Why the stack routines ask for the list
Several routines on this site send the reader to a use-case list. The stack audit tutorial pairs each tool with the use cases it supports and treats a tool that supports none as a candidate for retirement. The new tool request guide triages each request by asking which written use case it meets. The renewal review tutorial re-scores a tool against the use cases the business has now, not the ones it had on the day of purchase. And the stack management playbook takes the use-case map into its yearly audit and roadmap. Each speaks of the list as though it already existed.
Where it does not, a request has no written use case to be checked against, an audit has no use cases to map tools to, and a roadmap can drift into a purchasing schedule. The marketing technology hub names the list as one of the artefacts of a first stack review, beside the inventory and the data flow diagram. The audit produces the inventory, and the integration register tutorial produces the record the diagram is drawn from. This tutorial is the method for the third.
Two neighbours can be mistaken for the list. The choosing and evaluating tools guide converts the use cases for a layer into requirements for one candidate, whereas the list sits above any selection. The measurement plan tutorial works down from agreed objectives to events and a tracking specification, and the designing guide calls the list its operational twin: the plan says what is measured, and the list says what the stack must do. The stack recordThe live register of every marketing tool: owner, deputy, cost, notice date, users, use cases and status, kept current by onboarding, reviews and retirement. lists tools; the list names jobs, and the two meet in one column that says which tool does which job.
Before you start
Decide the scope. A list for the whole stack takes in all channels and teams; a narrower list can cover one layer or one team's work if it says so, so that the reader knows why a job is missing.
Gather what already exists, however rough. Look first for the measurement plan or, if the business has none, its agreed objectives and KPIsA metric tied directly to an objective., because they supply the outcomes the use cases serve; the measurement plan framework names the pieces. The campaign calendar shows what marketing does and when, and any automation inventoryThe single list of every active automated workflow, kept outside the platform, with its purpose, owner, trigger, exits, dependencies, measure and review dates. lists what runs without a person; the automation governance guide explains how one is kept. A stack record or request log helps too. None of these is essential: where they are missing, the first pass is conversations, and the list becomes the first written record of what the stack is for.
Set aside time with people rather than documents: time with each channel owner and with whoever prepares the regular reports, and a short conversation with whoever owns data protection before the consent column is filled in. Write the roles into a RACI matrix: a single accountable person for the list, each channel owner responsible for their own lines, and the data protection owner and engineering consulted on the columns that touch them. Keep the list as one shared table with one owner and a change history, beside the stack record.
The steps
Agree what the list is for and who answers for it
Write down the routines that will read the list: the audit, the request route, the renewal review and the roadmap. Naming the readers decides how the lines are written, because a line the request route cannot use to say yes or no is not yet a use case. Name the one person accountable for the list, and agree two rules before any line is drafted. First, no line names a tool or a feature, which keeps the list free of the stack it is meant to judge. Second, each line carries the name of the person who runs the job or asked for it, which keeps it from filling with lines nobody can speak for.
Collect the jobs the team already runs
Start from the places where a job leaves a trace, and write what you find as raw lines in the words people use. Do not tidy yet. Note where each line came from: a job that shows up in several sources is corroborated, and a job that shows up in only one deserves a question about whether anyone still needs it.
Where jobs hide What it shows What to ask The campaign calendar What marketing does and when Which part of this needs a tool to happen at all? Automations and flows What runs without anyone pressing a button What is each one for, in a sentence? Reports and dashboards people open What the business reads Which decision does each one inform? Exports and uploads done by hand Jobs where a person is the connector What stops if that person is away? Requests and workarounds Needs the stack does not meet Which request keeps coming back? The measurement plan The outcomes the business has agreed to chase Which of these needs a stack job behind it? Ask each channel owner a plainer question as well: if this tool were switched off tomorrow, what could they no longer do for customers or colleagues? The answers are use cases in the owner's own words. Include the manual routines: a spreadsheet that someone updates for a standing meeting is doing a job for the business.
Write each job as one sentence with the outcome it serves
Turn each raw line into a sentence that begins with what marketing does, names who or what it is done for, adds the timing only where timing matters, and ends with the outcome. As general practice, read the outcome aloud: if it could be the heading of a report, it is an outcome. The landscape published by chiefmartec.com is a useful reference for vocabulary when a team member describes a job in a vendor's words, but it is a map of products and will not tell you what your business needs.
Some examples of the form, written for a made-up shop. "Remind a customer about an order that has been dispatched, so that fewer deliveries are missed." "Add a show-stand sign-up to the customer list, so that the shop can reach them like any online buyer." "Tell the sales team when a form asks for a call, so that the call is made promptly." "Keep anyone who has unsubscribed out of any send, whichever tool sends it." Each names a doer, an audience and a result; none names a product, a feature or a setting.
When a line cannot be written without a tool, it is describing the tool. Ask what the tool is for and write that. "Use the email platform's automation builder for the welcome series" becomes "bring a new customer through onboarding so that they use what they bought", and the platform moves into a column where it can change.
Test the grain: split what bundles and merge what repeats
As general practice, read each sentence against four tests. One outcome: if the sentence ends two ways, with an "and" that joins different results, split it. One owner: if two teams run different halves, write a line for each, or name a single owner for the outcome and let the others be consulted. One judge: if two lines would be judged by the same measure and run by the same people, merge them, because counting one job twice can make later gaps and overlaps look larger than they are. One reader: if someone who has not seen the stack could not say how they would know the job worked, the line is too vague.
Then group the lines by the outcome they serve, using the groups the designing guide offers: acquisition, conversion, retention and reporting. The groups are for reading, not scoring: a reader scanning one group can see at once whether retention is crowded with tools and reporting has none.
Mark each line live, planned or an aspiration, and name who runs it
Give each line one of the three statuses the designing guide names, and define them for your own list. Live means the job is running now and someone could show it working. Planned means a team owns it and intends to build it. An aspiration has been asked for by someone, but no team has yet agreed to own it; it stays on the list as a record of demand, not a commitment. Keeping the three apart lets the audit read the live lines against the tools in use and the roadmap read the planned ones.
Add the owner: the team that runs the job and a named person within it, or for an aspiration the person who asked for it until a team agrees to take it on. Add one more flag: whether the job depends on a person doing something by hand. A live use case that runs on a manual export is real but fragile in a way the status does not show, and the flag is what sends it to the integration registerOne standing record of every connection between marketing tools: what moves, which way, keyed how, on what schedule, with whose credentials and owner., which treats a person acting as the connector as a connection like any other.
Record what each use case needs from data and from consent
For each line, write what it needs in order to work. Which events or records does it depend on, and where do they come from? The tracking plan defines the events the site and app emit, and the customer or order record sits in the CRMThe system that stores an organisation's contacts, companies, deals and interactions, and the processes built around it. or the commerce system, depending on the business. Which identifier does the job key on, and does each source the job touches carry it? How soon must it act? The data flow and integrations guide sets out why the latency a job needs should decide how it is connected.
Then record whether the job uses personal dataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier. and, if it does, which purpose it serves. The consent mapping tutorial works in purposes rather than tools, and its first step lists them. A use case that fits none means either a purpose is missing from the consent map or the job should not be running on personal data. Leave the basis blank until the person who owns data protection has decided it: the list records what a job needs, and it is not the place to settle what the business may do. The retention guide also works by purpose, so the same column saves work there.
Map each use case to the layers and then to the tools
Now place each line on the stack; the audit later reads the same map from the tools' side. The marketing stack hub describes the layers: collection, data, decisioning, activation and measurement. For each use case, mark which layers it touches and write the tool that does the work in each. Where the work is done by a person with a spreadsheet, write that, because it is the honest answer and the audit needs it. A tool whose category you cannot place can be looked up in the MarTech categories guide.
Use case Collection Data Decisioning Activation Measurement Remind a customer about a dispatched order Shop events through the tag manager The shop's order record Email platform rule Email platform Email platform report Add a show-stand sign-up to the customer list Paper sign-up sheet at the stand CRM customer list, typed in by hand None None Nothing; nobody checks it Tell sales when a form asks for a call Form event CRM lead record CRM assignment rule CRM alert None The last two rows show what a first map can reveal: jobs that work and are measured nowhere. Resist fixing them in the table. The map's task is to show the stack as it is, including a customer data platform (CDPUnified first-party customer profiles.) or warehouse that serves no line, so that the next step reads facts and not intentions.
Read the map for gaps, overlaps and tools that serve nothing
Read the finished map for what only the list can show. Down each use case, note any line that has a layer no tool serves; that is a gap. Along each tool: does it appear on any line? A tool on none may be running a job the list missed, so ask its owner what it is used for before anything else, and if there is such a job, that line goes in. And along the flags: which live jobs rest on a person or a manual step, and which use personal data under a purpose the consent map does not have? A job served by two tools shows on the map as well, but the audit marks overlaps and tools with no job when it reads this map from the tools' side and decides them, so this step only notes them and passes them on.
Sort what you find by where it goes next, and decide none of it here. Gaps go to the roadmap, which is where the audit sends them. Overlaps and tools with no job go to the audit, whose decisions are keep, fix, consolidate or retire. Manual dependencies go to the integration register (the audit also counts a use case served by a manual workaround as a gap, so they are noted for the roadmap too), and a record that exists in one place only is checked against what the backing up guide says deserves a copy of its own. Missing purposes go to whoever owns the consent map. The list reports what it sees; each of those routines decides.
Rank the gaps and agree the list with the people who fund and use it
Take the planned lines and the aspirations that now have a gap behind them and set each on the impact-effort matrix, by what the job would change and what serving it would take; the audit ranks its own findings the same way. A use case that needs a key the stack does not yet hold waits on whatever builds that key.
Then put the whole list in front of the budget holder, the channel owners and the data protection owner, and ask each for the thing only they can supply: the budget holder for what is wanted enough to fund, the channel owners for what is missing or mis-stated, and the data protection owner for purposes. As general practice, record what was left out and why, in the team's decision log, so that the next person to ask for it can see it was weighed. Issue the agreed list as one version with a date.
Put the list where the routines read it and set what changes it
Put the list beside the stack record, link it from the form that new tool requests arrive through, and name it as an input in each routine's own notes. Then agree the changes that reopen it, so that keeping it true is a habit and not a project.
Event What happens to the list A request goes to evaluation A planned line is added, written as a job, with a team that owns it A use case goes live The status changes and the tools column is updated at the same time A tool is retired Each line that named it is rechecked; a job left with no tool becomes a gap A renewal review starts The tool is re-scored against the lines as they now stand The audit runs The whole list is reread against the inventory before any decision The measurement plan changes The outcomes the lines serve are checked against the new plan A new channel, market, platform migration or privacy requirement arrives The affected lines are reread; the marketing technology hub names these as the triggers for an unscheduled review Record each change in the list's own history with a date and a name. Hold to the designing guide's rule for its own artefacts: a change missing from the list has not been made.
The use-case list
The finished list is one table. Each field below earns its place, and the last column says what a list of sentences alone would miss.
| Field | Content | Why it earns a column |
|---|---|---|
| Use case | One sentence, an outcome, no tool named | Lets anyone judge a request or a tool against it |
| Outcome group | Acquisition, conversion, retention or reporting | Shows where the stack is crowded or thin |
| Status | Live, planned or an aspiration | Separates what the audit reads from what the roadmap reads |
| Owner | The team and a named person | Gives each line someone who can say whether it still matters |
| By hand | Whether a person acts as a connector | Finds the jobs that stop when someone is away |
| Data and events | What the job needs and where it comes from | Shows whether the data exists before anyone shops for a tool |
| Identifier | The key the job matches on | Explains why matches between tools fail |
| Timing | How soon the job must act | Decides how it is connected |
| Purpose | The purpose it serves in the consent map, with the basis left to data protection | Shows a use of personal data that has no purpose behind it |
| Layers and tools | Which layers it touches and the tool in each | Makes gaps and overlaps visible |
| Finding | Gap, overlap, manual dependency, missing purpose or none | Carries the finding to the routine that decides |
| Last change | The date of the latest edit and who made it | Shows whether the line has been read lately |
A mushroom kit seller writes its first use-case list (illustrative)
Virask Mushroom Kits writes its first use-case list
Virask Mushroom Kits, an invented seller of mushroom-growing kits through its own online shop and a stand at garden shows, has a marketing team of 3 and a stack of 11 tools. Ahead of its first audit the team gathers what already runs and finds 26 raw lines: 9 from the campaign calendar, 8 from flows and automations, 5 from reports people open and 4 from exports done by hand. Eight are the same job seen from two sources, which leaves 18, and two lines that each bundled three outcomes, such as the order confirmation, the first-week growing guide and the review request, are split, which adds 4. The list stands at 22 use cases: 15 live, 4 planned and 3 aspirations.
Mapping them to the layers and tools shows 3 live jobs served by two tools each, one of them the growing-guide email, sent from both the email platform and the shop's own notifications. Two live jobs, a weekly digest of stock and orders for the growing room and the entry of show-stand sign-ups into the customer list, run only because one person prepares them by hand, so the team flags them for the integration register. Of the 4 planned jobs, 2 need a layer no tool serves, because the stack holds no join between show-stand sign-ups and online orders. And 2 of the 11 tools appear on no line. One belongs to a team that has left the business, and the other runs the restock alerts for sold-out kits, a job nobody had written down. That live line is added, which brings the list to 23 use cases, 16 of them live, and leaves 1 tool serving none.
The team ranks the 2 gaps on the impact-effort matrix and sends the join between show-stand sign-ups and orders to the roadmap as the phase it proposes first. Overlaps and the unused tool go to the audit, and the 2 manual jobs to the register, noted for the roadmap too. A few weeks later a request for a quiz tool to recommend a kit is checked against the list: it matches one aspiration that no team has agreed to own, so it is declined with the reason in writing, the line stays an aspiration, and the answer changes when a team will own it.
Common mistakes
- Writing the list from the tool inventory. A list built by walking the tools can only contain what tools already do. Start from the work, so that the jobs no tool serves appear as gaps.
- Putting the tool in the sentence. A line that names a product can hide an overlap and is harder to use to judge a replacement. Write the job and give the tool its own column.
- Bundling several outcomes in one line. A sentence that joins three results cannot be live, planned or an aspiration in any single way, and it hides the part that is missing.
- Leaving out the manual routines. The scheduled export and the hand-built list are jobs too, and they stop when their person is away.
- Counting aspirations as live. A wished-for job on the live side of the list can make later gaps look smaller and tools look more useful than they are.
- Deciding the consent basis or retention in the list. The list records which purpose a job serves and leaves the basis and the periods to whoever owns data protection.
- Writing it once and filing it. A list that no routine reopens can be out of date by the next request. Name the events that change it and the person who changes it.
- Letting any request become a line. A line needs someone who asked for it and can say what outcome it serves. A request nobody will own is declined with the reason in writing, and its line stays an aspiration.
