What marketing technology is for
Every marketing team already has a stack, whether or not anyone designed it. A website, an analytics property, an advertising account, an email platform and a spreadsheet of customers is a stack; so is a warehouse, a customer data platform and a dozen activation channels. The difference between the two is not sophistication but whether the pieces were chosen for a purpose and connected on purpose.
The purpose is always one of a few jobs. Collect: record what people do on the site, in the app, in email and in the sales process. Store and unify: hold that record in a form where one person is one person, whatever device or channel they used. Decide: segment, score, test and choose what each person should see next. Deliver: put the message or the offer in front of them through a channel, and report what happened. Each tool in a stack does one or more of these jobs, and the stack works when the jobs are covered once each and the hand-offs between them are reliable.
The landscape of vendors is famously large; the annual marketing technology landscape published by chiefmartec.com has grown for over a decade and groups thousands of products into a handful of categories. The scale is why this hub starts with the system rather than the market. A team that knows which jobs it needs done and how data should move between them can read any vendor map; a team that starts from the vendor map ends up buying jobs it already has. The designing a marketing stack guide is where that method is laid out, and the sections below give the shape.
The stack
The marketing stack section treats the technology as layers. At the bottom is collection: the tags, pixels, software development kits and server-side endpoints that record events. Above it is the data layer: the warehouse or customer data platform (CDPUnified first-party customer profiles.) that stores events and resolves them to people, and the identity and consent rules that decide what may be stored and used. Above that is decisioning: segmentation, scoring, experimentation and the automation that acts on rules. At the top is activation: the channels (advertising platforms, email, messaging, on-site personalisation) that deliver, and the analytics and reporting that close the loop.
The core guides cover the section. The design guide sets the method: start from the use cases the business has, map them to layers, decide what to build and what to buy, and choose tools for how well they connect before how many features they have. The data flow and integrations guide follows the data from source to activation, including identity resolution, the complementary roles of a warehouse and a customer data platform, and the reverse flows that push segments back out to channels. The tag governance guide deals with the collection layer as an organisational problem: what tags are on the site, who approved them, whether they respect consent, and what they cost the page. Further guides in the section turn the layers into routines: running a stack audit, writing the use-case list, reviewing access to marketing tools, mapping consent across the stack, documenting integrations, deciding how long tools keep personal data, test environments and test data and backing up tool data and settings.
The stack is where a tag management system such as Google Tag Manager (GTMTag management system.) earns its place: not as a tool among tools but as the control point for the collection layer. The Google Tag Manager section of the analytics hub covers the container mechanics; this hub covers how the container fits the rest.
The tools
The tool categories section is the vendor-neutral map. It explains how the landscape is organised, what each category does, where categories overlap and how to choose within one. The MarTech categories guide walks through the major groups (advertising and media, content and experience, social and relationships, commerce and sales, data and analytics, management), and the choosing and evaluating tools guide gives a method for selecting within a category: requirements from use cases, scoring criteria, a proof of concept with your own data, and the integration checks that most selections skip. Further guides follow a tool through the rest of its life in the stack: handling requests for new tools, reviewing a tool before renewal, reviewing licences and seats, retiring a tool, preparing for a tool outage, hiring a technology partner and the choice between a suite and best-of-breed tools.
This hub does not rank vendors, and the guides beneath it do not either. The reasons are practical rather than principled. Products in a category change faster than a page can be updated; the right product depends on the stack it has to join, which the page cannot know; and most selection failures come from an unclear requirement rather than a wrong vendor. The marketing tools resource lists the tools this site documents, and the marketing automation hub covers the category that most teams buy first, but neither is a league table.
Data as the connective tissue
Tools are the visible part of a stack. Data flow is what makes it a system. A stack where the email platform, the advertising accounts, the analytics property and the customer relationship management system each hold their own version of the customer is four stacks sharing a logo. The connective tissue is a shared definition of the customer and the events that matter, a place where that definition is enforced, and integrations that carry it to every tool that acts on it.
Three pieces matter most. Identity: a consistent way of recognising the same person across site visits, email
clicks, purchases and support tickets, within what consent allows. Event definitions: an agreed list of the
events the business cares about, with the same names and parameters wherever they are collected, so that a
purchase in analytics is the same thing as a purchase in the warehouse. Data quality: checks that the events
arrive, are complete and match the system of record, which the
data quality guide in the analytics hub covers in detail.
Older stacks were built around a data management platform (DMPPlatform that organises audience data for targeting.) fed by third-party cookies; the current direction is first-party data, consented, held in a warehouse or customer data platform and activated through server-side connections and the platforms' own matching. Google Analytics (GA4Google's event-based analytics platform.) and the advertising platforms sit at the ends of that flow as both sources and destinations. Privacy and consent rules shape all of it, and the privacy and consent guide sets out the general direction without being legal advice.
Keeping the stack lean
The natural tendency of a stack is to grow. Each new campaign brings a tag, each new team member brings a favourite tool, each vendor adds a module. Lean is not a size; it is a discipline of asking, for every tool, what job it does, who owns it, what it is connected to and what would break if it were removed. The same question now arises for AI, and the comparison of standalone AI assistants and built-in AI features helps decide whether a task needs a new tool or the AI already inside the stack.
| Question | Healthy answer | Warning sign |
|---|---|---|
| What job does this tool do? | One or two named jobs with a named use case | "It does lots of things", or nobody is sure |
| Who owns it? | A named person who can log in, configure and renew | A former employee or an agency |
| What does it connect to? | Documented integrations, with data moving both ways where needed | Manual exports and uploads |
| What would break if it went? | A known list of campaigns, reports or flows | Nobody wants to find out |
| Does another tool already do this? | No overlap, or the overlap is deliberate | Two tools doing the same job for different teams |
| Is it tagged on the site? | Only where needed, consent-gated, in the inventory | On every page, undocumented |
| When was it last reviewed? | At the last stack review, with a decision recorded | At purchase |
The questions are the stack review in miniature. Any tool that fails two of them is a candidate for consolidation or removal, and the process for deciding is in the choosing and evaluating tools guide, which works as well for keeping a tool as for buying one. The stack audit tutorial asks these questions of every tool at once, the licence and seat review checks what each one pays for against what is used, and the retirement tutorial removes a tool without leaving its tags, integrations or access behind.
A subscription box company audits its stack
A subscription box retailer with a marketing team of six lists every tool with a login or a tag on the site and finds 31 of them, 19 with a recurring cost. The audit maps each to a job: four tools send email in some form, three run surveys, two manage social scheduling and nobody can name the owner of 7 of the 31. Tagging shows 14 scripts loading on the product page, of which 5 belong to tools whose contracts lapsed. The team consolidates email to one platform, removes the lapsed tags, assigns an owner to every remaining tool and sets a rule that no new tag ships without an entry in the inventory. The recurring bill falls by roughly £2,400 a month; more usefully, the product page's largest contentful paint improves by about 0.6 seconds once the dead scripts are gone, and the quarterly review now takes an afternoon rather than a fortnight.
A first stack review
If the stack has never been reviewed, the first pass is less about choosing tools than about seeing what is there. It takes a few weeks of part-time effort for most teams and produces three artefacts. The stack audit tutorial runs the review in order, from the scope to a decision on every tool.
An inventory. Every tool with a login, a contract or a tag on the site, with its owner, cost, renewal date, the jobs it does and the integrations it has. The tag governance guide covers the tag half of this, and the review can start there, because the site's tags reveal tools nobody remembers buying. Beside the owner, record who else can log in; the access review checks every account against a current person, supplier or integration.
A use-case map. The marketing use cases the business runs or wants to run, written in plain language (recover abandoned baskets; follow up trial sign-ups; report revenue by channel), each mapped to the tools and data it needs. Gaps show where a tool is missing; duplicates show where tools overlap. The designing a marketing stack guide builds the whole stack from this map, and the tutorial on writing the use-case list writes and maps it step by step.
A data flow diagram. One page showing where each event is collected, where it is stored, how identity is resolved and which tools consume it. Most first diagrams reveal that the same event is collected three ways and stored twice, and that the report everyone trusts is built on a manual export. The data flow guide explains how to read and fix the picture, and the integration register records each connection the diagram is drawn from.
With those three in hand, the measurement plan that defines the business's KPIA metric tied directly to an objective. set has a stack it can be built on, and the sections of this hub give the depth. Start with the stack section if the problem is that the pieces do not connect; start with the tools section if the problem is choosing or consolidating. Both end at the same place: a small number of tools, each with a job, an owner and a connection.
