DMI Expert key terms

A staged framework describing how an organisation's digital capabilities, culture, and processes advance from ad hoc experiments to integrated, data-driven, continuously optimised operations. It is used to benchmark current state and plan capability investment.

A sequenced plan that links digital ambition to concrete initiatives, owners, budgets, and milestones across a multi-year horizon. It balances quick wins with foundational work such as data infrastructure and skills development.

A goal-setting method pairing a qualitative, inspiring objective with a small set of measurable key results. Digital teams use it to align marketing, product, and analytics work to shared outcomes rather than output volume.

A single measure that best captures the value customers receive from a product or service and correlates with long-term business growth. It guides prioritisation when teams face competing short-term requests.

A hierarchical decomposition that breaks a top-level business goal into the driver metrics that mathematically or logically produce it. It helps marketers see which levers actually move the headline number.

The practice of assigning credit for a conversion to the touchpoints a customer encountered beforehand. Models range from single-touch rules to data-driven and incrementality-based approaches, each with distinct bias trade-offs.

An experimental method that measures the lift caused by a marketing activity by comparing treated and untreated groups. It answers whether spend created outcomes that would not have happened anyway.

A randomly selected segment deliberately excluded from a campaign or treatment so its behaviour can serve as a baseline. Comparing holdout and exposed groups isolates the true causal effect of the activity.

A statistical technique that regresses historical sales against media spend, pricing, seasonality, and external factors to estimate each channel's contribution. It works without user-level tracking and suits privacy-constrained measurement.

The structured approach to helping people adopt new processes, tools, and behaviours during transformation. It addresses communication, sponsorship, training, and resistance so that digital initiatives deliver their intended value.

The combined set of software platforms a marketing organisation uses, such as CRM, automation, analytics, and content tools. Deliberate architecture prevents duplicated data, overlapping licences, and integration debt.

A system that unifies customer records from multiple sources into persistent, privacy-compliant profiles available to other tools. It enables consistent segmentation and personalisation across channels.

The page a search engine returns for a query, containing organic results, paid ads, featured snippets, local packs, and other modules. Layout and feature mix strongly influence click-through behaviour.

Any non-standard element on a results page, such as knowledge panels, image packs, shopping units, or people-also-ask boxes. Winning them can capture attention without a top organic link.

A highlighted answer box that search engines extract from a page to address a query directly. Earning one can raise visibility and brand authority even when the page does not rank first organically.

The underlying goal a person has when entering a query, commonly classified as informational, navigational, commercial, or transactional. Matching content format and offer to intent is central to search performance.

The process of discovering and prioritising the queries an audience uses, weighing relevance, demand volume, competition, and business value. Outputs feed content planning, paid search structure, and product messaging.

A highly specific, usually lower-volume query phrase that reflects a narrow need. Such terms often convert better and face less competition, making them efficient for content and paid campaigns.

The work of making a site crawlable, indexable, fast, and structurally sound for search engines. It covers robots directives, canonical tags, site architecture, rendering, and Core Web Vitals.

An HTML link element that tells search engines which URL is the preferred version of a page. It consolidates ranking signals when duplicate or parameterised URLs exist.

The approximate number of URLs a search engine is willing and able to fetch from a site in a given period. Poor architecture or excessive low-value pages can waste it.

A set of user-experience metrics covering loading speed, interactivity, and visual stability. They are used as page-experience signals and as practical diagnostics for improving conversion.

An ad platform's estimate of how relevant an ad, keyword, and landing page are to a user's query. Higher scores can lower cost per click and improve ad position.

The formula an ad auction uses to decide whether an ad shows and where, combining bid, ad quality, expected impact of extensions, and context. It explains why higher bids do not always win.

An assessment of how relevant, transparent, and easy to navigate the destination page is for someone clicking an ad. It feeds ad quality scores and directly affects conversion rates.

A paid search bidding strategy where bids are set automatically for each impression based on predicted likelihood of a click or conversion. It shifts control from manual keyword bids to target setting.

Revenue generated per unit of advertising spend, usually expressed as a ratio. It is a core efficiency measure but must be read alongside margin and incrementality to avoid misleading conclusions.

Monitoring online conversations, mentions, and sentiment about a brand, competitors, or category. It surfaces emerging issues and audience needs that surveys may miss.

Automatically classifying text as positive, negative, or neutral toward a subject. Accuracy varies by language and context, so results are best used as directional signals rather than precise measures.

The day-to-day practice of responding to comments and messages, moderating discussions, and nurturing relationships in brand-owned spaces. It converts passive audiences into engaged advocates and protects reputation.

Partnering with individuals who have audience trust to promote a product or message. Success depends on audience fit, disclosure compliance, and measurable objectives rather than follower count alone.

A document giving a creator the objectives, key messages, mandatory disclosures, deliverables, and creative guardrails for a partnership. Good briefs preserve authenticity while protecting brand and legal requirements.

Media created by customers or fans rather than the brand, such as reviews, photos, and videos. It often feels more credible than brand-produced assets and can be reused with permission.

Selling products directly within social platforms through shoppable posts, catalogues, and in-app checkout. It shortens the path from discovery to purchase but requires robust product feeds and trust signals.

Systems that produce new text, images, audio, or code from learned patterns in training data. In marketing they accelerate content production but require human review for accuracy, brand fit, and rights.

A neural network trained on vast text corpora to predict and generate language. It powers drafting, summarisation, classification, and conversational interfaces, subject to limits in factual reliability.

The practice of designing instructions, context, and examples so a generative model produces useful, consistent output. Clear constraints and evaluation criteria matter more than clever phrasing.

A confident but fabricated or incorrect output from a generative model. Mitigation includes retrieval of verified sources, constrained tasks, and human fact-checking before publication.

An approach where a model retrieves relevant documents from a trusted store and uses them as context when answering. It improves factual grounding and lets organisations control the knowledge a system draws on.

The policies, roles, and controls an organisation uses to ensure AI systems are lawful, fair, secure, and accountable. It covers data provenance, disclosure, human oversight, and model risk review.

Telling audiences when content, recommendations, or interactions are produced or materially shaped by AI. Transparency obligations vary by jurisdiction and platform, and undisclosed use can damage trust.

Systematic skew in model outputs caused by unrepresentative training data, flawed labelling, or objective design. In marketing it can produce unfair targeting, stereotyping, or exclusion of audience segments.