Structured Briefs for CRM CampaignsFor CRM teams who want to think before they send

The problem#

Most CRM teams operate in a cycle that looks like this: someone requests a campaign, a brief gets written (or doesn’t), copy gets drafted, the email goes out, and the team moves on. It feels productive. But underneath that activity, something is missing.

Nobody asked why this message needs to exist. Nobody defined what the customer should do differently after receiving it. Nobody checked whether this campaign conflicts with the three other journeys the customer is already in. Nobody designed the experience. They just designed the email.

The result is familiar to anyone who has worked in or around CRM:

• Campaigns that exist because someone asked for them, not because there was a clear customer or business need. The brief said “send,” so the team sent.
• Messages written before the thinking is done. The subject line gets debated for an hour. The question of whether the campaign should exist at all gets debated for zero minutes.
• Journeys that collide with each other. A customer receives an onboarding sequence, a promotional offer, a product update, and a satisfaction survey in the same week, all from the same company, because each journey was designed in isolation.
• Metrics that measure activity, not impact. Open rates and click rates fill the dashboard but nobody can say whether customer behaviour actually changed.
• No shared language for what “good” looks like. Every campaign is evaluated on its own terms. There is no common structure that connects strategy to execution across the team.

This isn’t a people problem. CRM teams are typically sharp, fast, and under real pressure to deliver volume. The problem is structural. There is no consistent framework between “we need a campaign” and “here’s the email.” The gap gets filled with instinct, habit, and whatever the last campaign looked like.

Why this matters#

CRM is not a channel that sends messages. CRM is the architect of how people experience your product over time.

Customers don’t experience your campaigns individually. They experience the pattern, and that pattern shapes how they feel about your product, how much they trust you, and whether they stay.

When that structure is designed intentionally, CRM drives activation, deepens engagement, and builds relationships that compound. When it isn’t (when each campaign is a one-off reaction to an internal request) it becomes noise. And noise teaches customers to ignore you.

The difference is not talent, tooling, or budget. It is whether the thinking happens before the doing.

What this framework does

This brief gives CRM teams a shared system for that thinking. It works at three levels of depth (from quick sends to deep strategic work) and asks the questions that, if answered clearly, make everything downstream easier and more effective.

It does not tell you what to write. It tells you what to think about before you write anything.

Everything in the framework flows through five layers, in this order:

1. 1
   Outcome What should change?
2. 2
   Behaviour What must people do differently?
3. 3
   Experience What should it feel like?
4. 4
   Delivery What do we send, when, where?
5. 5
   Measurement How do we know it worked?

Tier 1 covers these in six quick questions. Tier 2 gives each layer its own section. Tier 3 adds the operational detail that makes the whole system executable.

How to use this#

There are three tiers. Pick the one that matches the stakes.

Every campaign deserves at least Tier 1. Most deserve Tier 2. Tier 3 is for when you are stepping back to design the whole system, not just one part of it. You can always start at Tier 1 and go deeper if the stakes warrant it.

What you need before you start

This framework assumes nothing about your tools, team size, or CRM platform. But it does require some baseline capabilities, and they increase with each tier. If you do not have them yet, that is not a reason to skip the framework; it is a reason to start at the tier you can actually execute.

The most common mistake is waiting until everything is perfect before starting. The second most common mistake is attempting a Tier 3 architecture with Tier 1 infrastructure. Start where you are.

Tier 1: Snapshot#

For quick sends, one-off campaigns, and time-sensitive messages. Before anyone writes a word of copy, answer these six questions. If any answer is vague, the campaign isn’t ready.

1. Who is this for?

Not “everyone” and not “our users.” Get specific. What defines this group? Are they new customers in their first week? Long-time users who haven’t tried a specific feature? People whose subscription is about to renew? Customers who were active last month but have gone quiet?

The sharper the answer, the better everything else becomes. A message for everyone is a message optimised for no one.

Vague: “Our customers”
Sharp: “Users who signed up in the last 14 days and haven’t completed their profile”

2. What do you want them to do?

After they read this, what is the one action you want them to take? Not think, not feel. Do. Click, buy, try, return, reply, upgrade, book, complete.

Test your answer: could you watch someone do it? If not, it is too abstract. “Feel more engaged” is invisible. “Open the dashboard and create their first report” is something you can see, measure, and design for.

If you have more than one answer, you have more than one campaign. Pick the most important action and design around it.

Vague: “Engage with the new feature”
Sharp: “Open the dashboard and create their first report”

3. Why should they care?

This is the question most teams skip, and it is the most important one. Put yourself in the customer’s position. They are busy. They get dozens of messages a day. Why should they give this one any attention?

The answer has to be about them, not about you. “We’re excited to announce” is about you. “You can now do X, which means you’ll stop wasting time on Y” is about them.

A useful test: if a friend described this message to the customer over coffee, would the customer be interested or would they shrug?

About you: “We’ve launched a new analytics dashboard”
About them: “You’ve been exporting data to spreadsheets manually. You can now see it all in one place without leaving the app”

4. Why now?

This question has two parts, and both need clear answers.

Part one: Why is this the right moment for the customer? Something in their journey should make this message timely. Good timing signals:

• Something just happened: they signed up, made a purchase, hit a milestone
• Something is about to happen: trial ending, renewal date, seasonal event
• Their behaviour changed: they stopped using something, started exploring something new, viewed pricing
• There is a genuine deadline or limited window

Part two: Should this message exist at all? If you cannot name a timing signal (if the honest answer is “because we decided to” or “because we haven’t sent anything this week”) pause. A message without a reason to exist is a message that spends the customer’s attention without earning it. It is better to send nothing than to send something that trains them to ignore you.

5. Where does this reach them?

Email, push notification, SMS, in-app message. Each channel has a different context and a different level of interruption.

Think about where the customer is when they receive this. Email is something they check at their own pace. A push notification interrupts whatever they are doing right now. An in-app message meets them where they are already engaged with your product.

Match the channel to the urgency and the action. If you want them to do something inside your product, an in-app message might make more sense than an email that asks them to stop what they are doing and switch contexts.

6. How do you know it worked?

Pick one number that tells you whether this achieved its goal. Not five numbers. One.

This should connect directly to your answer in question 2. If you want them to create their first report, the metric is “number of recipients who created a report within X days.” If you want them to renew, the metric is renewal rate for this segment.

Avoid vanity metrics. Open rates tell you the subject line worked. Click rates tell you the message was interesting. Neither tells you whether the customer actually did the thing that matters.

The ready check

Look at your six answers together. If any of them are vague, the campaign is not ready. Vague inputs produce generic messages, and generic messages get ignored. Go back and sharpen the weakest answer before moving forward.

Worked example

Here is a Tier 1 brief filled in for a feature adoption campaign. Notice how every answer is specific enough to act on.

Every answer passes the tests: the audience is a real person in a specific moment, the action is observable, the reason is about the customer, the timing is grounded in their journey, the channel matches the context, and the metric connects directly to the desired action. This brief took four minutes. The campaign it produces will be sharper than one that took an hour without this thinking.

Tier 2: Journey#

For anything with more than one touchpoint: onboarding sequences, activation flows, win-back campaigns, feature adoption series, re-engagement programmes. A journey is a designed sequence of connected steps with logic between them. If the customer does X, they get message A. If they don’t, they get message B. If they complete the goal, they stop hearing from you about it.

Start here: Outcome

Every journey begins with a measurable result. Not “improve engagement” but a specific number you can point to before and after.

Define three things: the outcome you are driving (what changes if this journey works), the current baseline (where that number sits today), and the target you are aiming for (where it needs to be, and by when). Without a baseline and a target, you have no way of knowing whether the journey worked. You are just sending messages and hoping.

Then: Behaviour

Outcomes are produced by behaviours. Before designing any messages, identify what you need the customer to do differently and, critically, diagnose why they are not doing it already.

There are four root causes of inaction, and each demands a different response:

• They don’t know about it (awareness): CRM can solve this directly with well-timed, relevant information.
• They don’t see the value (motivation): reframe the benefit in terms that matter to them, or show social proof.
• It’s too hard or confusing (friction): CRM alone cannot fix this. It requires product or UX changes. More emails about a broken flow make things worse.
• The timing is wrong (context): wait and resurface the opportunity when their situation changes.

Name the root cause in your brief so the entire team designs for the right problem.

Then: Audience

Describe a person, not a segment. A segment label like “inactive users, 30+ days” tells you who to include. It tells you nothing about who they are, what they were trying to do, or what got in the way. For the journey to work, answer four questions about the person you are designing for: who are they, what are they trying to accomplish, what is stopping them, and where are they in the lifecycle?

Part A: Who enters and who leaves

Before designing any messages, define the boundaries of the journey.

Entry trigger

What event starts this journey? Be precise. “Signs up” is different from “signs up and confirms their email.” “Abandons cart” is different from “views a product page without buying.” The trigger is the specific moment that puts someone into the sequence.

Entry requirements

What else must be true about them at the point of entry? The trigger gets them to the door. The requirements decide if they walk through. For example, the trigger might be “hasn’t logged in for 30 days” but the requirement might be “has an active subscription and has used the product at least three times before.” Without the requirement, you risk sending a win-back message to someone who only ever signed up by accident.

Exclusions

Even if someone matches the trigger and requirements, are there reasons they should not enter? Perhaps they are already in another journey that would conflict. Perhaps they recently contacted support with a complaint. Perhaps they have already opted out of this type of communication. Think about who would have a bad experience receiving this sequence right now.

Exit conditions

There are two types of exit:

• Success exit: They did the thing. They completed onboarding, they came back, they adopted the feature. Stop messaging them about it. Continuing after the goal is met is one of the fastest ways to erode trust.
• Graceful exit: They didn’t do the thing, but you have reached the point where continuing would be annoying rather than helpful. Let them go. Not every journey converts everyone, and knowing when to stop is as important as knowing when to start.

Re-entry rules

If someone completes the journey (success or graceful exit) and later meets the entry conditions again, should they re-enter? A customer who churns and re-signs up might qualify for the onboarding journey a second time. A user who adopted a feature, stopped using it, and became inactive again might re-qualify for the activation sequence.

Decide three things:

• Is re-entry allowed? Some journeys are one-time by nature (first-purchase welcome). Others should be repeatable (re-engagement, win-back).
• Does the sequence still make sense? If the journey assumes first-time context (“Welcome! Here’s how to get started”), it will feel wrong to a returning customer. You may need a variant for re-entrants.
• What is the cooldown? How long after exiting must someone wait before they can re-enter? Without a cooldown, a customer hovering around the entry threshold could loop through the journey repeatedly.

Part B: Journey map

Before designing individual messages, walk through the experience from the customer’s perspective. For each stage of the journey, answer three questions: what are they thinking, what is stopping them, and where is the opening?

A message designed for someone in the awareness stage will fail if the customer is already past consideration and stuck on friction. Complete this map before designing any touchpoints.

Part C: Designing the sequence

Now map out each step. For every touchpoint in the journey, answer these questions:

What is this single step trying to achieve?

Each message in a sequence has its own job. The first might educate. The second might show proof. The third might create gentle urgency. Don’t try to do everything in every message.

When does it arrive?

How long after the trigger or the previous step? Timing matters more than most teams realise. An onboarding email two hours after signup catches someone while they are still engaged. The same email three days later feels like a stranger. For re-engagement, the opposite applies: one day feels desperate; two weeks feels natural. Think about the emotional state of the person at each point in time.

Why this channel for this step?

The best channel might change within a single journey. Step one might be email because the content is detailed. Step three might be an in-app prompt because by that point the customer is actively using the product. Step five might be nothing at all, a deliberate pause where you let the customer act on their own terms.

What is the intent and emotional tone?

Don’t write the copy yet. Define two things: the job the message is doing, and the emotional register it should strike.

The job might be:

• Educating: helping them understand something they need to know
• Prompting: reminding them of something they started but didn’t finish
• Reassuring: making them feel confident about a decision they have made
• Proving: showing evidence that others like them got value from this
• Creating urgency: making the case that acting now matters

The emotional tone is separate from the job. An educational message can feel warm or brisk. A prompt can feel like a gentle nudge or a formal reminder. For each step, ask: how should this feel when they read it? The copy comes later. The intent and tone come first.

What happens next depending on their response?

This is where journeys become more than a series of scheduled messages. If the customer takes the desired action at step two, they should not receive step three (which was designed to encourage the action they have already taken). If they open but don’t click, the next message might take a different angle. If they don’t open at all, you might try a different channel.

Map the branches. Not every branch needs to be complex, but every step should account for at least two scenarios: they moved forward, or they didn’t.

Part D: Creative direction

This section bridges strategy and execution. It tells the people who will write the copy, design the visuals, and create the assets what world this journey lives inside.

Tone of voice

How should this journey sound? Not your brand’s general tone, but the specific tone for this journey, for this audience, at this moment in their experience. An onboarding sequence for a first-time user will sound different from a win-back sequence for someone who cancelled. Name the tone explicitly so everyone involved creates from the same register.

The story you are telling

Every journey has a narrative arc, even if it is simple. What is the starting state? What is the transformation? What does the ending look like? Naming the story keeps individual messages coherent across the full sequence rather than feeling like disconnected fragments.

What this should feel like

Try this: describe the journey as if it were a scene in a film. Where is the character? What just happened to them? What do they need to hear next? This exercise sounds unusual for CRM work, but it unlocks a kind of empathy that spreadsheets and segment names cannot. It puts you inside the customer’s experience rather than looking at it from the outside.

Part E: System triggers and dependencies

What must exist in the system for this journey to actually run? This section makes the journey launchable. Without it, strategy stays on paper.

Events that trigger this journey

What specific system events start the sequence? These need to be real, trackable events in your data, not concepts. “Customer shows interest” is a concept. “Customer views pricing page twice in 7 days” is a trackable event.

Segmentation logic

What data fields, behavioural signals, or calculated attributes determine who qualifies? Write this precisely enough that someone configuring the journey in your CRM platform can translate it directly into segment rules.

Suppression rules

Who should be actively prevented from entering, even if they match the trigger? Customers in other active journeys, recent support contacts, users who have already converted, people who have opted out of this message type. Be explicit. Suppression gaps are one of the most common sources of bad customer experiences in CRM.

Dependencies

What features, data integrations, tracking events, or content assets must exist before this journey can launch? Map these clearly so you know what is ready and what is blocking.

Part F: Journey-level sanity checks

Once the sequence is designed, step back and pressure-test the whole thing.

How much is too much?

Count the maximum number of messages someone could receive if they never take the desired action. Is that number reasonable? Every message that doesn’t land erodes trust slightly. Set a cap you are comfortable with and design the journey within that constraint.

What else are they hearing from you?

Your customer does not experience your journeys in isolation. They might simultaneously be in an onboarding sequence, receiving a product newsletter, getting billing notifications, and being targeted by a promotional campaign. What does their inbox look like when you add everything together?

This is one of the most common CRM problems and one of the hardest to spot, because different teams often own different journeys. Someone needs to look at the total experience across all active programmes. In practice, this requires three things:

• Frequency caps. Set a maximum number of non-transactional messages a customer can receive per channel per week. The right number depends on your product and your audience, but most teams over-communicate. A reasonable starting point: no more than two to three marketing messages per week per channel. Transactional messages (receipts, password resets, billing alerts) are exempt, but they still count toward the customer’s total experience of hearing from you.
• Priority rules. When two journeys want to send to the same person on the same day, which one wins? Define a clear hierarchy. Typically: transactional > lifecycle-critical (onboarding, renewal) > behaviour-triggered (feature adoption, re-engagement) > scheduled campaigns (newsletters, promotions). The lower-priority message either waits or gets suppressed for that customer, not deleted entirely.
• A single owner of the cross-journey view. Someone (a person, not a team) needs to be responsible for auditing the total customer experience across all active programmes. Without this, every individual journey can be well-designed and the sum total can still be overwhelming. This person reviews the message calendar, flags collisions, and has the authority to delay or suppress sends that would breach frequency caps or conflict with higher-priority journeys.

What will you test?

Measurement tells you whether things worked. Testing tells you why and what to try next. Before the journey launches, name what you plan to experiment with: subject lines, send timing, channel choices, message angles, sequence length. A journey without deliberate experimentation only ever gets as good as the first version. One that tests gets better every cycle.

Worked example

Here is a partial Tier 2 brief for a win-back journey. It covers the strategic sections (Outcome, Behaviour, Audience) and Parts A–C. In practice, you would also complete Parts D–F before handing this to your team.

Notice the structure: strategy was completed before a single touchpoint was designed. Each step has a defined job, branching logic ensures nobody receives a message designed for someone in a different state, and the sequence caps at four messages over 45 days, restrained enough that even non-converters are left with a respectful impression.

Tier 3: Architecture#

For lifecycle design, CRM programme strategy, or any moment where you are stepping back and asking “how should all of this work together?” This tier is not about a single campaign or journey. It is about designing the system that all campaigns and journeys live inside.

Tier 3 extends Tier 2. Everything in a journey brief still applies, but at the programme level. Where Tier 2 designs one journey, Tier 3 designs the system all journeys live inside. If you have already completed a Tier 2 brief, Tier 3 builds on that work rather than replacing it.

The system is built on five strategic layers. Each one flows from the one before it:

A full architecture brief covers twelve dimensions. The first five map to those strategic layers. The remaining seven make the system deliverable. Here is the complete structure:

Here is how the twelve dimensions map to the five layers:

Resist the urge to start at the bottom (channels and tools) and work up. Start at the top (outcomes) and let everything else follow.

Layer 1: Outcome

What changes if all of this works?

Before designing any communications, get clear on what CRM is actually trying to achieve.

There are two sides to this:

The business outcome. What does the company gain? More revenue? Higher retention? Faster activation? Lower support costs? Be specific. “Growth” is not an outcome. “Increase 90-day retention from 62% to 70%” is.

The customer outcome. What does the customer’s experience look like when CRM is doing its job well? They get value from the product faster? They feel informed without feeling overwhelmed? They discover capabilities they didn’t know existed? They feel like the company understands where they are in their journey?

Where these two align is your sweet spot. The best CRM work serves both sides simultaneously. The customer genuinely benefits and the business measurably improves. Where they tension against each other (short-term revenue pressure vs. long-term customer trust, for example), name that tension honestly. It will not go away by ignoring it, and the team will make better decisions if the tension is visible.

Layer 2: Behaviour

What do people need to do differently?

Outcomes are the result of behaviours. If you want higher retention, what behaviours drive retention? If you want faster activation, what does an activated customer actually do?

Identify the three to five behaviours that matter most for the outcome you defined in Layer 1. Not thirty. Not ten. The vital few.

For each behaviour, ask:

What is happening now? What percentage of customers exhibit this behaviour currently? Is it increasing or decreasing? Do some segments do it naturally while others don’t?

What is the gap? How far is current behaviour from what you need? A small gap might need a nudge. A large gap might indicate a product problem that no amount of messaging will solve.

What is causing the gap? This is critical, because each cause requires a different response. There are broadly four reasons someone does not do something:

• They don’t know about it. An awareness problem. They literally don’t know the option exists or that they should care about it. CRM can solve this directly with well-timed, relevant information.
• They don’t see the value. A motivation problem. They know about it but don’t believe it is worth their time or effort. CRM can help by reframing the value in terms that matter to this specific segment, or by showing social proof.
• It’s too hard or confusing. A friction problem. They want to do it but something in the experience stops them. CRM alone cannot fix this. It requires product or UX changes. Sending more emails about a broken flow makes things worse, not better.
• The timing is wrong. A context problem. They might do it later, but right now is not the right moment in their journey. CRM can help by waiting and resurfacing the opportunity when the customer’s context changes.

Diagnose the cause before designing the response. An educational email sent to someone who already understands but finds the process confusing is a waste of everyone’s time.

Layer 3: Experience

What should the customer’s journey actually feel like?

This layer is about designing the experience before deciding on any specific messages or channels. Think about it entirely from the customer’s perspective, across time.

Map the ideal experience

For the lifecycle stage you are focused on (first 30 days, renewal period, post-purchase, re-engagement), what should the customer feel, learn, and do at each phase? Not what messages they receive. What they experience.

For example, in the first week after signing up, the ideal experience might be: “I feel welcomed. I understand the one thing I should try first. I have a small early success that makes me want to come back. I am not overwhelmed with information.”

That is not a message plan. It is an experience design. The messages come later to serve it.

Find the moments of highest leverage

In every lifecycle stage, there are a handful of moments where a well-timed interaction can meaningfully change the customer’s trajectory. The moment after their first success. The moment they hit a confusing point. The moment they are considering whether to renew. The moment they come back after being away.

These are your most valuable touchpoints. Design them with the most care. Everything else is supporting material.

Find the silences

Where are you currently not communicating when you should be? And equally important, where are you communicating when you would be better off saying nothing? Sometimes the most respectful, trust-building thing you can do is leave someone alone. A well-designed silence is a sign of CRM maturity, not a gap in the calendar.

Layer 4: Delivery

How does this get operated?

Only now do you translate the experience design into concrete delivery. This is where strategy meets reality, and the gap between the two is usually larger than anyone expects.

What journeys and campaigns does this require?

Map the experience from Layer 3 into actual sequences, one-off sends, and triggered messages. Each one should trace back to a specific part of the experience you designed. If a campaign does not connect to a moment in Layer 3, question whether it should exist.

Be explicit about the type of each communication:

• Triggered journeys: sequences initiated by a customer event (signup, purchase, inactivity). These are the backbone of most lifecycle programmes.
• Scheduled campaigns: time-based sends such as product updates, seasonal promotions, or content roundups. These run on a calendar, not a trigger.
• Transactional messages: receipts, password resets, billing alerts. Often owned by product or engineering, but they are part of the customer’s experience and need to be accounted for in the architecture.
• Deliberate silences: moments where the design says “do not communicate.” These should be documented as explicitly as any send.

What channels carry each touchpoint?

Match channels to moments. High-leverage moments might warrant multiple channels. Routine updates might need only one. Some moments might not be messages at all; they might be in-product experience changes, UI adjustments, or even human outreach from a success team.

For each touchpoint, ask three questions: Where is the customer when this arrives? (in-app, on their phone, at their desk), How urgent is the action? (immediate vs. whenever they get to it), and How complex is the content? (a single prompt vs. detailed information). These three factors narrow the channel choice to one or two sensible options.

What do you need to make this work?

Be honest about requirements. Map the gap between what the strategy demands and what you can actually deliver today. This usually breaks into four categories:

• Data: What behavioural events, customer attributes, or calculated fields do you need that do not currently exist? Which systems hold the data, and can you access it in real time or only in batch?
• Content: How many unique assets need to be created? Who writes, designs, and approves them? What is the realistic production timeline?
• Integrations: What connections between systems are required? CRM platform to product analytics, billing system to email tool, support platform to suppression lists. Each integration is a dependency that can delay launch.
• Tooling: Can your current CRM platform support the branching logic, personalisation, and frequency rules the design requires? If not, what compromises do you make, and what do you escalate?

Who owns what?

CRM rarely operates alone. Product teams own in-app experiences. Engineering teams own data pipelines. Content teams create assets. Design teams shape how things look and feel.

For each component of the delivery plan, name three things: the owner (the person accountable for it being done), the contributors (the people doing the work), and who needs to be informed (stakeholders who are affected but not directly involved). Ambiguous ownership is where good strategies go to quietly die. A RACI-style mapping is useful here, but the important thing is that every component has exactly one owner, not zero and not two.

Layer 5: Measure

How do you know it’s working, and how do you keep improving?

Measurement in CRM operates on two timescales, and you need both.

Leading indicators tell you quickly whether behaviours are shifting. These are things you can see within days or weeks: activation rates, feature adoption rates, email engagement patterns, journey completion rates. They don’t prove the outcome is improving, but they tell you whether you are on the right track.

Lagging indicators confirm whether the outcome is actually moving. These take longer: retention rates, lifetime value, revenue per customer, churn rates. They are the ones that matter most but they require patience.

Leading indicators without lagging indicators can mislead you. You might be driving clicks that don’t translate into meaningful behaviour change. Lagging indicators without leading indicators leave you flying blind for months.

Design your feedback loop

How often do you review performance? What triggers a redesign of a journey versus a tweak to a single message? Who reviews the data and who decides on changes?

A reasonable starting rhythm: a quarterly architecture review (revisiting Layers 1–3) combined with monthly journey reviews (Layer 4 performance) and weekly message monitoring (open rates, click rates, conversion rates at the individual send level).

Define clear escalation rules. Not every underperforming metric needs the same response:

• Message-level underperformance (low open rate on one send): tweak the subject line, the timing, or the creative. This is a weekly decision.
• Journey-level underperformance (the sequence is not driving the behaviour): revisit the journey design: the branching logic, the number of steps, the root cause diagnosis. This is a monthly decision.
• Architecture-level underperformance (the outcome is not moving despite active journeys): revisit Layers 1–3. The problem may be the wrong behaviours, the wrong audience, or a product issue that messaging cannot solve. This is a quarterly decision.

What will you test?

Measurement without experimentation is passive observation. At the architecture level, define what hypotheses you want to test across the programme. Which lifecycle stages are you least confident about? Where are you making assumptions about channel effectiveness, timing, or messaging that could be validated?

Structure your testing roadmap across three levels:

• Tactical tests (within a journey): subject lines, send times, CTA wording, creative variants. Fast to run, fast to learn from.
• Structural tests (journey design): sequence length, channel mix, branching logic, tone shifts between steps. Slower to learn from, but higher impact.
• Strategic tests (architecture level): does adding a whole new journey to the lifecycle improve the outcome? Does removing a journey that feels noisy improve trust and long-term retention? These are the tests most teams never run, and they often matter most.

Create a testing roadmap alongside the measurement framework so the system improves deliberately, not by accident.

Name your blind spots

What can’t you currently measure that you wish you could? Maybe you can’t track in-app behaviour after an email click. Maybe you have no way to see which messages a customer received across all journeys in the past month. Maybe there is no mechanism to measure whether customers feel over-messaged.

Write these down. They are your measurement roadmap. The gaps that, once closed, will make every other decision sharper.

Common blind spots that are worth naming explicitly:

• Cross-journey view: Can you see the total volume of messages a single customer received across all programmes in a given period?
• Channel attribution: When a customer acts, can you tell which touchpoint (or combination of touchpoints) contributed?
• Negative signal tracking: Are you measuring unsubscribes, spam complaints, and message fatigue as carefully as you measure clicks and conversions?
• Holdout measurement: Do you have a control group that receives no CRM communications, so you can measure the true incremental impact of the programme?

Beyond the layers: making it operational

The five layers are the strategic core. But a strategy that cannot be operated is just an essay. These additional sections complete the architecture brief by connecting the thinking to the doing.

Campaign overview

Why does this initiative exist? What business context created the need? A lifecycle architecture does not appear in a vacuum. There is usually a trigger: a retention problem, a new product launch, a strategic shift, a platform migration. Name it. This gives everyone involved a shared understanding of the stakes and the urgency.

A strong campaign overview answers three questions: What changed? (the event or insight that triggered this work), What is at stake? (the business impact if nothing changes), and What is the scope? (what this initiative covers and, equally important, what it deliberately does not). Scope boundaries prevent the architecture from growing into an everything-project that never launches.

Creative direction

At the architecture level, creative direction is broader than a single journey. It defines the tone, narrative principles, and visual language across the entire programme. How should the brand sound at each lifecycle stage? What messaging principles hold true across all journeys? What should the cumulative creative experience feel like when a customer moves through the full lifecycle? This section gives creative teams the rails to work within across every journey they design.

Think of it as a style guide for communication, not design. It should answer:

• Tone by stage: How does the voice shift as the customer moves from new user to long-term customer? Onboarding might be warm and guiding. Renewal might be confident and direct. Win-back might be understated and respectful.
• Messaging principles: The three to five rules that every message in the programme must follow. For example: “Lead with what the customer gains, not what we built.” “One message, one action.” “Earn the right to ask before asking.”
• What this should never sound like: Anti-patterns are sometimes more useful than guidelines. Name the tones that are off-limits (desperate, corporate, condescending) so the team has clear boundaries.

Product and system requirements

Everything that must exist for the architecture to function: tracking events, data integrations, segmentation capabilities, automation rules, feature flags, and platform configurations. Be thorough and honest. The gap between what the strategy requires and what the current systems support is one of the biggest risks in any lifecycle programme. Naming it early prevents painful surprises during implementation.

Structure this as a dependency audit:

• Have today: Capabilities that already exist and are ready to use.
• Need to build: Requirements that the team can deliver within the programme timeline.
• Blocked on others: Dependencies that require work from product, engineering, or third-party vendors. These are the highest-risk items. Flag them early and escalate them clearly.

Stakeholder ownership

Clear names. Clear responsibilities. CRM leads the engagement architecture, but the work touches product, engineering, content, design, data, and often commercial teams. For each dimension of the architecture, name who owns it, who contributes, and who needs to be informed. Ambiguous ownership is where good strategies go to quietly die.

Two common failure modes to watch for:

• Shared ownership with no single decision-maker. When two teams “co-own” something, neither is accountable. Every component needs exactly one person who can make the final call.
• Ownership that does not match authority. Naming the CRM team as owner of in-app messaging means nothing if the product team controls the roadmap for in-app features. Ownership only works when the owner has the ability (or explicit agreement) to make changes.

Delivery plan

Timeline, dependencies, asset requirements, and launch sequence. What gets done first? What can run in parallel? What has hard dependencies that gate everything else? A phased rollout is almost always better than a big-bang launch. It lets you learn and adjust before the full system is live.

A practical delivery plan has three layers:

• Phase 1: Foundation: The minimum set of journeys, data connections, and tooling needed to launch something real and start learning. Aim for the smallest version that delivers value and generates data.
• Phase 2: Expansion: Additional journeys, channels, and personalisation that build on what Phase 1 taught you. This is where most of the architecture comes to life.
• Phase 3: Optimisation: Refinement based on real performance data. Testing at scale, advanced segmentation, cross-journey orchestration, and the longer-term capabilities that require Phase 1 and 2 to be running.

For each phase, name: what launches, what data or content needs to be ready, who is responsible, and the target date. Be realistic. A plan that says “everything launches in four weeks” is not a plan. It is wishful thinking.

Worked example

Here is a partial Tier 3 brief for a SaaS company redesigning its first-90-days lifecycle programme. It covers the five strategic layers. In practice, you would also complete the operational sections (Campaign Overview, Creative Direction, Product & System Requirements, Stakeholder Ownership, Delivery Plan, Measurement Framework) before moving to execution.

Notice how the layers build on each other: the outcome (72% retention) defines which behaviours matter, the behaviour analysis diagnoses why each one is underperforming, and the experience design maps the customer’s journey before a single touchpoint is specified. A team working from this brief can design individual Tier 2 journeys that serve the architecture rather than competing with it.

Principles#

These apply at every tier.

Start with the outcome, not the channel

Know what you are trying to change before you decide how to say it. Start with what needs to change, not “we should send an email about X.”

Behaviour before message

Understand what people need to do differently, and why they are not doing it, before you write a single word. The diagnosis shapes the treatment. A message aimed at the wrong cause of inaction will, at best, be ignored and, at worst, feel tone-deaf.

Design the sequence, not just the send

Every message exists inside a larger experience, whether you designed it or not. The customer experiences all of your communications together. Better to design that deliberately than wonder why engagement is declining.

Attention is a finite budget

Every message you send costs the customer a small amount of attention and goodwill. Some earn that back. Others just spend it. Fewer, better messages almost always outperform more.

Measure to learn, not just to report

If a metric does not change what you do next, it is decoration. Every number you track should have an “if this, then that” attached to it. If activation drops below X, you revisit the onboarding journey. If open rates on step 3 fall below Y, you test a new angle. Measurement without action is just surveillance.

Using this with AI#

Once you have completed the framework at whatever tier you need, you have something valuable: a clear brief. That brief becomes a powerful input for AI tools, copywriters, internal teams, or anyone involved in execution.

What each tier gives AI to work with

A completed Tier 1 gives an AI tool everything it needs to draft a single message: a defined audience, one action, a customer-centric reason, timing context, channel choice, and a success metric. Hand it over with a prompt like “Draft the copy for this campaign based on the brief below” and the output will be dramatically better than asking for “an email about our new feature.”

A completed Tier 2 gives AI the structure to draft an entire sequence: the narrative arc across touchpoints, the emotional tone at each stage, the branching logic, and the creative direction. You can generate multiple message variants, test different angles for specific steps, or ask AI to identify gaps in your sequence logic.

A completed Tier 3 is too strategic for AI to execute directly, but it becomes a reference document. You can ask AI to audit individual journeys against the architecture, check for cross-journey conflicts, or draft the content for specific touchpoints within the system you have designed.

The difference in practice

Here is the same campaign drafted two ways. The first prompt has no brief behind it. The second is backed by a completed Tier 1.

The AI model was identical in both cases. The difference was entirely in the input. A sharp brief produces sharp output. A vague brief produces the kind of copy that gets ignored.

Where AI struggles without clear inputs

AI is exceptionally good at execution. It is poor at diagnosis. If your behaviour analysis is vague (“users aren’t engaging”), AI will generate confident-sounding copy that addresses the wrong problem. If your audience definition is a segment label instead of a person, AI will write generic messages that could be for anyone. The framework does the diagnostic work that AI cannot do for you. AI amplifies the quality of your thinking; it does not replace the need to think.

Watch for these patterns when reviewing AI-generated output from your briefs:

• Generic value propositions. If the AI writes copy that could apply to any product, your “Why should they care?” answer was too vague. Sharpen the brief, not the prompt.
• Tone inconsistency across a sequence. If each message in a journey feels like it was written by a different person, your creative direction section lacked specificity. Name the tone explicitly and give AI an example of what it should sound like.
• Friction-blind messaging. If your behaviour diagnosis identified a friction problem but your brief didn’t say so, AI will default to awareness-style copy (“Did you know you can…”) when the customer already knows and is stuck for a different reason.

How AI changes the economics

With AI handling execution, the cost of going from brief to draft drops dramatically. A Tier 1 brief that used to take five minutes of thinking plus thirty minutes of copywriting now takes five minutes of thinking plus thirty seconds of generation. This changes when teams should escalate: if producing a Tier 2 journey used to take days of writing, some teams cut corners and shipped Tier 1 campaigns instead. When the writing is near-instant, the only remaining cost is the thinking, and that thinking is exactly what separates campaigns that work from campaigns that exist.

One final thing. The best brief is the one your team actually uses. If it lives in a folder no one opens, it is not a framework. It is a document. Make it easy to reach for and fast to fill in.

Think first. Then move fast.