Core Ad Metrics glossary

Slots that could have carried an ad. Useful for spotting how much inventory you are leaving on the table.

Times the system asked for an ad to fill a slot. Larger than impressions, because plenty of requests go unfilled.

Items placed in a basket. A strong intent signal and a weak revenue predictor, since most carts are abandoned.

Every tap counted, including expands, profile clicks and reactions. Flattering, and almost never the number you want.

Conversions a channel touched without closing. Real, and routinely used to justify spend that was never incremental.

How long people spent with the ad, not just whether it rendered. Newer, harder to measure, and closer to what you were trying to buy.

Items or value per order, depending on who is asking. Agree which one before anyone reports it.

Impressions divided by reach. Climbs quietly when the audience is small, and drags performance down with it.

The B2B version, where one account can hold many users. Grows through expansion rather than through new logos.

Revenue divided by paying users only. Strips out the free tier and shows what monetisation looks like at the top.

Revenue divided by all users, paying or not. Falls when you grow the free base, which is not always bad news.

Revenue divided by units sold. Drifts down quietly whenever discounting becomes a habit.

What a customer spends over a defined window. Lifetime value without committing to a lifetime.

Revenue divided by transactions. The same arithmetic as AOV, usually in a retail or payments context.

The share of started checkouts that finished. Where payment friction, shipping cost and trust problems show up first.

How many clicks it takes to produce one lead. The bridge between what you paid for traffic and what you paid for a name.

Taps that sent someone toward your site. The first thing you pay for that has any intent attached.

Comments divided by reach. High comment rate with low save rate usually means the post was argued with, not liked.

How many people finished whatever you started them on, whether that is a video, a form or a checkout.

The revenue attached to a conversion. Send the wrong value and every value-based bid strategy inherits the error.

Google speak for ROAS, written as a ratio rather than a percentage.

Whatever you told the platform to count. Define it loosely and the algorithm will happily optimise toward noise.

Spend divided by carts. Useful mid-funnel target when purchase volume is too thin to bid on.

Spend divided by started checkouts. Watch the ratio between this and cost per purchase, because the gap is your checkout leak.

Everything spent divided by customers won. Also called CAC, and the number every other cost metric is a proxy for.

Occasionally used in small-budget or local buying, where a thousand impressions is a large number.

What it costs to get one person to see the page. The number that exposes slow pages faster than any speed tool.

What one marketing-qualified lead costs. Only comparable across companies if the qualification bar is the same, which it never is.

Spend divided by opportunities created in the pipeline. Where marketing and sales finally measure the same thing.

Spend divided by orders. The ecommerce version of CPA, and the one that has to sit under gross profit per order.

Spend divided by leads sales agreed to work. Always higher than CPL, often by three or four times.

Spend divided by revenue produced. The inverse of ROAS, and easier to reason about when the ratio is below one.

Spend divided by sessions. Slightly different from cost per visit depending on how the analytics tool defines a session.

What one sales-qualified lead costs. The last number before acquisition cost, and the one that predicts it best.

CPM in full. The price of reaching a thousand impressions.

Spend divided by visits. Sits between CPC and CPA, and catches the traffic that clicked but never loaded.

Spend divided by distinct people who arrived. Higher than cost per session, and closer to what a human costs you.

Ad spend divided by conversions. Fine as long as everyone agrees what counts as a conversion, which is rarer than it sounds.

What one click costs on average. You do not set this number, you set a bid and the auction decides what you pay.

Cost per completed view. Stricter than CPV because the person watched to the end, which makes it the more honest video price.

Cost per engagement. Whatever the platform decided engagement means that quarter, divided into your spend.

Cost per install. The app world version of CPA, and just as vulnerable to counting installs that never open the app twice.

Spend divided by leads. The cheapest number in the funnel to improve and the least meaningful on its own.

What a thousand impressions cost you. The base price of paid media, since almost every other cost metric is derived from it one way or another.

CPM for banner inventory. The cheapest impressions on the open web, and the ones most likely to be seen by nobody.

CPM on social feeds. Sits between display and video, and swings hard with audience size and seasonality.

CPM for video inventory specifically. Runs higher than display because attention is scarcer and the format is harder to ignore.

Cost per point, from broadcast buying. What it costs to reach one percent of a target audience once.

Cost per sale. Common in affiliate deals, where the publisher only gets paid when money changes hands.

Plain English for CPM. Same number, written out for people who do not live in the acronym.

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

Clicks divided by impressions. A relevance signal on search and a creative signal on social, and it means different things in each.

Conversion rate. The share of clicks or sessions that turned into the thing you were after.

New followers divided by reach. Worth tracking on content that is meant to build an audience rather than sell something today.

Ads served. What you bought, as opposed to what anyone saw.

New people a channel added that other channels had not already reached. The reason a second platform is worth buying.

Checkouts started. The last step before payment, and the one where cost surprises kill the sale.

Any deliberate action divided by impressions. Broad by design, which makes it easy to inflate.

Conversions divided by landing page views rather than by clicks. Strips out the people who never made it past the load.

Loads of the page the ad pointed at. The gap between clicks and landing page views is your bounce before the page even rendered.

The share of visitors who filled in the form. Rises every time you remove a field and falls every time you add one.

The share of leads that eventually bought. Multiply it into CPL and you get the number sales cares about.

People who identified themselves without buying. A lead is not a customer, and the gap between the two is where budgets die.

Likes divided by reach. The cheapest signal to earn and the least predictive of anything downstream.

Clicks on the actual link, ignoring likes, expands and other taps. The version that matters when traffic is the goal.

The action the business actually wanted. Usually a purchase, a signup or a booked call.

Small steps on the way to the real thing. Useful as diagnostics, dangerous as bidding targets.

How many marketing-qualified leads survive contact with sales. When this drops, the problem is usually the definition, not the leads.

How often a customer buys in a period. Multiply by average order value and you have most of lifetime value.

Clicks that left the platform for your site. The only CTR that costs the platform something, which is why it is buried.

Pages loaded. Fine for content sites, close to useless for judging a campaign.

Times a product page was opened. The first real signal of intent in an ecommerce funnel.

Gross profit divided by clicks. The honest version of value per click, and the only one that tells you your maximum bid.

What is left after cost of goods. The version of value per conversion that survives a finance meeting.

Gross profit per thousand impressions. Rarely reported, and the cleanest way to compare two very different media buys.

Gross profit divided by sessions. Moves when margin moves, which revenue per session will not show you.

Gross profit divided by distinct visitors. The ceiling on what a visitor can cost before the account loses money.

Purchases divided by sessions. The headline ecommerce number, and the one most affected by traffic quality rather than site quality.

Orders divided by customers over a window. Small changes here move lifetime value more than price changes do.

Completed orders. Everything upstream is a prediction, this is the fact.

Leads that met the criteria worth a salesperson time. The only lead count worth putting in a bidding strategy.

Money in, before any cost comes out. The number that makes ROAS look good and profit look confusing.

The same number, phrased for people who think in revenue rather than in conversion value.

The same figure, in the language of publishers rather than buyers.

Revenue divided by sessions. Tells you what one visit is worth, and therefore what you can pay for one.

Revenue produced per thousand impressions. Put it next to CPM and you have the whole media buy in two numbers.

Revenue divided by distinct visitors. Higher than revenue per session, and the better number for setting bids.

Saves divided by reach. On most social platforms this predicts long-term reach better than likes do.

Visits to the site, grouped by a time window. One person can produce several in a day, which is why sessions and users never match.

Shares divided by reach. The most expensive engagement to earn and the only one that buys you distribution for free.

The close rate on leads sales agreed to work. Slow to move, and the single biggest multiplier on acquisition cost.

The Meta version, counting fifteen seconds or the full clip if shorter. Not comparable to a YouTube view without adjustment.

Clicks from distinct people. Repeated clicks from the same person are usually curiosity, not demand.

Clicks from distinct people rather than total clicks. Lower than raw CTR, and closer to how many humans responded.

Impressions counted once per person. The gap between this and total impressions is your frequency.

Distinct people who saw the ad at least once. Always smaller than impressions, and the number that caps how far a campaign can go.

Items per order. One of the two levers behind average order value, and the easier one to move with bundles.

Distinct people, as far as the analytics tool can tell. Cookie loss and cross-device browsing make this an estimate, not a count.

Average revenue per conversion. If this drifts down while volume climbs, you are buying cheaper customers, not more of them.

Video completion rate. The share of starts that reached the end.

Completions divided by starts on video. Drops fast after fifteen seconds on social and after five on skippable pre-roll.

The same idea, named differently depending on which platform report you are reading.

Views divided by impressions on video. How many people stayed long enough to count as watching.

The share of served impressions that had a real chance of being seen. Anything below fifty percent means half your CPM bought nothing.

The same idea as users, usually in older reports. Treat them as interchangeable unless the tool says otherwise.

View-through rate. Views over impressions, expressed as a percentage.