-reported cost per acquisition, it includes agency fees, creative production, tooling and sales salaries, which is why it is always the larger number.
Campaign budget optimisation
Campaign budget optimisation holds one budget at campaign level and lets the platform distribute it across the ad sets beneath, moving money toward whatever is performing. Meta's version is the widely known one; the same pattern exists on other platforms under other names.
Campaign
A campaign is the top-level container in an advertising account, holding the objective, the budget arrangement, the bid strategy and the schedule. What sits at this level is what the platform optimises across, which is why structure decisions are made here.
Carousel ad
A carousel ad presents several cards in sequence, advanced by the viewer. The swipe is a filter: fewer people see the later cards, and the ones who do have chosen to, which is what the format trades reach for.
Click-through conversion
A click-through conversion is credited when somebody clicked an ad and converted within the click window. It carries the strongest evidence available in ordinary attribution, which is not the same as evidence that the ad caused the sale.
Click
A click is one counted interaction with an ad. Platforms count several different sets of interactions under the same word, so the figure in a report is only interpretable once the set it counted is named.
Contribution margin
Contribution margin is what a sale leaves after every cost that varies with it — goods, shipping, payment fees, returns. It is the money available to pay for advertising and fixed costs, which makes it the ceiling behind every acquisition target in the account.
Conversion rate
Conversion rate divides conversions by the traffic that produced them — clicks, sessions or visitors, depending on who is asking. The denominator is the argument, and stating it is what makes the figure comparable to anything.
Conversion
A conversion is an event an account has configured itself to count — a purchase, a form, a call, an install, a page view. Everything downstream reads from that choice: the reporting, the bidding, the optimisation and the budget.
Conversions API
A conversions API is the server-side path for sending conversion events to an advertising platform, bypassing the browser entirely. Meta's name for it is the widely used one; every major platform offers an equivalent.
Cost cap
A cost cap holds the average cost per result at or below a stated figure, leaving the bidder free to pay more in one auction and less in another. TikTok calls it Cost Cap; Meta now labels the idea a cost per result goal.
Cost per result
Cost per result is spend divided by the count of whatever event the campaign was set to optimise for. The event changes with the objective, which makes the column convenient inside one campaign and meaningless when compared across campaigns with different objectives.
CPA
Cost per acquisition is what one conversion cost: ad spend divided by the number of conversions attributed to it. The conversion can be a sale, a lead, a signup or a booked call, and the figure means nothing until you say which.
CPC
Cost per click is what one click cost: ad spend divided by the number of clicks the platform counted. Which clicks it counted is most of the argument, because an all-clicks figure and a link-clicks figure from the same campaign are different numbers wearing the same three letters.
CPE
Cost per engagement is spend divided by the number of interactions the platform counted — reactions, comments, shares, saves, expansions, taps on a profile. The set of events in the denominator differs by platform and by placement, and widening it always lowers the price.
CPI
Cost per install is app-campaign spend divided by the number of installs attributed to it. The install is the first event in an app funnel and rarely the one that matters, which is why the figure is read beside what happens after the app opens.
CPL
Cost per lead is ad spend divided by the number of leads it produced. A lead is whatever the account decided to count — a form, a call, a chat, a downloaded file — so the figure means nothing until the event behind it is named.
CPM
Cost per mille is the cost of a thousand ad impressions: spend divided by impressions, times a thousand. The mille is Latin for thousand, which is why the abbreviation has an M where the metric has none.
CPV
Cost per view is video spend divided by the number of views the platform counted. Each platform defines a view differently — two seconds, three seconds, fifteen seconds, or the whole thing — so the figure is only comparable inside one platform's own definition.
Creative rotation
Creative rotation is the deliberate replacement of ads on a cadence, rather than in response to a decline that has already happened. It is a production schedule as much as a media setting, and its right frequency is set by how fast the account accumulates exposure.
Creative
Creative is the ad itself — the image, the video, the copy, the offer as it is stated. As targeting has moved to delivery models, creative has become the main thing an advertiser still controls and the main way a platform is told who an ad is for.
CTR
Click-through rate is the share of impressions that produced a click: clicks divided by impressions, expressed as a percentage. It measures whether an ad earned attention, and nothing about what happened next.
Custom audience
A custom audience is built from data the advertiser holds — a customer list, site or app activity, engagement on the platform — rather than from platform inference. Meta's name for it is the widely used one; every major platform has an equivalent.
Customer Match
Customer Match is Google Ads' name for targeting built from a customer list the advertiser uploads. The identifiers are hashed before sending, matched against signed-in accounts, and available across the surfaces the policy permits.
Daily budget
A daily budget is the average amount a campaign is instructed to spend each day. Platforms treat it as a pacing target rather than a hard ceiling, spending more on days with better opportunity and averaging out over a period.
Data-driven attribution
Data-driven attribution distributes credit across the touchpoints before a conversion using patterns a platform learned from journeys that converted and journeys that did not. It is better than a fixed rule and it is still attribution.
Dayparting
Dayparting restricts or weights advertising delivery by hour of day and day of week. It is worth doing where the business genuinely differs by hour — a sales team's opening times, a kitchen's service window — and is frequently illusory where the only evidence is a small-sample performance report.
Demographic targeting
Demographic targeting limits or weights delivery by attributes such as age, gender, parental or household status. Most of those attributes are inferred rather than known, which makes the targeting less precise than its labels suggest.
Dynamic creative
Dynamic creative takes components an advertiser supplies — images, videos, headlines, descriptions — and assembles combinations per auction, choosing what to show each viewer. The advertiser supplies the range; the platform supplies the arrangement.
eCPC
Effective cost per click is total spend divided by clicks delivered, computed after the fact whatever the campaign was bought on. The same abbreviation named a Google Ads bidding feature, Enhanced CPC, which has since been retired.
eCPM
Effective CPM is the cost of a thousand impressions worked backwards from what was actually spent and delivered, whatever pricing model the campaign was bought on. It exists so that buys priced on clicks, views or actions can be compared on one axis.
Engagement rate
Engagement rate divides counted interactions by impressions, by reach or by followers, depending on the platform. Both halves of the fraction vary, so the same campaign can report several different engagement rates and all of them are correct.
Enhanced conversions
Enhanced conversions is Google Ads' feature for sending hashed customer data alongside a conversion so it can be matched to a signed-in account the browser could not identify. It improves matching rather than measuring anything new.
Enhanced CPC
Enhanced CPC was a Google Ads bid strategy that took a manual cost-per-click bid and raised or lowered it in each auction by the predicted chance of conversion. Google has retired it and moved the campaigns that used it to manual cost-per-click bidding.
Exact match
Exact match makes a keyword eligible for searches that mean the same thing as it. It stopped requiring the same words some time ago, so the tightest match type available is tighter than the others and looser than its name.
Exclusion audience
An exclusion audience is a list of people a campaign is forbidden to reach, applied alongside whatever it targets. Exclusions encode facts a platform cannot infer, which is why they remain the most reliable audience work in an account.
Flight dates
Flight dates are the start and end of a campaign's run. The dates are not only a schedule: with a fixed total they set the daily spend rate, which decides how far down the inventory curve the campaign has to reach.
Flighting
Flighting is scheduling advertising in defined bursts with deliberate gaps between them, rather than running continuously. It concentrates a budget where demand is, and it accepts that recognition decays during the gaps.
Frequency
Frequency is impressions divided by reach — how many times the average person in the audience saw the ad. The average conceals a long tail, so a campaign at a comfortable mean can be showing a small group the same ad dozens of times.
Geo lift test
A geo lift test measures incrementality by running advertising in some geographic regions and withholding it from comparable ones, then comparing outcomes. It works on any channel, including ones with no targeting or tracking at all.
Geo-targeting
Geo-targeting restricts or weights delivery by location. The consequential setting is not the boundary but the presence option, which decides whether a location means people who are there or people showing interest in it.
Guardrail
A guardrail is a limit set in advance that prevents an account doing something costly, regardless of what any optimisation would otherwise choose. It is a backstop rather than a target: it should rarely bind, and when it does something has gone wrong.
Hold rate
Hold rate divides a longer-threshold view count by a shorter one: the share of people who got past the opening and were still there later. On Meta the two counts are usually ThruPlay over three-second video plays. It measures the body of the ad rather than its hook.
Holdout test
A holdout test withholds advertising from a randomly chosen portion of an audience and compares outcomes against the portion that received it. Random assignment is what makes it the cleanest available measure of what advertising caused.
Hook rate
Hook rate divides three-second video views by impressions: the share of people served the ad who stayed past the opening. It is built from Meta's three-second video plays, or TikTok's two-second equivalent, and is a practitioner metric rather than a platform one.
Hook
The hook is the opening of an ad — the first frame, the first words, the first movement — whose only job is to earn the next second of attention. On scrolled surfaces most of the audience is lost there, which makes it the highest-leverage part of the creative.
Impression share lost to budget
Impression share lost to budget is the estimated share of eligible impressions an account did not receive because its budget was exhausted or its pacing restricted delivery. It is the closest thing in advertising to a priced estimate of unspent demand.
Impression share lost to rank
Impression share lost to rank is the estimated share of eligible impressions an account did not win because its ad rank was too low. Rank combines the bid with the platform's quality estimate, so the remedies include several that cost nothing.
Impression share
Impression share is the share of the auctions you were eligible for that you actually won. It is the only common metric that measures the size of the opportunity rather than the size of the result, which is why it answers a question no performance metric can.
Impression
An impression is one delivery of an ad to a screen. It is the denominator of most rate metrics in advertising and it carries no claim that anybody saw the ad, which is why viewability exists as a separate measure.
Incrementality
Incrementality is the portion of results that would not have happened without the advertising. It is measured by withholding advertising from a comparable group and comparing outcomes, which is the only design that separates causing a sale from being present at one.
Keyword match type
A keyword match type controls how closely a search has to relate to a keyword before the ad becomes eligible. Google Ads and Microsoft Advertising offer broad, phrase and exact, and all three now match on meaning rather than on the words alone.
Last-click attribution
Last-click attribution credits a conversion entirely to the final ad interaction before it. It is the simplest rule, the most auditable, and the one that most reliably rewards whatever was present at the moment of decision.
Learning phase
The learning phase is the period after a material change during which an automated bid strategy rebuilds its predictions. Delivery is uneven and the reported cost is unrepresentative, which makes anything measured during it a poor basis for the next decision.
Lifetime budget
A lifetime budget sets the total a campaign may spend across a defined start and end date, leaving the platform free to distribute it unevenly across the flight. It suits fixed-window campaigns and it removes the daily control that some accounts depend on.
Link click
Link clicks count only the interactions with an ad that sent somebody to a destination, as distinct from the wider all-clicks figure that also counts reactions, comments, shares, profile taps and image expansions.
Lookalike audience
A lookalike audience is a set of people a platform believes resemble a seed audience the advertiser supplied. The model is only as good as the seed, and the percentage setting controls how far from it the platform is willing to go.
LTV:CAC ratio
The LTV:CAC ratio divides what a customer is worth over the relationship by what they cost to acquire. It answers whether acquisition is economic at all, and it says nothing about whether the business can afford the cash it takes to get there.
LTV
Lifetime value is what a customer contributes over the whole relationship, after the cost of serving them. Stated as revenue rather than gross margin it is not a number any acquisition decision can be made from, because revenue cannot pay for advertising.
Manual bidding
Manual bidding applies a figure the advertiser sets to every auction an entity enters, with no per-auction adjustment. It trades the platform's prediction for predictability, which is worth having on thin accounts and expensive on busy ones.
Maximise conversion value
Maximise conversion value is a Google Ads bid strategy that spends the budget pursuing the greatest total conversion value rather than the greatest number of conversions. It is only as good as the value the account sends back with each conversion.
Maximise conversions
Maximise conversions is a Google Ads bid strategy that spends the whole budget pursuing as many conversions as possible. Without a target attached it has no cost ceiling, so the budget becomes the only constraint and cost per conversion is whatever the auction produced.
Media plan
A media plan states what advertising is expected to produce, what it will cost and on what assumptions. Its value is less in the forecast than in making the assumptions explicit enough to be wrong in public.
MER
Media efficiency ratio divides all revenue in a period by all advertising spend in the same period, attributed to nothing. It is deliberately blunt: because it ignores attribution entirely, it cannot be inflated by platforms all claiming the same sale.
Modeled conversions
Modeled conversions are estimates a platform adds for conversions it believes happened but could not observe. They fill gaps left by consent, tracking restrictions and cross-device journeys, and they cannot be reconciled line by line against anything.
Multi-touch attribution
Multi-touch attribution is the family of approaches that distribute credit across several touchpoints rather than assigning it to one. The models are easy to describe and the identity resolution underneath them is the part that decides whether any of it works.
Naming convention
A naming convention is the agreed pattern for naming campaigns, ad groups, ad sets and ads. It is a reporting decision rather than an administrative one: the fields in a name are the dimensions an account can filter and group by.
Negative keyword
A negative keyword prevents an ad from showing for searches containing it. It is the one control in search advertising that refuses rather than influences, which is why maintaining the list matters more as matching has loosened.
Payback period
Payback period is the number of months of contribution it takes for a customer to repay what they cost to acquire. It is the cash-flow question behind every acquisition decision, and it binds long before any profitability ratio does.
Phrase match
Phrase match makes a keyword eligible for searches that contain its meaning, with additional words allowed around it. It no longer requires the keyword's literal words in order, which is the change that makes its name misleading.
Pixel
A pixel is a snippet of code on a website that reports visits and events back to an advertising platform. It was the standard way conversions were measured, and browser restrictions have made it the less reliable half of a pair.
Placement
A placement is the specific surface an ad appeared on — a feed, a story, an in-stream break, a partner site, an app. Prices and response differ enormously between them, so a figure blended across placements describes none of them.
POAS
Profit on ad spend divides the gross profit a campaign produced by what it cost, rather than dividing revenue by cost. It ranks campaigns by what the business keeps, which is a different order from the one revenue produces whenever margins differ across products.
Portfolio bid strategy
A portfolio bid strategy is a Google Ads arrangement where one shared strategy and target span several campaigns, ad groups or keywords, so the bidder can run one campaign above target and another below while the portfolio average holds.
Quality Score
Quality Score is a Google Ads diagnostic, reported per keyword on a one-to-ten scale, summarising expected click-through rate, ad relevance and landing page experience. Microsoft Advertising reports its own equivalent. It reports the estimate rather than being the estimate the auction uses.
Reach
Reach is the number of distinct people who saw an ad at least once in a period. It is an estimate rather than a count, it is bounded by the size of the addressable audience, and it is the number a launch or a brand campaign is actually buying.
Remarketing
Remarketing is advertising to people who have already interacted with a business — visited, enquired, bought, or engaged on a platform. Google Ads uses the word where most social platforms say retargeting, and in practice the two describe the same activity.
Responsive search ad
A responsive search ad is the Google Ads text format where the advertiser supplies a set of headlines and descriptions and the platform assembles a combination per query. Microsoft Advertising offers the same format; it is now the standard way search text ads are built.
Retargeting
Retargeting is advertising to people who have already visited a site or app, or engaged with a business on a platform. It is the same activity Google Ads calls remarketing, and it reports the most flattering figures in almost any account.
ROAS
Return on ad spend is the revenue attributed to a campaign divided by what that campaign cost over the same window. A ROAS of 3 means three dollars of attributed revenue for every dollar spent.
Safe zone
A safe zone is the region of an ad's frame where content will not be obscured by platform interface elements or lost to cropping. It differs by placement and by surface, which is why one asset rarely survives every placement a campaign buys.
Search term
A search term is the query a person actually entered, as distinct from the keyword that made an ad eligible for it. Since all match types now match on meaning, the search-term report is the only view of what an account is really buying.
Server-side tracking
Server-side tracking collects and forwards events from the advertiser's own infrastructure rather than from the visitor's browser. It recovers measurement the browser loses and makes the advertiser responsible for deciding exactly what is sent.
Smart Bidding
Smart Bidding is Google Ads' label for the subset of automated bid strategies that optimise toward conversions or conversion value using its per-auction prediction. It is a family name rather than a strategy, which is a frequent source of confusion.
Spend cap
A spend cap limits the total an account, campaign or period may spend, whatever that spend achieves. Meta calls it a spending limit, Google Ads an account budget, TikTok a lifetime cap. It is a financial control, not a performance one.
Static ad
A static ad is a single non-moving image with its accompanying text. It states one idea simultaneously rather than sequentially, which bounds what it can communicate and makes it the fastest and cheapest thing in an account to produce and test.
Target CPA
Target CPA is the cost per acquisition an automated bid strategy is told to aim for. It is an instruction rather than a cap: the bidder pursues the average, which means some conversions cost more and delivery stops where the target cannot be met.
Target impression share
Target impression share is a Google Ads bid strategy that bids to appear in a chosen share of eligible auctions, optionally at the top or absolute top of the page. It buys visibility rather than results, which makes it right for a narrow set of jobs.
Target ROAS
Target ROAS is the return on ad spend a value-based bid strategy is told to aim for. It works on the conversion value the account sends back, which means it is only as good as that value, and a target set above what the market allows suppresses delivery rather than raising returns.
ThruPlay
ThruPlay is Meta's video metric. It counts a play only when the video was watched to the end, or for at least fifteen seconds if it runs longer than that. It replaced a shorter threshold precisely because the shorter one counted people who had scrolled past.
Thumbstop rate
Thumbstop rate is the share of impressions where somebody stopped scrolling long enough to register the ad, computed from Meta's three-second video plays or TikTok's two-second equivalent. In most accounts it is the same arithmetic as hook rate under a different name.
UGC ad
A UGC ad is advertising produced in the visual language of a person's own post — handheld, unpolished, first-person — whether or not an actual customer made it. The format borrows the credibility of organic content, which is also what makes the disclosure rules matter.
UTM parameters
UTM parameters are tags appended to a destination URL so that analytics can record where a visit came from. They are the advertiser's own attribution, independent of any platform, and their usefulness depends entirely on naming discipline.
Video ad
A video ad carries a message over time rather than all at once, which makes it the only format that can demonstrate, sequence or build. It also has to earn its opening before any of that is delivered.
Video view
A video view is counted when playback passes a threshold the platform defines — two seconds, three, six, fifteen or completion. Because the thresholds differ, view counts from different platforms measure different things and cannot be compared directly.
View-through conversion
A view-through conversion is credited when somebody was served an ad, did not interact with it, and converted within the view window. The evidence that the ad caused anything is much weaker than for a click, which is why the counting rules matter so much.
Viewability
Viewability measures the share of impressions that met a standard for being on screen: the standard asks for at least half the ad's pixels for at least one second on display, two seconds on video. It asks whether an ad could have been seen, never whether anybody looked.
Questions
Put frontier AI to work on every ad account you run.
Tempio connects your ad accounts into one workspace and hands your own AI the numbers and the tools. Every change it writes is a draft until you approve it, or until a rule you wrote does.
Request early access- Read and draft by default. Approve is a separate permission you grant.