Glossary

What is a marketing engineer?

A marketing engineer works at the intersection of marketing, data and software engineering — building and operating the technical stack that AI-era marketing runs on: prompt pipelines, agent workflows, AdCP and MCP integrations, GEO tooling and custom measurement.

Marketing practice

All 220 terms

Why the role appeared

Marketing has employed technical people for years — analysts, marketing operations, growth engineers. What is new is that the marketing work itself has become software.

Measuring AI share of voice means building a sampling harness that runs prompt sets against several assistants on a schedule. Making a product available to agents means exposing it over MCP. Buying conversational inventory programmatically means integrating AdCP. Measuring anything credibly now means running your own holdouts rather than reading a platform's report.

None of that is configuration. It is engineering, and it sits inside the marketing function because the questions it answers are marketing questions.

What the role actually builds

Prompt harnesses for measuring brand presence across assistants, repeatably and at fixed sample sizes.
Agent integrations — MCP servers exposing catalogue, pricing and availability so agents can use them accurately.
Measurement infrastructure, particularly holdout design and geo experiment tooling.
Data pipelines joining platform exports to first-party outcomes — see ETL.
GEO tooling for monitoring how the brand is described, and detecting when it is described wrongly.

Why it is not a growth engineer

A growth engineer optimises a product's own funnel — onboarding, activation, conversion. A marketing engineer works on the systems through which the brand is discovered and evaluated outside the product, most of which the company does not control.

The distinguishing skill is comfort with measurement under uncertainty. Much of what this role builds produces noisy, partial signals about systems that are opaque by design, and the value depends on knowing which conclusions the data can support — which is closer to applied statistics than to feature work.

Common questions

Is this a marketing role or an engineering role?

Both, and reporting lines vary. What is consistent is that the person needs to write real code and to understand why a marketing question is being asked, since the failure mode is building instrumentation that measures the wrong thing precisely.

Do I need one?

If AI-surface presence matters to your business and you are currently measuring it by hand or not at all, yes. The work does not fit into an existing analyst or agency remit.

Back to the glossary