The vocabulary we work with when advising retailers and brands — written plainly, with the marketing rounding taken out. 66 terms across five areas.
Every entry carries its Polish equivalent, so the glossary works in both directions for teams operating across the CEE region. Definitions follow the terminology used in the IAB and IAB Europe measurement standards where those settle a question.
Advertising sold by a retailer against its own ad surfaces and purchase data — in its own channels, in its stores and beyond them. What separates it from conventional advertising is that the seller owns the transaction, so for an identified shopper it can tie ad exposure to an actual purchase rather than to reach alone.
A retailer's advertising business run as an organised operation: inventory, ad serving technology, a sales team and measurement. The retailer acts as a publisher and manufacturers become its advertisers.
Advertising built on transaction and purchase-intent data and settled on a sales outcome, whoever holds the data. It covers retail media, marketplaces and delivery platforms, plus the branches now separating out: media networks run by banks and insurers, and by carriers and hotel groups.
Inventory in the retailer's own digital channels: the online store, the app, product search results, email. The equivalent inside a physical store is in-store retail media; outside the retailer's channels it is offsite.
Campaigns run outside the retailer's channels but on its data — across the open web, social platforms or CTV. Measurement returns to the in-store transaction, and that is what separates offsite retail media from buying reach.
Advertising inside the physical store: screens, audio, shelf-edge media, electronic shelf labels. What distinguishes it from shopper marketing — sampling, tastings, displays — is that it is planned, served and settled like media.
The digital slice of in-store retail media: screens inside the store and around it — window, aisles, checkout zone, car park. Increasingly sold like media, priced against estimated contact with the screen, though many networks still sell fixed time packages.
The ability to connect ad exposure with a later purchase — at the level of a person, a loyalty account, a household or comparable test groups. The foundation of retail media: without a closed loop what remains is ordinary display inventory.
Data a retailer collects in its direct relationship with the shopper: transactions, loyalty programme, behaviour in the app and on the site.
A controlled environment where a retailer and a brand match their data — usually pseudonymised and joined on hashed identifiers — without either side reaching the other's record-level data. The environment's rules limit what may leave it: most often aggregates, in some deployments audience segments for activation. Pseudonymisation does not take data outside GDPR, so a lawful basis and an agreement defining each party's role remain necessary.
An endemic advertiser sells its products through the retailer; a non-endemic one does not and buys access to its shoppers — a bank, an insurer, a telco, a car brand. The split defines a network's growth path once supplier budgets are exhausted, which decides the ceiling of the whole business in markets where endemic budgets are shallow.
Paid promotion of a single product in search results, on category listings and on product detail pages, including competitors'. Usually settled in a CPC auction and targeted by keyword or by specific product.
A brand format in the retailer's search results — logo, headline and a set of products, also as video — targeted by keyword. The name comes from Amazon Ads and Walmart Connect; other vendors call their equivalents brand banners or brand pages.
Graphical and video modules in the retailer's online store and app. Bought in a CPM auction targeted on the retailer's first-party data, or as guaranteed inventory: fixed packages and category takeovers.
Advertising in the retailer's product search, served alongside organic results. A high-intent format, because the shopper states the need.
Everything a retailer has to sell: slots in the online store and app, product detail pages, search results, in-store screens, email sends, and offsite inventory sold against its data. Available volume follows traffic and is counted in impressions, not in the number of slots. Note that in a retail organisation the word inventory first means stock.
A specific serving position with defined specifications and visibility. In auction formats it is also an optimisation dimension — bids can vary by placement; in guaranteed sales it carries its own price.
The product page carrying sponsored carousels and cross-sell modules. The largest surface in an online store and the source of a standing dispute over who controls exposure on someone else's product page.
The share of ad requests that returned a paid creative. A supply-side delivery metric: it shows whether there is demand in the auction for a given surface at the moment of serving.
The share of impressions sold against impressions available for sale in a period; a measure of how well the asset is commercialised. Watch the collision: in merchandising the same term means units sold against units received, so the context has to be stated.
A brand's share of the retailer's search results for a defined basket of category keywords, within a stated range of positions, counting sponsored and organic results together. Not to be confused with the demand-side reading — a brand's share of query volume — used in brand tracking.
A brand's share of exposure: physical on the shelf or digital on a category listing.
The selling unit for in-store screens. The loop is a closed content cycle played on a screen; the advertiser buys a share of it — a number of spots of a given length per cycle — rather than impressions. This is why fill rate means something different for screens than for onsite inventory: the loop always fills with the retailer's own content.
The mechanism where advertisers' bids compete for position, with ranking decided by the bid together with quality and relevance. The auction generates most of a network's revenue and sets what a brand actually pays — a rate card covers guaranteed sales only.
Cost per click. The standard in auction formats, because it ties cost to interest rather than to serving alone.
Cost per thousand impressions. Typical for display and DOOH. In retail DOOH impressions are estimated from footfall rather than counted as in display.
Cost per action, most often a purchase. It moves risk onto the seller of the inventory, so in retail media it tends to be offered conditionally.
A manufacturer's activity directed at the trade channel and at the shopper in store: displays, promotions, POS materials, in-chain activations. It carries its own budget, separate from the brand's media budget and from the commercial terms negotiated with the retailer.
Activity aimed at the shopper on the path to purchase and at the shelf. In practice it is a third, distinct pool alongside the media budget and commercial terms — and onsite retail media most often competes with this pool rather than with the media budget. Where the money sits decides who on the brand side takes the decision: commercial, shopper, or media with the agency.
A shared growth plan between retailer and manufacturer: agreed objectives, a category plan and mutual commitments from both sides, reviewed on a cycle. Distinct from the negotiation of commercial terms, even though it runs on the same calendar. Increasingly it carries retail media investment commitments.
A model where a partner takes a share of the advertising revenue generated instead of a fixed fee.
The share of advertising spend retained by an intermediary or technology vendor, expressed as a percentage — the percentage side of a revenue share agreement. Comparable only when the base is stated: the advertiser's gross spend or the network's net revenue.
The party that invoices the advertiser and recognises the revenue. It determines whether a network reports advertising revenue at the full value of spend or only its own share after settling with a vendor — and therefore whether two networks' revenue figures are comparable at all.
The retailer or its partner runs the campaign on the advertiser's behalf. It dominates where a network has no self-service platform yet, and — regardless of maturity — for complex, offsite and DOOH campaigns. It usually carries a service fee and a minimum budget threshold.
The advertiser buys and runs the campaign itself in a platform. It lowers the cost to serve but requires mature technology and predictable availability.
Managing availability, price and ad density to extract the highest revenue from a constrained asset without degrading conversion and the shopping experience. It also covers splitting inventory between guaranteed and auction demand. A network that maximises ad revenue alone degrades the underlying business.
The price below which a network will not sell a given surface, even when demand is weak. In auction formats it acts as the minimum accepted bid and is often set per surface and per deal.
A one-to-one programmatic transaction with reserved inventory, guaranteed volume and a price agreed upfront, with no auction. Both sides commit: the network reserves the inventory, the buyer takes the agreed volume.
A programmatic transaction open only to invited buyers and executed through a deal ID — most often a closed auction with a floor, sometimes a fixed price without an auction. Unlike programmatic guaranteed it does not guarantee volume: it grants priority of access, not a reservation.
The binding order document for a campaign: formats, dates, volume, prices and delivery terms together with the accepted general conditions. The link between proposal and launch, and the basis for billing and for any make-good claim.
Compensation in additional delivery or an invoice adjustment when a campaign fell short of the agreed volume or ran outside the terms of the order. It applies to guaranteed buys; in auction campaigns there is no volume guarantee and therefore nothing to make good.
Revenue attributed to a campaign divided by media spend. On its own it says nothing about whether the sales would have happened anyway. Nor is it a profitability measure: the numerator is revenue, not margin, so the break-even point depends on product margin and on whether network and agency fees sit in the denominator.
Incremental sales divided by campaign cost. It counts only sales that would not have happened without the campaign, which makes it an estimate carrying uncertainty and reported with a confidence interval. It can be lower than attributed ROAS where attribution collects demand that existed anyway — and higher where attribution cannot see part of the effect: in-store sales, halo onto other products, conversions outside the window.
The portion of sales that would not have happened without the campaign. A counterfactual quantity: it cannot be read off a report, only estimated through an experiment. It separates real effect from claiming credit for purchases that would have occurred regardless.
A randomly withheld portion of the population eligible for targeting, to which the campaign is deliberately not served. The difference in sales between exposed and control groups is the campaign's effect only where assignment was random and the control group consists of people who would have been targeted. The control can be drawn at user, store or geographic level.
An experiment run on randomly assigned regions or stores instead of individual users. The primary — often the only — way to measure incrementality where the user cannot be identified: in store, on screens, in CTV and across much of offsite.
Techniques for controlling selection bias in a test: serving the control group a neutral advertisement instead of nothing, or logging won auctions without serving. Without them the control group differs from the exposed one not only by the absence of an ad but by the fact that the algorithm shows ads to those closest to purchase.
The range within which the true value of an estimate falls at a stated probability. Incrementality, iROAS and sales lift are estimates, so without an interval they cannot be read: a 3% lift ±1pp and a 3% lift ±9pp are entirely different results.
The rule for assigning a conversion to one or more prior ad exposures. It is a reporting convention, not proof of causation: it shows what happened after contact with an advertisement, not whether it happened because of it. The second question is answered by incrementality measurement.
The period after contact with an advertisement during which a purchase still counts as its effect; usually set separately for clicks and impressions. Different windows produce different results for the same campaign, which makes ROAS without a stated window incomparable across networks. The window alone is not enough: comparison requires a matching attribution model, product scope and treatment of in-store sales.
A purchase attributed to a campaign after an impression was served, with no click on any of its advertisements. Counted in its own window, usually shorter than the post-click window. An impression does not prove the advertisement was seen — viewability is a separate measurement. Not to be confused with VTR, which in video means the share of completed views.
The increase in sales in the exposed group against the control group. A comparison with the period before the campaign is not sales lift — it controls for neither seasonality, price promotion nor distribution changes — and cannot be used to calculate incrementality.
Purchases attributed to a campaign made by shoppers who had not bought the brand at that retailer within an agreed look-back period. Reported as order count, sales value and their share of the total. A retailer-local measure: a shopper loyal to the brand at another store counts here as new.
A campaign's effect on sales of products other than the advertised one — most often the rest of the same brand's portfolio, including in a channel other than the one that carried the advertisement. A competitor's uplift within the category is not halo for the advertiser.
Sales displaced rather than generated: from another product of the same brand, from another retailer, or from a future period. The reason incrementality measured at one retailer does not add up to incrementality at brand level.
Statistical decomposition of sales into a base and causal drivers — media, but also price, promotion, distribution and seasonality — on aggregated data. It needs no personal data but does need long time series and real variation in spend: a channel on a flat budget is invisible to the model. A model that explains sales by channels alone assigns the effect of price and promotion to media. Results are calibrated with experiments, not substituted for them.
The technical layer of delivery: trafficking, creative review and approval, targeting and cap configuration, delivery monitoring, resolving serving errors and discrepancies between systems. Accountable for the campaign working technically as written.
The process and commercial layer: running a campaign from brief and proposal request, through availability checks, proposal, order and launch, to reporting and billing. Accountable for the campaign being delivered as contracted. In small networks one person holds both roles.
Entering a campaign into the ad serving system together with creatives and targeting rules.
The rate at which budget is spent and delivery is distributed over time. In guaranteed campaigns the network manages it to deliver the contracted volume evenly. In auction formats it is an advertiser setting: a daily budget spent evenly or accelerated until exhausted.
The gap between impressions or clicks reported by the network and the count from the advertiser's or agency's system, settled against an agreed tolerance threshold. The most common escalation to AdOps and the most common dispute at billing.
The network's requirements for a format: technical parameters, the product data pulled from the retailer's catalogue, and the retailer's editorial and branding rules. Some formats are not supplied as a finished file but assembled dynamically from the product feed. The absence of shared specifications across networks is one of the barriers to scale — one answer is IAB Europe's Flexi Ad Sizes Guidelines of January 2026, recommending flexible aspect ratios instead of fixed pixel sizes.
A request from an advertiser or its agency, answered by the network with a proposal built on an availability forecast and a rate card. It covers guaranteed and manually sold inventory; auction formats are bought self-service, outside the proposal process.
A closed ecosystem where the advertiser buys, measures and optimises only inside the owner's tools, and data leaves at most as reports or through a clean room. The consequence is no deduplication of reach and frequency across networks. It does not mean the absence of audit: measurement inside a walled garden can be accredited or certified.
Not one document but a family of standards. In the US, IAB together with MRC published the Retail Media Measurement Guidelines (January 2024). In Europe, IAB Europe published Retail Media Measurement Standards (April 2024), updated as Commerce (Incl. Retail) Media Measurement Standards V2 (January 2026). IAB and IAB Europe jointly issued In-Store Retail Media: Definitions and Measurement Standards (December 2024) and Guidelines for Incremental Measurement in Commerce Media (November 2025).
The mechanisms for demonstrating compliance with the standards. MRC accredits specific metrics of specific networks following an audit. IAB Europe runs the Retail Media Certification Programme as an audited route to compliance with its standard. A claim of being aligned with IAB standards, without accreditation or certification, remains the network's own assertion.