Reference

Marketing terms, in plain language.

Every term below is defined the way we use it in a brief - short enough to quote, specific enough to argue with. Where the industry disagrees with itself, we say which side we take and why.

Marketing terms, in plain language.

Marketing agency

A marketing agency is an outside team that plans and produces a company's brand, content, advertising and analytics work - either instead of in-house staff or alongside them. The useful distinction is not size but ownership: an agency worth hiring leaves you owning the accounts, the files and the audience it built.

Agencies differ far more in operating model than in service list. Some sell hours, some sell deliverables, some sell a retained system. The question that separates them is what remains yours when the relationship ends - the ad accounts, the pixel history, the source files, the audience lists. If the answer is 'ours', you were renting your own marketing.

Related: Retainer (vs project) Brand system

Brand system

A brand system is the set of rules, assets and decisions that let a team produce consistent work without asking a designer every time. It is larger than a logo and smaller than a strategy: colour and type, yes, but also the decision rules, the format kits and a named owner with a process for changing them.

A brand book that only describes appearance stops being useful the first time somebody needs a format it does not cover. A system answers the next question too: who decides, what happens when a rule does not fit, and how a change gets recorded. Consistency is not a design outcome - it is an operating one.

Related: Brand book Content system

Brand book

A brand book is the document that records a brand's visual and verbal rules: logo use, colour, typography, imagery, tone of voice and the wrong ways to use each. It is a reference, not a system - it describes what the brand looks like, without saying who decides when a new case appears.

Most brand books are read once and then consulted only in disputes. That is fine, provided something else carries the operating half: templates people actually open, and a named person who owns changes. A brand book with no owner ages into a PDF everyone quotes and nobody follows.

Related: Brand system

SMM (social media marketing)

SMM is the planning, production and distribution of content on social platforms, together with the community and paid promotion around it. It covers what gets posted, when, in what format, and what happens in the replies - which is the half most plans leave out.

SMM is usually sold as a number of posts per month, which measures production rather than results. The more useful contract states what the account is for - reach, enquiries, service load - and which two or three numbers will be watched. Posting volume is an input; nobody buys inputs on purpose.

Related: Content system Conversion

Content system

A content system is the repeatable process behind publishing: where ideas come from, who approves them, which formats exist, how assets are produced and how performance feeds the next round. It is what turns content from a series of one-off requests into something a team can run without heroics.

The test for a content system is whether the second month is easier than the first. If every piece still starts from a blank brief and ends in an ad-hoc approval chain, there is no system - there is a person working hard.

Related: SMM (social media marketing) Evidence loop

SEO (search engine optimisation)

SEO is the work of making a site findable and trustworthy in search results: the technical health of the pages, the structure and depth of the content, and the references pointing at it from elsewhere. It is a compounding channel - slow to start, and slow to lose once it works.

SEO fails most often for one of two boring reasons: pages exist that answer nothing anyone searches, or pages exist that answer well and nothing links to them. Both are content problems wearing a technical costume.

Related: GEO (generative engine optimisation) Conversion

GEO (generative engine optimisation)

GEO is the work of being found, quoted and cited correctly by AI answer engines rather than only by search results pages. It rewards clear entity information, self-contained factual passages, structured data and corroboration from sources the model already trusts.

The practical difference from SEO is what gets returned. A search engine returns your page and the reader decides; an answer engine returns a sentence and names a source. That changes what is worth writing: passages that answer completely on their own, and facts a model can check somewhere other than your own site.

Related: SEO (search engine optimisation) Structured data

Structured data

Structured data is machine-readable markup added to a page - usually JSON-LD following schema.org - that states plainly what the page is about: an organisation, a service, an article, a question and its answer. It does not change what a visitor sees; it changes what a machine can be certain of.

Structured data cannot make a claim true, and search engines increasingly ignore markup that a page does not visibly support. Treat it as a way to remove ambiguity from something already on the page, never as a way to assert something that is not.

Related: GEO (generative engine optimisation) SEO (search engine optimisation)

Performance marketing

Performance marketing is paid promotion bought against a measurable outcome - a lead, a purchase, an install - rather than against exposure. The defining feature is not the platform but the accounting: every campaign has a target cost per outcome, and spend follows the result.

Performance marketing works exactly as well as the measurement underneath it. Buying against a conversion the site cannot reliably record produces confident optimisation towards noise, which is more expensive than not optimising at all. Build the measurement before the media runs.

Related: Cost per lead (CPL) Attribution Conversion

Conversion

A conversion is a visitor action a business has decided to count: an enquiry, a call, a booking, a purchase, a signup. Conversion rate is that count divided by the visitors who could have taken it. Both numbers are only as meaningful as the decision about which action counts.

The common mistake is counting whatever is easy to record. A newsletter signup and a request for a quote are both conversions, and treating them as one number hides which of the two the campaign actually produced.

Related: Qualified lead Conversion rate optimisation (CRO)

Qualified lead

A qualified lead is an enquiry that meets a stated definition of a real potential client - the right kind of business, a real need, a workable budget and someone who can decide. Without that definition written down, 'leads' counts form submissions, which is a measure of traffic rather than of demand.

Agreeing the definition before a campaign starts is the single cheapest improvement available to most marketing. It changes what the media optimises towards, what the report means, and whether the sales team trusts the number at all.

Related: Cost per lead (CPL) Conversion

Cost per lead (CPL)

Cost per lead is the media spend divided by the number of leads it produced. It is useful only against a definition of 'lead' and a known close rate: a cheap lead that never becomes a client is more expensive than an expensive one that does.

CPL is the number most often quoted between agencies and clients, and the one most easily improved by lowering the standard for a lead. Ask for it alongside the qualification definition and the eventual close rate, or it measures nothing.

Related: Qualified lead Performance marketing

Attribution

Attribution is the set of rules that decide which marketing touchpoints get credit for a result. Last-click gives everything to the final step, first-click to the first, and data-driven models spread it - so the same campaign can look like a success or a failure depending only on the rule chosen.

No attribution model is correct; they are different simplifications of something unobservable. The practical approach is to fix one model, state it in every report, and use holdouts or controlled tests when a decision is large enough to deserve real evidence.

Related: Performance marketing Evidence loop

Retainer (vs project)

A retainer is an ongoing monthly engagement covering continuous work - running campaigns, publishing, measurement and iteration. A project is a scoped piece with a definite end, such as an identity or a website. The difference is not price but whether the work has a finish line.

Work that compounds - search, content, media optimisation - loses most of its value when it stops and restarts, which is what a retainer is for. Work that ends when it is done should not be a retainer; paying monthly for a finished website is the most common way an agency relationship quietly turns into rent.

Related: Marketing agency

Conversion rate optimisation (CRO)

CRO is the practice of increasing the share of visitors who take the action a page exists for, by changing the page rather than buying more traffic. It covers offer clarity, page structure, form friction, page speed and proof - tested rather than assumed.

CRO is usually the cheapest growth available, because the traffic is already paid for. It is also the easiest to fake: a test that ran for a week on a few hundred visitors can show anything at all. If a result cannot survive being stated with its sample size, it is not a result.

Related: Conversion Qualified lead

Evidence loop

An evidence loop is the working cycle in which measurement changes a decision, the decision changes the next piece of work, and that work produces new measurement. It is the difference between reporting - which describes what happened - and learning, which changes what happens next.

The test is whether any number in the last report changed a decision. If not, the report is a record rather than a loop, and the metrics in it were chosen for availability rather than for consequence.

Related: Attribution Content system

AI in marketing

AI in marketing means using generative and analytical models inside the working process - research, segmentation, first drafts, production, reporting - rather than selling them as a separate product. It accelerates the middle of a task; the framing at the start and the judgement at the end stay human.

The useful split is by task, not by tool. AI compounds value on work with a checkable answer - analysis, structure, variation, routine automation. It erodes trust on work whose value is that a person is accountable for it: strategy, claims, brand voice, and final approval.

Related: Content system GEO (generative engine optimisation)

Ask about your own case

Free audit

Not sure where to start? Start with the audit.

We review your brand, content, media and digital presence, then send back a short, honest read of where the leverage is.

Request a free audit