A controlled comparison of two versions of an ad, offer, landing page, subject line, or other element to determine which performs better against a defined metric.

The maximum amount a marketer can afford to spend to generate a lead, order, or customer while still meeting the campaign’s profit requirements. Actual CPS and CPL can move as campaign inputs and performance change, but the allowable is the fixed economic ceiling used to build and judge that campaign.

The method used to assign credit for a response, conversion, or sale to the marketing touchpoints that influenced it.

The average revenue generated by each order. Formula: total order revenue ÷ number of orders.

Revenue and profit generated after the initial transaction through repeat purchases, renewals, upsells, cross-sells, or continuity programs.

The point at which campaign revenue or contribution margin equals campaign cost, producing neither a profit nor a loss.

The minimum ROAS needed to cover the costs included in the calculation. It depends on gross margin and any other variable costs assigned to the sale.

A person or organization that has completed a purchase; often distinguished from a prospect, lead, or inquiry.

The specific instruction that tells the audience what to do next, such as call, click, request a quote, subscribe, or buy now.

A coordinated set of messages, offers, audiences, media placements, and response mechanisms designed to achieve a measurable objective within a given period of time.

A testing structure in which the current best-performing treatment is the champion and a new alternative is the challenger. Promote a challenger only after it clears the predefined performance and reliability thresholds; repeated testing against a weak control produces little strategic value.

The percentage of customers or subscribers who cancel, lapse, or otherwise stop buying during a defined period.

The percentage of delivered impressions or messages that produce a click. Formula: clicks ÷ impressions or delivered messages × 100.

A recurring-revenue arrangement in which customers receive products, services, or access on a repeating schedule until they cancel or the term ends.

Revenue remaining after the variable costs required to generate and fulfill a sale. It contributes to fixed costs and profit.

The unchanged or established version against which a test variation is compared. A valid control provides the baseline for measuring improvement relative to the existing treatment. It supports a conclusion about true incrementality only when it receives no treatment. Do not replace or alter the control during a test, because doing so compromises the comparison.

The completion of the action a campaign is designed to produce, such as a form submission, phone call, appointment, subscription, or purchase.

The percentage of people who complete a defined action after entering a measured audience or process. Formula: divide conversions by eligible opportunities, then multiply by 100. In a sales funnel, the term may refer specifically to the percentage of leads that become customers.

The written language used in an ad, email, landing page, direct-mail piece, script, or offer to persuade the audience to respond.

The average marketing cost required to acquire one new customer. Formula: acquisition spend ÷ new customers acquired.

The average amount paid for each click generated by an advertising campaign. Formula: ad spend ÷ clicks.

The average marketing cost required to generate one lead. Formula: lead-generation spend ÷ leads generated.

The average marketing cost required to generate one order. Formula: campaign spend ÷ orders attributed to the campaign.

The average marketing cost required to generate one completed sale. Formula: campaign spend ÷ sales attributed to the campaign.

The audience-facing assets used in a campaign, including copy, design, images, video, audio, format, and presentation.

A controlled comparison of copy, design, format, imagery, or execution. Creative tests should be built around a clear hypothesis and held to the same audience, offer, media, and measurement conditions wherever possible.

An offer for a related or complementary product or service made to an existing buyer.

The average marketing cost required to acquire one new customer. Formula: acquisition spend ÷ new customers acquired.

The estimated value a customer contributes over the full relationship, usually based on revenue or contribution margin, purchase frequency, retention, and service costs.

The ongoing process of correcting, standardizing, deduplicating, validating, and removing unusable records from a customer or prospect database.

A measurable marketing message delivered physically to a selected audience through the postal system or another delivery service.

Marketing that communicates directly with selected individuals, households, or organizations using addressable channels such as mail, email, telephone, messaging, or targeted digital media. Direct marketing is broader than direct mail and may incorporate direct response methods to produce measurable actions.

DMCG’s operating principle that every campaign should begin with explicit economics, a defined audience and offer, a measurable response mechanism, disciplined testing, reliable source attribution, and a decision rule for rollout or revision. It should not be reduced to a creative style or a demand for immediate sales; its purpose is accountable, economically grounded learning and growth.

Marketing designed to prompt a specific, trackable action and connect that response to an audience, offer, creative, channel, or campaign. Often used interchangeably with the phrase: Direct Marketing

A lower-priced or lower-commitment offer presented when a prospect declines the primary offer or upsell.

Information an organization collects directly from its customers, prospects, website visitors, transactions, and owned channels.

The average number of times a person or household is exposed to a message during a defined period.

The initial offer and first transaction used to acquire a customer, before later backend purchases or renewals.

The process of delivering the promised product, service, information, or benefit after a response or purchase.

A multivariate design that runs every possible combination of the selected variables and variations. Three variables with three variations each create 27 test cells. Full-factorial designs can reveal interaction effects, but the required sample grows rapidly and may make the design impractical.

Revenue remaining after cost of goods sold. Formula: revenue − cost of goods sold; gross margin percentage is gross margin ÷ revenue × 100.

A portion of the eligible audience intentionally not exposed to a campaign so the marketer can estimate incremental impact.

A list or database of customers and prospects owned and maintained by the organization, rather than rented or purchased from an outside source.

A description of the type of customer most likely to receive strong value from the offer and produce attractive economics for the business.

One recorded display or delivery opportunity for an advertisement or marketing message; it does not necessarily indicate that a person noticed it.

The additional responses, conversions, revenue, or profit caused by a treatment beyond what would have occurred without it. Incremental lift is estimated by comparing the treated group with a valid holdout group that does not receive the treatment. A control can measure improvement over an existing treatment, but it establishes whether the result is truly incremental only when the control receives no treatment. Attributed outcomes should not be treated as incremental because some would have occurred without the campaign.

The additional responses, conversions, revenue, or profit caused by marketing beyond what would likely have occurred without it.

A metric selected to evaluate progress toward a specific business or campaign objective. Direct response KPIs may include response rate, conversion rate, cost per lead, cost per sale, average order value, customer lifetime value, and the allowable acquisition cost established by the campaign economics.

A focused web page built to receive campaign traffic and move visitors toward one primary action.

A person or organization that has expressed interest or supplied information indicating potential interest, but has not yet purchased.

A useful resource or incentive offered in exchange for contact information or another initial response.

A defined collection of customer or prospect records selected for communication, analysis, or campaign use.

An intermediary who helps marketers identify and arrange access to third-party mailing or audience lists.

A limited-use arrangement in which a marketer pays to contact a list owner’s audience under agreed terms without taking ownership of the underlying records.

A controlled comparison of a prospect or customer list, list segment, or selection criterion against a control list. The list is the most significant predictor of response, so the offer, creative, timing, and fulfillment must remain consistent for performance differences to be attributed to the list.

A comparison of customer lifetime value with customer acquisition cost. Formula: LTV ÷ CAC.

A blended measure comparing total revenue with total marketing spend. Formula: total revenue ÷ total marketing spend.

The percentage of submitted audience records that a platform or data provider can recognize and connect to usable profiles or identifiers.

The channels, placements, formats, and inventory used to distribute a marketing message to an audience.

A controlled comparison of individual media channels, placements, formats, schedules, or coordinated channel combinations. A test may compare one channel with another, a single channel with an omnichannel treatment, or different channel combinations. Keep the audience, offer, creative strategy, timing, and measurement rules as consistent as practical so you can attribute differences to the media plan. The objective is to identify the channel or combination that produces the strongest performance against the defined KPI and is economically suitable for rollout.

A test that changes two or more elements simultaneously to estimate the effect of each variation and identify the combination predicted to perform best. If three elements each have three variations, a full-factorial test creates 27 test cells. Because the audience is divided among many cells, each cell must receive enough observations for reliable analysis. The predicted winning combination should then be tested in a separate A/B split against the existing control to verify that it produces an actual, repeatable improvement before rollout.

The complete value proposition presented to the prospect, including the product or service, price, terms, bonuses, guarantee, urgency, and requested action.

A controlled comparison of value propositions, prices, terms, guarantees, bonuses, or incentives. The offer is often a stronger performance lever than minor creative changes. Avoid combining an offer change with unrelated audience or media changes unless using a valid multivariate design.

The percentage of delivered emails recorded as opened. Because privacy controls can affect tracking, it is best interpreted alongside clicks and conversions.

An affirmative action through which a person gives permission to receive specified communications or participate in a program.

A small, complementary add-on offered during checkout that can be accepted without leaving the primary purchase flow.

The time required for the contribution margin from a customer to recover the cost of acquiring that customer.

A small tracking mechanism used to record events such as page views, advertising exposures, and conversions and send that information to an analytics or advertising platform. Pixel data may support measurement, audience creation, and retargeting. Coordinating digital advertising with direct-mail recipients generally requires a separate identity-matching or household-targeting process.

The average profit generated by each order after subtracting the costs included in the campaign’s profit model, such as cost of goods, fulfillment, payment processing, and acquisition expense.

A person or organization that appears to fit the target market and could become a customer.

A lead that meets defined criteria indicating fit, need, authority, timing, intent, or another meaningful likelihood of becoming a customer.

The number or percentage of distinct people, households, or accounts exposed to a campaign at least once during a defined period.

A segmentation method that scores customers by how recently they purchased, how often they purchase, and how much they spend.

The percentage of delivered or contacted prospects who complete the defined response action. Formula: responses ÷ delivered pieces or eligible contacts × 100.

The percentage of customers or subscribers who remain active over a defined period. Formula: retained customers ÷ customers eligible to be retained × 100.

Revenue attributed to advertising divided by advertising spend. ROAS measures revenue efficiency, not profit.

The gain produced by an investment relative to its cost. A common formula is: (return − investment cost) ÷ investment cost × 100.

The average revenue generated for each delivered recipient in a campaign. Formula: attributed revenue ÷ delivered recipients.

The controlled expansion of a winning treatment to a larger qualified audience after testing. The qualified universe determines the maximum available scale. Roll out in stages when operational, financial, or customer-experience risk is material, and continue monitoring because timing, seasonality, exposure, and scale may change performance.

The number of eligible observations assigned to a test or test cell. The test sample should be randomly selected from—or demonstrably representative of—the qualified audience intended for rollout. In an omnichannel test, it should represent the audience that can be identified, reached, and coordinated across the combination of channels being tested. Eligible observations should then be randomly assigned among test cells whenever practical. Sample size should be planned before launch using the baseline rate, minimum meaningful improvement, required confidence, and expected response volume. An unrepresentative sample may not predict rollout performance, while stopping a test when early results appear favorable can produce a false winner.

The division of an audience into meaningful groups based on characteristics, behavior, value, needs, or likelihood to respond.

A unique identifier placed on a campaign, list, creative, offer, or media placement so responses and sales can be attributed and compared.

Another name for an A/B test: a controlled test that divides traffic or an audience among variations to compare performance.

A measure of whether an observed test difference is unlikely to be explained by random variation under the test assumptions. Statistical significance does not prove strategic or financial importance; the result must also be large enough to matter economically.

A file of records that must be excluded from a campaign, such as opt-outs, recent buyers, employees, duplicates, or ineligible contacts.

The percentage of eligible customers or prospects who accept a particular offer, option, upsell, or add-on. Unlike a general response rate, take rate measures acceptance of the specific offer.

The acquisition cost a campaign is expected or required to achieve based on the business’s economics and goals. It should not exceed the allowable CPA.

One defined audience-and-treatment combination in a test. Each cell must receive enough eligible responses or conversions for a reliable comparison. Adding cells without increasing sample size weakens the test.

The element intentionally changed in an experiment, such as the audience, offer, price, headline, format, or call to action. Limit variables according to the test design; changing several elements in an A/B test prevents a clean conclusion about what caused the result.

One alternative version of a test variable. Each variation should represent a distinct, deliberate hypothesis about what may improve performance. Adding variations that differ only slightly—or that have no clear strategic rationale—creates additional test cells and allocates fewer observations to each one, reducing the test’s ability to produce meaningful comparisons and useful guidance for rollout.

A URL containing identifiers or parameters that help connect visits, responses, and conversions to their campaign source. A tracking link may contain source codes or UTM parameters identifying the channel, medium, campaign, or creative.

An offer that encourages a buyer to choose a higher-value version, larger quantity, premium option, or additional benefit. Upselling may occur before, during, or after a purchase and is often used to increase average order value or customer lifetime value.

A cost that changes with the number of responses, orders, customers, units, or transactions, such as fulfillment, payment processing, or commissions.

A campaign designed to reactivate former customers, lapsed buyers, or inactive subscribers.