The concepts used in planning and evaluating advertising: TV and radio metrics, digital metrics, buying models, programmatic, data collection and business outcomes. Each term comes with an explanation, a formula and an illustrative calculation.
Average audience of a programme, commercial break or airing as a percentage of the chosen base: the total population or a target group. Without a base, the figure is meaningless.
How many different people were exposed to the campaign at least once, in absolute numbers or as a percentage of the base.
How many ad contacts, on average, each person reached received.
Sum of the ratings of all airings, calculated against the same base. It counts contacts, not people, so it can easily exceed 100.
The equivalent of GRP, with ratings taken only for the campaign’s target audience.
Number of contacts in absolute terms, usually in thousands: GRP expressed in people. ⚠ Some sources use the term for average frequency.
The proportion of viewers with the TV on in a given time slot who chose a particular channel.
Percentage of households (HUT) or people (PUT) with the TV on in a given time slot.
How much more concentrated the TA is in the audience of a channel or programme than in the base.
The number of contacts from which, by the planner’s estimate, the message starts to deliver results. The 3+ benchmark is a convention and depends on the objective.
The share of the audience that received at least as many contacts as the effective frequency requires.
Share of the audience with N or more contacts. Reach 1+ is the same as reach.
The price of each GRP or TRP point delivered.
The price of a thousand ad contacts; in digital it is called CPM.
Creative wear-out: with too frequent repetition, attention and response decline.
The group of people the campaign is aimed at; TRP and reach within the TA are calculated against it.
Target values fixed in the media plan before launch: ratings, reach, frequency, airings, budget.
Values the campaign actually achieved, according to the measurement provider or the platforms.
Actual as a percentage of plan, separately for each metric.
The gap between actual and plan, in the metric’s own units or in percent.
Total number of times the ad was displayed on screen. One person can generate several impressions, so impressions are not the same as people.
Number of different users who encountered the ad at least once.
How many impressions, on average, each user reached received.
How many times users clicked on the ad.
Counted plays of a video ad; the minimum duration is set by the platform.
Visits to the website; each visit is counted separately, even from the same person.
Number of different website visitors over the selected period.
Share of impressions that could actually be seen. Under the MRC standard, this means at least half of the banner’s area on screen for 1 second or more, or 2 seconds or more for video.
Payment for every thousand impressions, regardless of clicks.
The advertiser is charged only for clicks.
Payment per counted video view.
Payment per target action; the advertiser defines which action counts.
Payment for a prospect’s contact details: a form, a call, a message.
Payment for each order placed.
Payment for each mobile app install.
The calculated cost per thousand impressions for a buy under any pricing model; it helps compare CPC or CPA with CPM.
Share of impressions that resulted in a click: how well the ad attracts attention.
Share of clicks or visits followed by a target action.
Share of video impressions watched to the end. VTR is sometimes used for the share of counted views.
Share of visits classified as bounces; each analytics system uses its own criterion.
The portion of revenue spent on advertising; the inverse of ROAS, expressed as a percentage.
A brand’s share of advertising presence in a category or channel; measured by impressions, ratings or budgets.
Total reactions to a post divided by the number of followers.
Reactions divided by the post’s reach; more accurate than ER for evaluating advertising.
Reach is the number of different viewers of the content; impressions are the total number of times it was displayed.
The play counter for Reels and other videos.
Actions on a post: likes, comments, saves and shares.
The number of accounts following the creator.
Return on investment: the percentage the investment earned over and above the amount invested.
Return on marketing spend. The return is the campaign’s gross profit, not revenue; otherwise the result is overstated.
Revenue per somoni of ad budget; usually a ratio, sometimes a percentage.
Average cost of acquiring a new customer, including advertising and sales.
Gross profit from a customer over the entire relationship; as a rule, a forecast value.
The buyer’s system for automated inventory buying: audiences, bids, sites, restrictions.
The publisher’s system that puts its ad slots up for auction to many DSPs.
Real-time bidding for an individual impression, within fractions of a second while the ad slot loads.
A repository of audience data: it combines sources, builds segments and passes them to ad platforms.
A closed auction for invited buyers only; access is via a Deal ID, often with a price floor.
Automated buying through platforms: in auctions (RTB, PMP) and through direct fixed-price deals (Preferred Deal, Programmatic Guaranteed).
A limit on the number of impressions per user over a set period.
Measures that keep ads from appearing next to content unacceptable to the brand.
Users who resemble a brand’s seed audience, such as its customers, in traits and behaviour.
CTV — video ads on internet-connected TVs: Smart TVs, set-top boxes, consoles. OLV — video ads in the video players of websites and apps.
Parameters appended to a link that tell analytics the source, medium and campaign; utm_term and utm_content are also used.
A snippet of code on site pages that records visitor actions, such as a payment, and sends them to the ad platform.
The principle by which a conversion is credited to a channel, campaign or ad.
The scheme for dividing a conversion’s value among touchpoints: Last Click, First Click, Linear, Data-Driven, etc.
Sending events from your own server, not only from the browser; less data is lost to ad blockers and cookie restrictions.
Meta’s server-side interface for sending events, including offline conversions. Other platforms have equivalents; for TikTok it is the Events API.
A Google mechanism that passes the visitor’s cookie choice to GA4 and Google Ads tags and adjusts data collection accordingly.
Eliminating double counting of a conversion reported by two sources, such as the pixel and CAPI.
A diagnostic Google Ads score from 1 to 10 per keyword, based on expected CTR, ad relevance and landing page experience. The score itself is not used in the auction, but it reflects the factors that affect ad position and cost per click.
Words that stop the ad from showing; they filter out irrelevant queries.
How many impressions the ad received out of all those it was eligible for.
Nothing found — try another query.
Take a roadside billboard. Reach is how many different people noticed it. Frequency is how many times, on average, each of them saw it. GRP is reach in percent multiplied by frequency, i.e. the total volume of contacts.
↑ — the metric should go up: CTR, CR, ROAS, ER, reach. ↓ — it should go down: DRR, CAC, bounces, cost per contact or click. Amounts in the examples are illustrative.
Most often it is the total number of contacts in absolute units, i.e. GRP in thousands; less often, average frequency. Before combining data, find out what the source means.
GRP is usually calculated on the whole population and TRP on the TA, but ratings within the TA are sometimes also called GRP. The convention is set by the measurement provider’s methodology and the agreement between the parties.
They cannot be compared directly. In Yandex Metrica, a bounce by default is a single-page visit shorter than 15 seconds. In GA4, bounce rate is the share of non-engaged sessions, i.e. sessions with no key event, shorter than 10 seconds and with a single page view.
Reach is given in people or as a percentage of the base (universe). Always fix the base, otherwise the figures are not comparable.
Some divide reactions by followers, others by reach. For evaluating an integration, ERR is more reliable: it does not depend on the number of followers, including bought ones.
Two views of the same ratio: DRR = 1 / ROAS × 100%, so ROAS 4 equals DRR 25%. The report format is agreed with the client.
It reflects working practice, not an industry standard.
© SKY Central Asia, 2026 · A reference for practical work. How metrics are calculated depends on the measurement system and is agreed with the client.
We use cookies for analytics and to improve the site. Learn more