The short answer

Agile Marketing primarily concerns how marketing teams organise and execute work. Adaptive Marketing concerns how marketing changes in response to the outside environment.

Agile Marketing borrows its practices from agile software development: short iterations, small cross-functional teams, visible backlogs, frequent review, a preference for shipping something small over planning something large. Its subject is the team.

Adaptive Marketing, as defined on this site, is the continuous adjustment of strategy, messaging, channels, evidence, and execution in response to meaningful changes in customer behaviour, market conditions, performance, technology, and context. Its subject is the market.

Where the two actually differ

DimensionAgile MarketingAdaptive Marketing
Primary subjectThe team and its processThe market and the customer
OriginAgile software development, extended to marketingMarketing capability research, notably Day (2011), and market-sensing practice
Core questionHow do we work in smaller, faster increments?What has changed, and what should we change because of it?
Unit of workThe sprint, the backlog item, the standupThe signal, the interpretation, the decision
Failure modeExecuting the wrong plan very efficientlyRecognising the change and having no ability to act on it
Evidence it is workingCycle time falls, throughput rises, work is visibleDecisions trace to signals, and some of them are decisions not to act
Can exist without the otherYes, and frequently doesYes, on a slower rhythm

Neither column is a criticism. Agile practices solve a real and common problem: marketing work that takes three months to reach the market and cannot be corrected once it has. Adaptive practice solves a different one: marketing that is delivered promptly and is aimed at a customer who has moved.

Three combinations, all of which occur

Agile but poorly adaptive. The most common of the three in well-resourced teams. Sprints run cleanly, the board is current, work ships every fortnight, and nobody has asked in eighteen months whether the positioning the work expresses still matches what buyers care about. The velocity is real. The direction has not been examined, because no ceremony in the process owns that question.

Adaptive without formal agile methodology. Common in small businesses and in senior teams that have been together a long time. There is no board and no sprint. There is somebody who talks to customers weekly, notices when the questions change, and is willing to change the plan. This is genuine adaptation, and its characteristic weakness is that it lives in one person's head, so the organisation cannot tell which of its changes worked.

Both. The two compound, because agility shortens the interval between deciding to adapt and having adapted. The value of a signal decays: an insight about a shifting objection is worth considerably more in the week it appears than in the quarter it finally reaches a brief. Agile practice is the most reliable way of shortening that interval, which is why the pairing is worth having even though neither requires the other.

Why the confusion persists

Both involve changing things often, and from outside a single organisation the two look identical in any given week. The distinguishing evidence is not the rate of change but what preceded it. An adaptive change names the condition that moved and states what it should shift. An agile change names a backlog item.

There is also a vocabulary problem. Agile practice uses the word adapt in its own manifesto, meaning responding to changes in requirements and in the work. That is a different referent from adapting to a change in the market, and the shared word does a lot of quiet damage.

What this means in practice

If marketing is slow to reach the market and impossible to correct once launched, the problem is probably agility, and agile practices address it directly.

If marketing reaches the market quickly and performance is drifting for reasons nobody can name, the problem is probably adaptive, and no amount of process improvement will touch it. The specific thing usually missing is ownership of the interpretation step: somebody accountable for deciding which observed changes matter enough to act on. Sensing is usually distributed across analytics, sales, and support. Response is usually distributed too. Interpretation is the step with no deliverable, which is why it is the one that ends up belonging to nobody.