Advertising and unit economics glossary

The score Google uses to decide whether your ad shows and in what position. Bid is only one input; expected click-through rate, ad relevance and landing page experience carry as much weight.

How closely the ad copy matches the intent behind the keyword. A component of Quality Score, and usually the cheapest one to fix.

A conversion where a channel appeared somewhere in the path but was not the last touch. Useful for spotting undervalued channels, easy to double-count.

The rules that decide which touchpoint gets credit for a conversion. Every model is a choice, not a measurement, and every platform picks the one that flatters itself.

How long after a click or impression a conversion still counts. Shorten it and every channel looks worse; lengthen it and everything looks better.

The real-time process that picks which ads show for a query. You are not bidding against a fixed price, you are bidding against whoever else wants that person right now.

Letting the platform set bids toward a goal you define. Works when conversion data is clean and plentiful, fails quietly when it is neither.

Revenue divided by orders. The one growth input you do not have to buy.

The top of the funnel, where the goal is being known rather than being bought from. Measured in reach and frequency, not in ROAS.

What you are willing to pay for a click, impression or conversion. In modern accounts you rarely set it directly, you set the target that produces it.

The rule set that turns your goal into bids. Maximise clicks, target CPA, target ROAS and maximise conversion value all behave differently under the same budget.

Total sales and marketing cost divided by all new customers, paid and organic together. The number the business feels, as opposed to the one the ad platform reports.

The share of sessions with no second interaction. A weak signal on its own and a strong one when it moves after a landing page change.

Queries containing your own name. Cheap, high-converting, and the single most over-credited line in most accounts.

The return on ad spend at which a campaign stops losing money. One divided by gross margin, before anything else is considered.

The loosest keyword match type. Modern broad match plus smart bidding can work well; broad match with manual bids is a way to donate money.

Spending the budget evenly across the period rather than exhausting it in the first ten days. Front-loading distorts every weekly comparison that follows.

Customer acquisition cost. Everything spent to win a customer divided by the customers won. Larger than CPA whenever not every conversion is a customer.

How many months of gross profit it takes to earn back what a customer cost to acquire. The number that decides whether growth is fundable.

The container that holds budget, targeting and bidding rules. Structure decides what you can control and what you can only observe.

The share of customers who leave in a period. Its inverse is average customer lifetime, and it compounds far faster than most people assume.

Clicks divided by impressions. A relevance signal on search and a creative signal on social.

A group defined by when they joined. Comparing cohorts rather than months is the only way to see whether retention is genuinely improving.

Price minus every cost that moves with the unit. The only per-unit number that answers whether one more sale makes you better off.

The action you decided counts. Define it too loosely and the bidding algorithm will happily optimise toward noise.

The share of clicks or sessions that convert. The strongest lever in the chain, because it multiplies everything downstream.

The period after a click during which a conversion is attributed to it. Different from the attribution window in that it applies per conversion action.

Ad spend divided by conversions. Not the same as customer acquisition cost unless every conversion is a customer.

What a single click costs on average. A function of bid, competition and quality, not a price you set.

Spend divided by leads. The first number in a chain that ends at cost per customer, and meaningless without the rest of it.

The price of a thousand impressions. The base currency of paid media, since every other cost metric is derived from it.

What a counted video view costs. Compare across platforms only after checking what each one counts as a view.

The decay in performance as the same audience sees the same creative repeatedly. Shows up first as falling click-through rate and rising frequency.

A path that starts on one device and finishes on another. The reason platform-reported and analytics-reported conversions rarely agree.

The gross profit a customer produces over the whole relationship. Sets the ceiling on what you can afford to pay to acquire them.

Restricting or adjusting spend by hour and day. Worth doing when conversion rate genuinely varies by time, not because the graph looks bumpy.

Removing double-counted conversions when several channels claim the same sale. Without it, channel reports add up to more revenue than the business made.

Advertising to people already looking for what you sell. Cheap and finite, because you cannot capture more demand than exists.

Creating want where none existed yet. Expensive, slow, and the only way past the ceiling that demand capture eventually hits.

Banner inventory across websites and apps. Cheap impressions, weak intent, and the first place a careless budget disappears.

Ads generated from your site content rather than from keywords you chose. Useful for discovery, dangerous without a tight exclusion list.

Effective cost per thousand impressions, derived after the fact from whatever you paid. The common unit for comparing buys priced differently.

Interactions divided by reach, impressions or followers. Three definitions that produce three different numbers from the same post.

The tightest keyword match type. Still matches close variants, so it is exact in name rather than in behaviour.

Placements, audiences or search terms you have told the platform to avoid. The most neglected asset in most accounts.

The platform prediction of how likely your ad is to be clicked. A Quality Score input you influence through relevance rather than through bidding.

Data you collected directly from your own customers. Increasingly the only durable targeting and measurement asset you own.

Average impressions per person in the period. Rises quietly when audiences are small, and takes performance down with it.

A limit on how often one person sees the ad. The cheapest defence against creative fatigue.

The narrowing path from impression to customer. Every metric on this site sits at one specific point in it.

Restricting delivery by location. Also the cleanest way to run a holdout test, since you can switch a region off and watch what happens.

Revenue minus cost of goods, as a share of revenue. Decides what break-even ROAS has to be.

Revenue minus cost of goods, in currency rather than percent. The money that has to cover acquisition.

The lift a campaign produces in channels other than itself. Real, frequently claimed, and rarely measured properly.

Deliberately withholding advertising from a group or region to see what happens without it. The only honest measure of incrementality.

One serve of an ad. What you buy, as opposed to what you get.

The share of available impressions you won. The gap tells you whether the constraint is budget or rank.

People the platform believes are actively shopping in a category. Useful, and only as accurate as the signals behind it.

The conversions that would not have happened without the ad. The number that matters, and the one no platform reports.

Incremental return on ad spend. ROAS measured against a holdout rather than against platform attribution, and usually far lower.

The term you bid on. Increasingly a hint rather than an instruction, given how loosely modern matching works.

The platform assessment of what happens after the click. A Quality Score input, and the one with the largest effect on conversion rate.

Giving all credit to the final touch. Simple, auditable, and systematically biased toward the bottom of the funnel.

A person who identified themselves without buying yet. A lead is not a customer, and CPL is not CAC.

An audience modelled on the traits of an existing one. Quality depends entirely on the quality of the seed list.

The share of impressions you missed, split into budget and rank. Tells you which constraint to attack.

The return on the next dollar of spend, not the average across all dollars. Always lower than blended ROAS, and the number that should drive budget decisions.

Total revenue divided by total marketing spend, across all channels. Immune to attribution arguments, which is exactly why it is useful.

How much you added on top of cost, as a share of cost. Not the same as margin, and the confusion between them is expensive.

Statistical modelling of how each channel contributed, using aggregate data rather than tracking individuals. Slow, expensive, and privacy-resistant.

Splitting credit across several touchpoints. More honest than last-click in principle, and heavily dependent on assumptions in practice.

A term you have told the platform never to match. The single highest-return maintenance task in a search account.

Revenue from existing customers this period against last, including expansion and churn. Above one hundred percent means the base grows without new customers.

Sending sales that happened outside the website back into the ad platform. The fix for optimising toward leads that never close.

A platform recommendation score. A checklist of what the platform would like you to do, which is not the same as what you should do.

The increase in unpaid traffic caused by paid activity. Real in brand campaigns, and the reason pure paid ROAS understates some spend.

Months of gross profit needed to recover acquisition cost. Shorter payback means the same capital funds more growth.

A campaign type that runs across all Google inventory with minimal manual control. Effective with clean data and strong assets, opaque when either is missing.

The specific site, app or surface where an ad appeared. Reviewing placements is how you find out where the cheap impressions came from.

Profit on ad spend. ROAS calculated on gross profit rather than revenue, which is the version that reflects reality.

A conversion that followed a click. The stricter and more defensible of the two conversion types.

A conversion counted after an impression with no click. Legitimate in principle, wildly over-claimed in practice.

Automated buying of display and video inventory through exchanges. Efficient at scale, and layered with fees that erode the working media percentage.

A lead that meets the criteria worth a salesperson time. The gap between leads and qualified leads is where most cheap CPLs die.

A one to ten diagnostic of expected click-through rate, ad relevance and landing page experience. Directional rather than exact, and worth watching as a trend.

The number of distinct people who saw the ad, as opposed to the number of impressions served.

The share of orders returned or refunded. Applied after revenue, before margin, and routinely left out of ROAS targets.

Advertising to people who already interacted with you. Posts excellent attributed ROAS and delivers modest incremental ROAS.

Revenue divided by ad spend. Says nothing about profit until you put margin next to it.

Revenue divided by clicks. The most direct way to see what a click is worth, and therefore what you can pay for one.

Recency, frequency and monetary value. A simple segmentation that usually beats far more elaborate models.

The actual queries that triggered your ads, as opposed to the keywords you bid on. The gap between the two is where the money leaks.

Predictable variation across the year. Comparing this month to last without accounting for it produces confident nonsense.

A subset of an audience defined by shared traits or behaviour. Averages hide problems that segments reveal.

Your share of the available impressions or conversations in a category. A competitive measure rather than a performance one.

Machine-set bids optimised toward a conversion goal. Needs volume and accurate conversion data, and degrades badly without either.

The probability that a difference between two variants is not chance. Ninety-five percent is convention, not law.

A bid strategy aiming for an average acquisition cost. Set it below what the account has ever achieved and delivery simply stops.

A bid strategy aiming for a revenue-to-spend ratio. Only sensible once you know the break-even ratio it has to clear.

Activity aimed at people who do not know you yet. Judged on reach and cost per view, never on last-click ROAS.

The URL structure that carries tracking parameters through to the site. Break it and every downstream report goes blank.

What one sale is worth after everything variable is paid. The foundation under every media decision on this site.

A tag appended to a URL so analytics can identify the source. Inconsistent naming across a team destroys more reporting than any tool ever fixes.

Revenue or gross profit generated per click. Multiply by expected clicks to sanity-check any budget.

A cost that moves with volume. Goods, shipping, processing and returns, as opposed to rent and salaries.

Cost per thousand viewable impressions, rather than served ones. Always higher than CPM, and closer to what you bought.

A conversion attributed to an impression that was never clicked. Treat as a directional signal, never as revenue.

Whether an impression was capable of being seen. An impression that rendered below the fold is a cost without an audience.

A campaign aimed at customers who already left. Cheaper than new acquisition, and capped by how many lapsed customers exist.